GM’s original Hydra-Matic transmission was one of the most important innovations in the history of the automobile. It wasn’t the first automatic transmission, but it was the first one that really worked and its resounding commercial success paved the way for every subsequent auto-shifter. This week, we take a look at the origins of the Hydra-Matic and its originator, Earl Thompson, who also developed the first Synchro-Mesh gearbox back in the 1920s.
SHIFTING GEARS
In 2010, Ferrari raised the hackles of automotive purists with the announcement that it would shortly phase out its conventional manual transmissions in favor of F1-style sequential gearboxes. The announcement gave new fuel to an old debate: whether a conventional manual transmission and separate clutch pedal are fundamentally obsolete.
Outside of a small contingent of enthusiasts and professional drivers, the automotive world has long regarded the manual gearbox as at best a necessary evil. The multi-speed transmission, which dates back to the 1890s, evolved to compensate for the limitations of early engines, which had modest power and narrow rev bands. Getting a heavy car moving from rest required short (high numerical) gear ratios that would have the engine thrashing its guts out above about 15 mph (25 km/h), but a ratio optimized for brisk cruising — at say, 40 mph (65 km/h)– wouldn’t have the torque multiplication to deal with steep hills. One of the great attractions of early electric cars, despite their severely limited range, was that they seldom required any gear changes at all, since electric motors produce their maximum torque from 0 rpm. Had early automakers devised a more efficient means of storing electricity, the evolution of passenger-car powertrains might have been very different indeed.
Actually shifting an early manual transmission was seldom a pleasant experience. Even upshifts often required careful timing and patience to avoid grinding gears, while downshifts required double-clutching and rev-matching. Few people ever mastered those techniques, particularly since engine speed generally had to be judged by ear; tachometers were not common on mundane automobiles in those days. Then as now, there were a few who prided themselves on being adept with the gearbox, but they were definitely in the minority. Many drivers opted instead to shift to high as quickly as possible and then stay there for as long as possible.
This hatred of gear-shifting was not limited to the public. Henry Ford strongly preferred planetary (epicyclic) transmissions, so much so that he allegedly did not even learn to use a conventional gearbox until the development of the Model A in the mid-1920s. The Ford Model T’s pedal-operated planetary gearbox at least avoided clashing gears, although in practice it was scarcely less labor-intensive to use than a conventional gearbox.
Shortly after the Great War, a number of engineers, including Britain’s Walter Gordon Wilson and France’s Jean Cotal, developed more sophisticated “preselector” planetary transmissions. With a preselector gearbox, you chose a ratio with a selector lever and then engaged that gear by pushing the gear selector pedal, which took the place of the traditional clutch. Preselector gearboxes were easier to use than was a conventional manual transmission, but they were also less efficient, substantially costlier, and often too complex to be completely trustworthy. As a result, they never quite caught on for passenger car use, although Cotal and Wilson preselectors were used in a number of pricier British and French cars (and quite a few British buses) into the fifties.
The preselector transmissions were automatic in a sense, but, with a number of rare exceptions beyond the scope of this article, they did not relieve the driver of the need to select the appropriate gear ratio for any given circumstance. Although there had been experiments with autonomously self-shifting transmissions since at least 1904, the technical challenges remained substantial, and none of the various attempts had been reliable enough or practical enough to have much success. A conventional dual-shaft transmission, whatever its deficiencies, was at least a known quantity.
SILENT SYNCHRO-MESH
Among the many people searching for easier ways to change gears was a young hydraulics engineer from Oregon named Earl A. Thompson. In 1918, Thompson applied for a patent on a three-speed preselector transmission that used a drum-shaped synchronizer to match the speed of each newly selected gear with the speed of the transmission output shaft prior to engaging that gear. The idea was that the driver would select the desired gear and then press the clutch pedal, which would disengage the clutch, synchronize the speeds of the selected gear and the output shaft, complete the shift, and then reengage the clutch automatically.
Thompson continued to develop and refine this idea with the aid of his younger brother Kirk, filing a second patent disclosure in 1923 that included a new synchronizer mechanism using cone clutches to match the speeds of gears to be meshed. He also managed to build functional prototypes of his preselector transmission, which he and Kirk installed in a new Cadillac and several other cars.
Thompson and his brother made several trips to Detroit, where they attempted to sell their invention to the auto industry. Although Detroit’s usual reaction to outside inventions bordered on the categorically hostile, the Thompson brothers eventually managed to secure an audience with Cadillac chief engineer Ernest W. Seaholm and then entered preliminary discussions with GM’s New Devices Committee about the possibility of GM’s purchasing or licensing Thompson’s patents. Those negotiations went nowhere, but Seaholm, who had found Thompson’s design crude but interesting, convinced Cadillac general manager Herbert Rice that Cadillac itself should take an option on Thompson’s transmission design.
In 1924, Thompson moved to Michigan and went to work as a Cadillac consultant, developing his original ideas into production form. His preselector transmission concept was discarded — judging by the patent description, it added a lot of complexity to no obvious benefit — but Cadillac remained very interested in his gear synchronizer mechanism, which would work just as well in an otherwise conventional transmission. After exhaustive testing and more than two dozen prototypes, Cadillac finally put Thompson’s invention into production in August 1928. The new transmission, dubbed “Silent Synchro-Mesh,” debuted that fall on the 1929 Cadillac and LaSalle.
For cost reasons, the early Synchro-Mesh transmissions provided synchronization only between second and third gears, so shifting into first still generally required coming to a complete stop to avoid clashing. (The “all-synchro” transmission with synchronized low gear wouldn’t become universal until around 40 years later.) Still, the system was a considerable improvement on earlier “crashbox” transmissions and made driving a good deal less painful. Synchro-Mesh quickly spread to other GM divisions and was subsequently licensed or imitated by many other automakers in the U.S. and Europe. By the mid-thirties, most passenger cars used something like it. (After the war, Thompson’s gear synchronization design would eventually have a strong rival in Porsche’s patented balk ring system, but that’s another story.)
Thompson became a Cadillac employee roughly a year after the first Synchro-Mesh cars debuted. About a year after that, Seaholm and general manager Lawrence P. Fisher (who had succeeded Rice in May 1925) promoted Thompson to assistant chief engineer. He had taken much of the teeth-gnashing (both literal and figurative) out of shifting. The next step was to make the process automatic.
I very much enjoyed the first installment of the Hydramatic story. However, I think Henry Ford deserves a little more credit for the basic blue print of a mass produced planetary gearbox. The Model T which debuted in 1908 had the bands, clutch pack and planetary gear sets which are easily recognized in modern automatics.
Henry Ford didn’t invent the planetary gearset — epicyclic gear trains were used for a whole variety of industrial purposes in the 19th century — but he was definitely their most enthusiastic proponent for automotive use (and he also did an enormous amount of work with planetary gears for tractors). I don’t know how much of a direct link there were between the Model T and later planetary gearsets (including the various preselectors, most of which were invented in the 1918-1920 period), but certainly any automotive engineer would have been aware of its design.
Ironically, Ford’s own efforts to develop an automatic transmission pretty much came to naught; he was keen to include automatic with his "X car" (which would also have had an X-8 engine), but he couldn’t make it work. Ford Motor Company ended up turning to Borg-Warner to develop Ford-O-Matic. Even more ironically, Howard Simpson, who worked very closely with Ford at Fordson (the tractor division) and at Ford on planetary gears, subsequently designed the Simpson gearset used variously by Cruise-O-Matic, TorqueFlite, and Turbo Hydramatic.
I always wonder why Ford didn’t develop an early automatic. As you stated, they already had the transmission design, all that was needed was to hire a good fluid power engineer or two to work out the automation.
Well, Ford did do the Liquamatic, optional on 1942 Lincolns and Mercurys, although by all accounts it didn’t work out well. Interestingly, the car Henry Ford wanted to succeed the Model T was supposed to have automatic transmission as well as the peculiar X-8 engine, but neither worked out. I have no details about the transmission design, but I have to wonder if it was another case of Henry Ford I having a strong fixed idea that he refused to relinquish even when it proved to be impractical. (For Ford prior to WW2, a crucial component of any engineering change was getting Henry Ford to agree to not only the idea, but also the method of achieving it. Ford had a variety of talented engineers whose opinions and suggestions he was not at all interested in hearing.)
I’m starting to feel that I’ve done everyone a disservice in glossing over the mechanics of the hydraulic control systems for the Automatic Safety Transmission and the early Hydra-Matic, giving the mistaken impression that the controls were a trivial part of the system. That’s really not so — as the text says, the hydraulic controls even of the earliest Hydra-Matic formed a fairly complex analog computer, combining inputs from several different sources to make each shifting ‘decision.’ Developing a system that would perform all the necessary actions at the appropriate times was not a simple or easy task and at least at GM happened in stages over a six- or seven-year period.
Today, yes, a skilled engineer could probably sit down and design a basic analog hydraulic control system on a home computer, but that’s with the benefit of 80 years of design precedent and a mountain of readily accessible technical literature on the subject. In the 1930s, this was new territory and the designers and inventors were to a large extent making it up as they went along. At the time, even hydraulic brakes were a relatively recent innovation and a hydraulic brake system only has to transmit motion; there’s no logic or computation involved. (And before anyone says it, yes, as Oscar Banker and others strove to demonstrate, it was also possible to design an automatic transmission that would not use hydraulics at all, but a purely mechanical analog computer is not less complex than a hydraulic one — potentially quite the opposite.)
So, to be very clear: The planetary transmission in a Model T Ford was NOT eight-tenths of an automatic transmission, even a fairly simple two-speed automatic like the 1953-vintage Powerglide, any more than a hang glider is almost a B-17 because they both have wings.
Yet zero automatic controls or shift schedule. Yes, the T was revolutionary in 1909. Not so in 1927.
The planetary system is not what actually seperates an automatic transmission from a gearbox. The implication of gear train components and there progression in automotive history is an entirely different subject. The inventor of the synchro mesh is allways noted on the conception of the synchro, but it’s just a side note to his connection to the development of the hydro matic inside the walls of general motors. As a side note to that side note the article should reference it’s battle testing inside tanks and trucks during the war effort before it’s return to being developed for the automotive industry
Er, all of these things are mentioned in the article at some length.
The Continental R-670 is not a rotary (Wankel) engine, but a radial engine.
Whoops — a slip of the keyboard. Thanks!
The original rotary engine was actually a radial engine in which the crankshaft was stationary and the cylinders rotated (Gnome was a big name in rotary engines at the beginning of aviation).
Great article guys!! Keep it coming.
I had vaguely heard of the fire that destroyed the Hydramatic plant in 1953, but this excellent article got me wondering about it. After a bit of searching, I found an interesting photo spread in the November 2, 1953 issue of Life magazine on page 102. It brings home the enormous industrial enterprise and effort that went into the Hydramatic.
I’m already anticipating next week’s installment…
The Willow Run plant that GM bought to replace the Livonia factory was truly vast. It was built during the war to build heavy bombers. At its peak in 1944, it was churning out 25 B-24 Liberators — four-engine heavies much more complicated than any contemporary car — every day. Kaiser-Frazer leased it after the war, although it had far more capacity than they needed.
It’s still there, and still vast. Note the scale; Willow Run is roughly 4 sq. mi., including runways.
Willow Run is both a busy freight airport and a GM transmission plant (or [i]was[/i], I’m not sure of Willow Run’s role in New G.M.).
On a semi-related note, Connie Kalitta’s successful air freight operation is based at Willow Run. Many will recognize his name from drag racing circles.
In addition to transmissions, Willow Run was the final assembly point for various GM vehicles. For anyone in the Ypsi area, I highly recommend a visit—the grounds are pretty well wide open, since it’s a freight airport. Kind of surreal.
I would like to know the series numbers of the tranmissions that were built at the Hydromatic Plant in Three Rivers, MI. Ther years should be between 1979 to around 1994. This is when General Motors owned it.
I’m afraid I don’t have that kind of technical information. The best I can suggest is contacting the GM historical archives and seeing if they have those records. Sorry!
Hello – This morning I found your nice site while attempting to establish the invention and first use of synchonized shift transmissions, thanks for the great read.
I want to draw your attention to the GM Automatic Safety Transmission for automobiles of 1937-39, which I had understood was the first commercial application of a practical automatic transmission by any maker. This transmission was invented and the prototype(s) made by Oscar Banker in his garage shop in northern Ohio, and subsequently pitched to GM. Although at once interested, by that date GM had invested heavily in H-M development and proceeded to the H-M transmission you outlined (read:’not originated here’ syndrome). Banker’s transmission was a much simpler design, which was offered essentially unchanged as the AST, and he also originated the safety shift pattern we now know (which H-M never had) and lobbied tirelessly as a ASE member for this shift pattern moving reverse range off the bottom in the name of safety. Banker’s transmission design evolved (slightly) in to GM V-Drive used in busses, etc. for decades. His and the transmission’s story is in a well written out of print biography of Banker by Robert ‘Bob’ Hull, and outlines this story in detail with footnotes, including theft of design lawsuits and what followed. I am also a fan of the H-M transmission, but this great story of one inventor getting crushed under (like Ford and Ferguson) rates a mention as his design and use of it passed the test of time – Best & Thanks
I have read Banker's account and most of the underlying patents and addressed this issue on p. 2 of the article. However, I want to note the following:
– Oscar Banker's transmission designs, as represented in his patent filings between 1927 and 1937, actually encompass at least FOUR substantially different designs with different layouts and different types of gearing. So, when one talks about “Banker's transmissions,” it's important to specify which one. (I assume the one he showed to GM in 1930 would have been the early dual-planetary variety.)
– None of Banker's patents is identical or, at least from the layman's viewpoint, substantially similar to the GM designs.
– When I initially glanced at Banker's 2,262,747 patent, I though it was surprisingly similar to the Automatic Safety Transmissions in some respects, but examining it and Thompson's 1934-1935 patents makes it clear that they're actually quite a bit different in both how they're laid out and how they work.
– Even Banker did not claim that the Automatic Safety Transmission was his design, although he did allege that Hydra-Matic was an infringement of his patents (without specifying in what ways or aspects).
– Some of the aspects that were similar between Banker's designs and GM's were not original to either party. Planetary transmissions, for instance, had been used in automobiles for more than 20 years at that point and several of the preselector transmissions had used multiple planetary gearsets to provide several different ratios. Banker certainly had no claim to the idea of an automatic transmission, designs for which were patented when he was only 9 years old.
I am certainly not an attorney, much less a patent attorney, but studying the history of any kind of technology makes it pretty clear that patent cases are seldom as cut and dried as “bad actor steals innocent inventor's invention.” In situations like this, where you have a host of different, separately patented ideas and a mountain of prior art, it more often comes down to making a legal argument that certain specific design elements are similar enough to constitute infringement. That line typically has to drawn by a court and even then the verdicts are often appealed until one side or the other decides to cut their losses or seek an out-of-court settlement.
So, no, the Automatic Safety Transmission was not Banker's transmission with a new name.
Also, by Banker's own account, the bus transmissions Banker designed, which Yellow Coach did use, were produced in very small numbers. Banker said some of the buses using those transmissions remained in service for decades (which isn't difficult to believe), NOT that the design that replaced them — which Banker considered inferior — was a variation of his transmission. Building a few devices that remain in service for 30 years is categorically different than creating a device that remains in *production* for decades.
I was surprised to read, “Deliveries to non-GM users didn’t resume until later in the 1954 model year,” because I’ve read elsewhere that GM took the noble step of delivering the first Willow Run Hydra-Matics to Hudson, Kaiser and Nash when the first units rolled off the line in October 1953.
However, I’ll also concede that your article cites its sources, which is something that can’t be said for the other articles stating that such a gallant gesture was carried out by GM.
You’re right, I’ve read that, as well, although it slipped my mind.
The 1954 part I believe is correct — although it’s important to emphasize that it was the 1954 model year, not the ’54 calendar year. By the time the Willow Run plant had been converted, I think all H-M users (GM or not) were at least a few weeks into ’54 production. Even if GM sent the first units off the line to its outside customers (which they may well have done, as much for contractual reasons as magnanimity), early ’54 models would already have been built with alternative automatics. However, the “to non-GM users” part is possibly inaccurate, or at least misleading, so I’ve deleted it from the text.
Thanks!
the 1953 pontiac production ended on november 20th,1953, quite some time before that date, pontiac was receiving regular shipment of the dual range hydra-matic from the new hydra-matic plant, during the time after the fire, a little over 17,000 pontiacs were built with the powerglide transmission, the 1954 pontiacs started production in december of 1953, and ALL 1954 pontiacs ordered with automatic transmissions had the dual range hydra-matics, the same goes for cadillacs and oldsmobiles. General Motors told rolls royce that they would stop suppling rolls with the dual range hydra-matics in 1966, not 1978 like you wrote. GM had rolls adapt to the buick version of the turbo hydramatic 400 transmission. you should know that the dual range hydra-matic trans was B&M high performance transmission choice, handling alot more horsepower after B&M modifications. the dual range hydra-matic was replaced by GM with the controlled coupling hydra-matic because the public wanted a smoother shifting transmission, not because of rising horsepower ratings from cadillac, oldsmobile, and pontiac engines.
charles coker
1953 pontiac technical advisor
Charles,
Thanks for the comment. Rolls-Royce did indeed switch to the TH400 for most of its cars around 1966, but the big Phantom VI limousine retained the old four-speed until 1978, finally switching to the Turbo Hydra-Matic when the 6,750 cc engine was adopted the following year.
hi aaron, i don’t know where you got your information, but when general motors told rolls royce in 1966, that GM wanted rolls to adapt to the newer turbo hydramatic 400 transmission, that meant for all rolls royce models, not just most models. if it wasn’t profitable to supply dual range hydra-matics for all rolls royce models, it certainly wasn’t proitable to supply for one rolls royce model. i have search the internet, and can find nowhere any search results that backs up your claim of 1978. i know a rolls royce transmission rebuilder in the west los angeles, calif. area, 1966 was the last year for any rolls royce model using a dual range hydra-matic transmission.
Charles,
I don’t claim to be a Rolls-Royce expert, but if you don’t believe me, I would suggest looking up the Phantom VI in any number of Rolls-specific websites or something like the old World Cars volumes released annually by the Italian Auto Club in the seventies and early eighties.
The Phantom VI was a formal limousine produced in very, very limited numbers for VIPs like members of the British royal family. The model wasn’t included in most price lists and I think it was available only on a special-order basis. As with the old Toyota Century and Nissan President or the Daimler DS420, which did similar duty, the Phantom VI was almost exclusively chauffeur-driven and then mostly for full-dress affairs. Between that and the extremely limited volume, the Phantom tended to lag well behind other contemporary models in technology. For example, the Phantom VI also retained drum brakes and the older 6,230cc engine long after most other Rolls-Royce models adopted four-wheel discs and the bigger 6,750cc V-8.
I suspect that with the Phantom VI, there is an important distinction to be drawn between when components were *supplied* and when they were *used*. My assumption is that Rolls-Royce had a stockpile of the older Dual-Range transmissions — which the company would also have needed for service replacements and the like — and just kept using them in the Phantom VI until the supply ran out. Keep in mind that total lifetime production of the Phantom VI was in the hundreds, and that was spread out over a period of more than 20 years!
My father was part of the team that perfected the automatic transmission. He was in GM’s experimental engineering group from 1937 to 1948. He was also chosen to be the driver of the Olsmobile that received the first working automatic after he designed the passing gear fluid system. This test made Michigan news paper headlines when they drove it across the state “first production car without a clutch peddle”. My father left GM and opened the first automatic transmission shop in the US. This shop was located in Phoenix Az, from here he supplied rebuilt HM transmissions all over the western half of the US. One of my fathers passions was horsepower and the method to transfer it the pavement. When you see the picture of a B&M Hydramatic opened up with all the modifications you are looking at my fathers transmissions/inventions, B&M bought all thier racing Hydro’s from my father, the only thing B&M ever produced was the B&M sticker they put on it.
Did you father work with Earl Thompson?
Any idea where I can find information on the old B&M hydra-matic modifications? I just bought a 1954 Starchief with a V-8 389 hooked up to a 1955 Slant pan Hydra-matic. I’m looking at what options I might have or recommendations on maintenance and the best ATF to run in it. Thanks.
I can’t help with that, but perhaps someone else can offer some suggestions. Good luck!
Regarding when GM sold Hydramatics to other car builders: My uncle bought a new ’51 Kaiser with an HM. I am quite sure they sold both to Studebaker and Nash about the same time.
My father-in-law drove a 1953 Cadillar, which he claimed would regularly return over 20 mpg on the highway at 55-60. I don’t doubt that number.
I’d heard from what I thouht were good sources that the net slippage was in the 3% range. I’m 80, and have had direct experience with most of the early automatics and semi-autos from Chrysler. As well as servicing some of them.
Currently I am just finishing a novel, with as much good history as I can find, about how the small six car companies could have/should have been saved.
The question wasn’t when GM originally started selling Hydra-Matic to other carmakers, but when they *resumed* deliveries after the fire destroyed the original Hydra-Matic plant in Livonia in 1953. As the text indicates, GM certainly sold transmissions to outside companies before then, but the destruction of the plant obviously interrupted deliveries both to GM divisions and outside customers.
Net slippage in fourth gear was 3% or less, but it’s important to note that the way the fluid coupling is arranged relative to the planetary gearing means that you have significantly different rates of slippage in each forward gear, which was an intentional design choice. It’s really very clever, although it takes a while to get your head around.
Not to Studebaker: they developed automatic drive 3-speeds with a lockup 3rd gear for the 1950 model year and later bought non-lockup 3 speeds from Borg Warner from about 1955.
Kaiser, Jeep, Hudson, Lincoln, Rolls Royce, Nash, and GMC did source 4 speed HydraMatic.
You may be interested to know that our next article will deal in some detail about the workings of the Studebaker/Borg-Warner DG.
Actually, Studebaker used a 3-speed automatic transmissions with a lock-up torque converter. It was developed and manufactured by Borg Warner and introduced in 1951. I have driven Studebakers with that transmission. Some years after 1951 they deleted to lock up clutch on the torque converter.
Rolls Royce, under license from GM, basically copied the GM 4-speed Hydramatic.
In I believe 1955, Lincoln dropped the GM Hydramatic and used a Borg Warner 3-speed automatic.
Incidentally, the early Studebaker Automatic Drive/Borg-Warner DG is discussed in some detail in my article Giving Slip the Slip.
If I were going to be really pedantic, I would describe the DG as having a direct-drive clutch rather than a lockup torque converter. Typically, a lockup torque converter allows the torus cover or flywheel/flex plate to drive the transmission input shaft, so the converter lockup and the actual transmission gear are independent of one another. In the DG, the high clutch allows the flywheel to drive the output shaft, so direct drive is a completely mechanical connection. It doesn’t allow converter action in third gear, nor does it permit the converter to lock up in any other gear.
GM sold HydraMatic to Hudson, Nash, Lincoln, Kaiser, Rolls-Royce, and GMC. Studebaker developed it’s own Automatic Drive 3 speed, and Packard it’s own Ultramarine independantly.
Your last line: does it refer to an input from me a time back? It is exactly what I have been working on for four years!!! Every time I sit down to do some research, it seems I find a new bit of info I need to address. Hopefully, I will be able to publish soon. All 300 plus pages. Lots of car industry info, as well as cars, components, coulda/shoulda beens, etc.
hello, i find some of your statements about the hydra-matic in error. the fluid coupling does not waste fuel or slips alot at low speeds, my own 1953 pontiac chieftain custom catalina would always get 15 to 17 mpg city, and 19 to 20 mpg hwy. the fluid coupling barely slips enough at idle to allow the pontiac straight eight to keep running with a in gear idle speed of 365 rpm’s. the hydra-matic’s low slippage equal ideal transmission fluid temps that only heavy duty cars like taxi’s and police cars needed a fluid cooler installed. as far as handling more than 165 horsepower, you only have to look at what transmission made B&M transmissions famous. the two best automatic transmissions to ever come out of detroit is the the 1955-56 dual-range hydra-matic, and the 1964-67 variable pitch turbo-hydramatic 400. charles l. coker, 1953 pontiac tech advisor, poci.
Mr. Coker,
Thanks for your comment. The remark to which you’re referring was a general description of the nature of fluid couplings, not specific to the Hydra-Matic. To function as a clutch, a fluid coupling has to have enough slip to absorb the engine’s idle torque, otherwise the engine will stall if the brake is engaged. The amount of slippage at engine idle speed or just off idle is therefore a lot higher than the slippage at cruising speed — more than 90% versus less than 5%.
That said, the Hydra-Matic’s fluid coupling had considerably less slip than most, if not all, of its early rivals, and as you note, HM-equipped cars could return quite good fuel economy. Some of that was achieved by using a lower numerical axle ratio than manual-shift cars, but the Hydra-Matic was certainly far more efficient than the early Dynaflow, Powerglide, or Ultramatic. I imagine that if one did a proving grounds comparison between a Hydra-Matic car and a manual-shift car (without overdrive) with the same axle ratios and tire sizes, the manual shift car would probably have a slight edge, but in the real world, it would likely be a wash.
Not to Studebaker: they developed automatic drive 3-speeds with a lockup 3rd gear for the 1950 model year and later bought non-lockup 3 speeds from Borg Warner from about 1955.
Kaiser, Jeep, Hudson, Lincoln, Rolls Royce, Nash, and GMC did source 4 speed HydraMatic.
Did you mean this in response to Dean’s comment? You’d replied this back in August, which you can see elsewhere in this thread.
But yes, in regard to this comment specifically, the Studebaker Automatic Drive/Borg-Warner DG (which is discussed in more detail in the split torque article) had the edge in cruising efficiency, since its converter is completely ineffective in third gear, whereas Hydra-Matic has, in effect, a partial lockup in third and fourth gears.
Actually, the Hydramatic had far less slip in 2nd gear than in 1st gear. I’ve driven 1953 Pontiacs with Hydramatic and a 1953 Oldsmobile with Hydramatic.
“L” on the selector was really 2nd; you could actually start in 2nd. With the car standing still and the brakes on, if one shifted from either drive position to 2nd the engine would noticeably slow down and, on the Pontiac, almost feel as if it was lugging. When accelerating from a stop in 2nd the engine revved very little at first and the car felt sluggish partly because the minimal slippage would not permit the engine to develop much torque.
In first gear the power flow was through the front planetary, then the fluid clutch, then the second planetary. That was why there was more slip in 1st gear than in 2nd gear.
Yes, the object of the Hydra-Matic’s unusual layout was to have fairly steady reductions in slip as you moved from first through fourth. If you want to be technical about it, the power flow is always through the front planetary gearset in the forward gears because the impeller of the fluid coupling is driven by the front gearset’s planet carrier rather than by flywheel. However, in first (and third), the front gearset is in reduction, so the front gearset drives the impeller at 69 percent of engine speed with 1.44 times engine torque. With the shift to second, the front clutch engages so the engine is now driving the impeller in direct drive, without the additional mechanical advantage; that’s probably why the engine loads up when you shift to “L.”
In terms of the performance difference between first- and second-gear starts, the substantial difference in overall ratio is probably more significant than the difference in hydraulic slippage. With a ’53 Dual-Range Hydra-Matic, first gear is 3.82:1. In second, you’re running only on the rear gearset, which has a ratio of 2.53:1. Since Hydra-Matic cars typically also have a taller axle ratio, it’s no wonder they end up feeling sluggish, especially with the old Pontiac straight six and straight eight, which weren’t exactly torque monsters to begin with. That’s about the same ratio as first gear on a three-speed manual car, but with the taller axle ratio, the coupling slippage, and the added load of the transmission oil pumps, the automatic car was obviously at a disadvantage.
Thanks for the great story and information. My late mother, who began driving before WWII, boasted that she never learned to drive a car with a manual transmission. She learned to drive on her mother’s 1937 Oldsmobile with AST, which my grandfather bought from an Oldsmobile rep traveling through the South. My mother later drove the car throughout the WWII years. She said that other than using the clutch pedal to change the gear quadrants, she left it in “High” and that it shifted very much like Hydra-Matic, which she drove for many years afterward. And speaking of Hydra-Matic, I remember as a boy back in the 1950’s, the women in the family talking about the Hydra-Matic “jump”. When starting out, they would always make sure that the gear selector was in the correct quadrant (especially when choosing “Reverse” on the far right side), give it the gas, and then release the parking break. I also remember Hydra-Matic having a rather unique and pleasing sound that is not heard in cars today.
Two points concerning the fluid coupling in 1940 to 1955 HydraMatic transmissions:
1. In first gear, the flouid coupling is driven through the front planetary gearset. Thus, the coupling is driven at about 3/4 of engine speed. This reduces “creep” and allows the engine to reach a higher RPM at takeoff.
2. When the transimssion is in fourth gear, the fluid coupling is effectively locked out and there is no slippage at all. The original HydraMatic was an engineering tour de force, a brilliant design with many patents that frustrated competitors for years.
Mr. Andrews,
Some of these issues were unclear in the previous version, which since I've taken pains to rectify. However, to your points:
1) The impeller is always driven the through the front planetary gearset, but it's only through the front planet gears (and thus in reduction) if the front brake band is engaged, which means in first, third, and reverse. (In second and fourth, the front clutch is engaged, so the elements of the front gearset are locked together in direct drive.) The purpose is as you say: to minimize creep in first or reverse with the throttle closed.
2) The coupling is never actually locked out completely. In third and fourth, torque is split between the coupling and the rear planetary gearset, but the intermediate shaft turns both the impeller and the rear planetary gearset at the same speed, so some power still goes through the coupling. Locking out the fluid coupling entirely would have been a neat trick, but the only provision for doing that is in Neutral.
It is a clever piece of work, there's no doubt about that.
The first car I ever drove was a 1958 Pontiac with Hydramatic – still, after all these years, the best Automatic ever made. Unlike modern computer controlled automatics, the H-M was analog – shifts were controlled by speed, oil pressure, and many other factors too technical for my level of expertise. In short, they made H_M seem as if it had a personality; indeed, each one I drove (Dual Fluid Couplings from 1956 on)had a distinct feel. Can’t prove it, but my feeling is that GM paid more attention to the smoothness of the shifts the more expensive the model. My friend’s parent’s ’59 Olds 98 Convertible had a 2d to 3d shift that you couldn’t feel, only hear the revs drop. My Pontiac, on the other hand, had a very pronounced 2 to 3 shift. And yes, echoing an earlier comment, they had a distinctive sound.
I’m not sure exactly how much difference there was in the dual-coupling Hydra-Matic across divisions. With some of the other transmissions, there was a lot; Turboglide and Flight Pitch Dynaflow, for example, were different transmissions, even though they were each based on the same Engineering Staff concept. On the other hand, those transmissions were manufactured by the divisions themselves, whereas the Hydra-Matic came from Hydra-Matic Division. I would assume that there were at least minor adjustments to tune the transmission for each division’s engines — the Oldsmobile, Pontiac, and Cadillac V8s in those days were quite different, and didn’t have the same torque curves — but I don’t know how extensive those changes may have been, or if there were also significant variations in internal materials, transmission mount design, etc.
Complicating that question is the fact that there were a lot of minor revisions to the dual-coupling Hydra-Matic during its first few years, to improve reliability, deal with teething problems, and so forth. Furthermore, the dual-coupling transmission was a very, very complex bit of business (part of the reason GM eventually phased it out). Minor production variations between individual transmissions might also make for small but perceptible differences in feel.
Of course, it’s not improbable that someone at Pontiac told Hydra-Matic Division, “Hey, we’d like a little firmer shift quality, for a sportier feel.” On the other hand, it’s also possible that the transmission in a ’59 Pontiac might feel a little different than one in a ’58, especially if they were at opposite ends of the allowable production tolerances. Hard to say.
Sir:
I am curious as to what oil is used the original Hydra-Matic trans. The military used the Hydra-Matic in the GMC (1950-1955) and used Heavy-duty engine oil as a hydra fluid.
Can a SAE 10W High-detergent work in this trans?
Marty Meaney
Ludlow, VT
Marty,
I’m afraid I’m a historian, not a mechanic, so I’m really not qualified to give technical advice. I’d suggest talking to a shop that overhauls early Hydra-Matics and see what they recommend.
Oldsmobile in the early days advised that 10W motor oil could be used in an emergency. I bought a 55 Olds 88 in 1978. It had 46,000 miles on it. The “original” transmission fluid did not have the red dye we associate with ATF or Dexron. I taught for Allison Transmission locally from 1978 to 2003. They reccomend Dexron for all automatics but for “off Highway” use you can use “C” approved fluids which are motor oils that have been approved by Allison
the answer is type A, which is still available today, although most original hydra-matic owners today are using dexron with good satifaction.
I am involved part-time with forest fire suppression and in that capacity have driven several different early fifties GMC 6X6s converted for hauling water. Most have 1,000 to 2,000 Imp. gallon water tanks installed. They all had the GM dual-range automatic transmission. They have got to be the roughest shifting automatics I have ever had the displeasure to drive, crashing into gear with no way to control the shift. With shift kits installed that turned them into manually shifted automatics most operators could learn to rev match the shift for a good deal smoother operation. But the factory transmission is no fun to drive.
Setting aside any maintenance issues, I would guess that Dual-Range Hydra-Matics built for heavy-duty applications shifted more harshly than ones intended for passenger cars. Heavy-duty clutch packs and a hydraulic valve body set up to take a lot of punishment aren’t known for smoothness in general; conversely, slow, slurred, gentle shifts are easier on passengers, but not good for transmission longevity under heavy loads.
It’s certainly true that some Hydramatics were rough shifting. I remember riding in a 1953 Cadillac where each upshift was very firm and gave the passengers a slight jolt in the back. However, I suspect that that resulted in longer transmission life by reducing wear on the clutches and bands.
I also remember riding in a 1949 Lincoln. The upshifts were so smooth that one could not feel them although one could hear the changes in engine speed. I remember riding in a 1950 Pontiac convertible which was just as smooth.
Why some Hydramatics were smoother than others I don’t know. I suspect that Hydramatics designed for heavy duty applications were designed to shift more firmly to reduce wear on the friction surfaces, but that would not explain why that 1953 Cadillac had such firm shifts.
the original fluid is type A, if you had to, in a pinch, you could use 20w oil, then change as soon as possible.
Can anyone here tell me how many of these 3100 pickups came with the hydromatic? I have the serial# and also the Detroit Transmission Division #off the side of tranny.Please contact me if you can help, and I will give the #s’ Thanks, Rod
I honestly don’t know. The figures GM supplied for Hydra-Matic production unfortunately did not include the truck units, and I have not seen totals elsewhere.
i have a hdra-matic that has the b&m automotive chevy bolt pattern bellhousing on it that is a very interresting transmission love to hear about the history on that transmission it is hard to get information on old transmission
Hi, I was tring to find some information, on the 1938 Oldsmobile AST transmission. I cannot understand why the information is so hard to find? I owned a 1938 Olds with the AST and a 6 cylinder engine. I did not like it and let my brother use it. He complained that, it cost more to put oil in the transmission, than it did gas. They used motor oil in the trany. So I sold it.
My first boss had a 1937 Hudson, with there Electric vacumm shift transmission, which I drove, at times. Both of these early tranys are hard to find info on.
Art Baethke
Art,
The AST was short-lived, and really quite rare. As best I can figure, production was no more than about 25,000 units, and as you saw, customer reaction to it was not great. I suspect that there weren’t too many AST-equipped left on the road by the end of the war. Also, I don’t think Oldsmobile encouraged dealers to take them apart; there was a program that let people trade a defective unit for a replacement for about $75. So, I’m not sure if there was even a dealer repair manual for it.
The Hudson system is a little better documented. (It was made by Bendix, known as the “Electric Hand.”) It was sort of an add-on vacuum/electric preselector system — it added a vacuum cylinder to the gearbox that would execute the actual shifts. There was a selector switch on the steering column that you would use to choose what gear you wanted, and another (usually attached to the clutch pedal) that would trigger the actual shift via solenoid. You could also order Electric Hand with a vacuum-operated clutch that would disengage automatically if you lifted your foot completely off the throttle and then reengage when you pressed the accelerator. (Hudson promoted the combination as Selective Automatic Shift, but it wasn’t exactly automatic, and as far as I know, the preselector and the vacuum clutch were technically separate options.) Hudson-Terraplane clubs would probably know where you could find service manuals or other technical info, if you wanted a more detailed explanation. Likewise, Cord collectors are familiar with it, as the Electric Hand was used on the 810 and 812.
I had a ’36 Terraplane (built by Hudson) with the magic hand. All of these cars were delivered with a “stick” that was under the front seat. It plugged into the transmission, just like any floor shifter of the era. That was the only way to shift the one I had. The standard joke was that the magic hand was gauranteed clear to the end of the dealers driveway, or until the first time you used it.
The Hydra-Matic was used in Lincolns from 1950 through 1954. In 1955, it was replaced by Turbo-Drive, a three-speed/torque converter automatic. I’m not sure if it was a bigger Fordomatic in concept and design; I’ve always assumed so, but I haven’t looked into it in any detail.
I have a 1940 oldsmobile 6 cylinder can you tell me what engine size it has its a hydromatic.
Manuel — The 1940 Oldsmobile six was 230 cubic inches (3,764 cc). It had 95 gross horsepower.
The 1905 Sturtevant offered a patented “automatic” transmission.This device was a was a “three speed,consisting of a series of clutches,operated by centriugal force.”The make was high dollar($5,000)and was
offered 1905,’06,and ’07 with the automatic.
Push-button electric shifting made by Hammer-Cutler came out about 1913 and appeared in the SGV,Pullman,Haynes,etc.I believe it was called
a Vulcan,and was a 4 speed.
Around 1917 another expensive car came out with
an electric transmission.The Owen-Magnetic had no gears to shift.
I took a look at the Sturtevants’ patent for the device (U.S. Patent No. 766,551, issued 2 August 1904), which is an interesting piece of work. The patent describes a two-speed transmission, although it notes that the same principles could be used to add additional speeds by adding additional clutches. I wonder how well it actually worked.
I’m not familiar with the Hammer-Cutler system or the Owen-Magnetic, but I’ll look into them — thanks!
Hmm, there was also a continuously variable transmission invented by Frank Hayes that was offered briefly in 16 and 18 HP Austins from 1933. It appears to have been broadly similar to the modern roller-type CVT (such as Nissan’s Extroid unit), but it retained a manual clutch, which had to be used for stopping and starting.
The Frank Hayes transmission may have been the basis of the Buick “Roller” transmission (which was apparently based on an outside design), but I'm still not certain about that.
Here’s an interesting page on the Hayes transmission:
www. austinmemories. com/ styled-23/ index.html
I read somewhere that the transmission was troublesome, and Austin retrofitted their standard stick-shift transmission to most of the cars that had had the Hayes transmission.
One of the British motor museums has a Hayes-equipped Austin, and they’ve somehow gotten the transmission to work, although I don’t imagine the car sees many miles.
Probably not. I’m still not entirely sure if the Buick transmission was the Hayes design. (It may have been, but there are several contradictory accounts.) It’s an interesting idea, but it took many years of metallurgy to make roller-type CVTs practical.
I’ve got car 74(1905 REO)and I forgot to mention that 1933 was the year REO offered the
Self-Shifter.I think it came on either the
straight eights or the sixes.It must have been
functional as they were offered multiple years
and there are several surviving cars with this
feature.
I recently saw a 1938 Studebaker for sale with an electric hand Bendix Shifter.It was my
understanding that it was a factory option.
The Reo Self-Shifter was introduced in the spring of 1933; it was standard on eights, optional with the six. It apparently worked reasonably well, but it was not an automatic transmission in the modern sense — instead, it was a four-speed semi-automatic, analogous in concept (though not in mechanical layout) to the later GM Automatic Safety Transmission.
One of the things that can be very confusing about this subject is that for many years, the terms “automatic” and “self-shifting” were applied rather loosely to all manner of transmissions, only a few of which we would consider automatic. Some were semi-automatic transmissions, like the Reo Self-Shifter or the AST, while others were preselector transmissions along the lines of the Cotal or the Bendix Electric Hand. Preselectors were not automatic except in the sense that the actual gear change was executed by some kind of remote mechanism, rather than through the direct movement of a shift lever or cable linkage. The driver still had to decide what gear to use and initiate the actual shifts. (I would draw an analogy to the way the term “automatic” is applied to pistols or rifles; it’s often used to describe weapons that are simply auto-loading, rather than truly automatic.)
Like the AST, a lot of the semi-auto and preselector transmissions retained the clutch pedal (although in a preselector transmission, it was essentially converted to an engagement switch to execute the actual shift) and generally required at least some manual shifting. Few of them did well commercially, particularly during the Depression — buyers were reluctant to spend the extra money and risk the potential reliability hassles if they still had to change gears and operate a clutch pedal.
GREAT GREAT ARTICLE…It helped me a lot
I have a 1941 Cadillac Series 61 (Model 6109D). Is my shift pattern as shown above?
Neutral, Drive, Low, and Reverse
Do I need to pull in (or push out) the shifter to get it into any gear?
Once again Thank you
My Great Uncle, Kenneth E Snyder was an engineer at GM for 30 years. He worked there from approx 1936 to 1966. On the back of the watch given to him by GM states: Service Award 1961 Detroit Transmission Division GMC K.E. Snyder 25 years. My Great Aunt Marion, his widow, gave the watch to my son after my uncle died. Uncle Kenneth was on the team which developed the hydramatic, from all the stories that I’ve been told, plus, the last time I was at his home in FL , I saw the award GM gave to him which was a small trophy of a transmission. I wish I knew exactly what it said. Our family has always been very proud of his work at GM and always knew how smart and gifted a man he was. I’m sure the whole team knew they would never receive individual recognition, auto companies never worked that way. Anyway, I just thought all you car people would enjoy hearing my story. My brother, Kenny, lives in FL and really knew my Uncle MUCH better than I and would be a much better source of information.
Hello! Certainly interesting historical information here. Thanks for the attention to detail. I was hoping that someone here might offer some advice on an HM unit that I’ve put in my 41 Cadillac 6227D coupe. The car was originally HM, rare for a coupe and I’ve just about completed a frame off restoration. I found an older gentleman who used to work for GM HM to do a rebuild of a ’48 HM unit (41’s are really not too good). He did it and I have pictures of each step of the rebuild. He’s now passed on. I have installed the unit, filled it properly, and adjusted the bands according to the 48 shop manual. The car moves in Hi, Lo, and R. R seems fine but the manual shift out of R is a little tough. [b]Lo has one speed, no shift and I’m not sure if that’s correct. In Hi, the car takes off, but misses the shift and feels as if it’s shifted to N. [/b]I’m hoping that an adjustment to the throttle control linkage may be the issue and that it’s not something internal. I’ve tried adjusting the bands 1/2 and 1 full turn tighter with no effect. There are no strange noises associated with the problem. Again, in Hi/D upon acceleration the car gets up to about 10mph and then feels as though it shifts to neutral rather than up shifting to 2nd.
Any help would be appreciated.
Thanks,
Ken Karrer 1941 Cadillac 6227D coupe
Ken,
I’m not qualified to perform troubleshooting or repair advice, but to the best of my knowledge, the early Hydra-Matic IS supposed to shift between first and second in Lo range (comparable to the “2” position on later automatics). I would suggest digging up a shop manual for the ’48. You might even find at the library — I was able to find and check one out while researching this article.
Ken
It’s been a while since your post, so you probably got your questions answered … but if you’re still having issues here’s a few comments.
I agree with Aaron, don’t want to try to diagnose your problem here … but I can’t help making a couple quick comments …
I’m an Olds guy … so I did a little quick research on your questions since Olds, of course, used the same Hydra-Matic as Cadillac (or mostly the same as far as I know). We had several Oldsmobiles with Hydra-Matic back when I was in high school, that I used to work on. And I have always been interested in transmissions (I’m an engineer by training, though I don’t work in engineering).
It sounds like you have two issues/questions:
1) What is correct operation in “Low” range?
2) Bad 2nd to 3rd shift (goes into neutral)
Item #1 – In Low range the trans should start in first and shift to second, but at a somewhat higher speed than in Drive range. If shifting into Low range at cruising speed, the downshift sequence gets a little more complicated, but that’s another discussion.
Item #2 – The 2nd to 3rd Hydra-Matic shift is quite complicated due to the required timing of the apply and release of the friction elements. The 2-3 shift is probably the major contributor to the Hydra-Matic’s reputation as having harsh shifts.
In second the front planetary is locked up with the clutch applied, the rear planetary is in reduction with the rear band on. In third this is reversed: front planetary goes into reduction by releasing the front clutch and applying the front back; the rear planetary up shifts by releasing the rear band, applying the rear clutch. If the timing of these “release” and “applies” is off then one gets 2-1-3 (front planetary downshifts faster than rear planetary upshifts) or one can get 2-4-3 (rear planetary upshifts before front planetary downshifts) … or 2-neutral-3 if either the front band apply is too slow (after the front clutch released) or the rear clutch apply is behind the rear band release. As I think about it, mis-adjusted bands seems like the most likely cause of a 2-neutral-3 shift, though it isn’t explicitly listed in the Diagnosis section of the Service Manuals I looked at. If you have a mis-adjusted band (singular) you should be able to figure out which by observing other shifts, but not necessarily.
Despite having adjusted the bands once, I’d recheck adjustment.
If bands are ok, next stop would be looking for sticking valve in valve body.
If you’re still having problems .. post something here and we can communicate outside this forum.
Jim
This article is fascinating to me. At my current age–66–I drive a Grand Marquis LS 4 door sedan. What I would give for a dual range hydramatic drive transmission. Before starting at OSU in Stillwater, OK, I worked at Edward’s AFB and purchased a fast back 1948 Cadillac with hydramatic drive. I wish I had never parted with that car! Thanks to the authors of this engrossing article/history!
This article contains much useful information about the Hydramatic. Earl Madman Muntz put the Hydramatic in the Muntz Jets from 1950-1953. A project Jet I’m working on has a 337 cui Lncoln flathead mated to a 1952 Hydramatic. Take 8 qts. of Type A fluid througha funnel in the dipstick hole. Hope to get it running. Right now it seems to have mineral oil in it. I guess it was used back in the day.
Hi,
first of all thanks for the nice reading, it’s the first time i see this site, i’ve run into it casually!
concerning:
"The best I can suggest is contacting the GM historical archives and seeing if they have those records."
I’m afraid they don’t have that information, I already checked.
Bye,
Jerry
I really enjoyed the story and have a question about the production facilities. GM built the Livonia facility in 1948 according to the Livonia history website. I’m curious as to where were they produced from 1939 until 1948? Was there a dedicated factory?
You know, I don’t know. That’s a very good question.
Ah-ha! Finally something I can contribute to! lol. I was actually searching for info on the Detroit Transmission Division of GM when I stumbled on this page. Except for the real technical stuff, I actually found it very interesting! Especially the GM history. (my dad retired after 42 yrs at GM and my mom got an early retirement at 27 yrs so we lived and breathed GM. lol)
Anyway, the plant you are asking about, where the transmissions were produced starting in 1939? That plant is the reason I’m scouring the internet on a sunny Saturday in Detroit. The first building GM built the Hydra-Matic transmissions was located on the corner of Riopelle and Farnsworth in Detroit. It was near Warren and I-75, on the east side of Detroit. The building started out sometime around 1917 as Fisher Body Plant #10, where presumably, Fisher made car bodies for GM and many other car companies at that time. (until a deal was struck in 1926 for them to become the in-house body maker for GM, according to Wikipedia) It’s a little unclear to me if they continued to make bodies for GM there until 1939 when the Detroit Transmission Division was formed and housed there or if there were other operations in between.
The H-M was built here until the new facility in Livonia was ready in 1949. We all know what happened within a few years of that. Below is a link to a brochure on the GM Heritage Center’s website, showing a timeline history of the Detroit Transmission Division, with the first line saying that the division was organized “in one-third of an old six story building on Riopelle Street in Detroit”.
[www.gmheritagecenter. com/docs…elcomesYou.pdf]
Although I don’t see a date this was printed, it was after 1957 since they list that year on the timeline but I’m sure it was before 1960. But what I find funny is, they write off the original building as “old” when, in 1939, it was just barely over 20 years old and even if it was 1959 when they printed this brochure, it was just barely over 40 years old! Now, this was an Albert Kahn building, top of the line, re-enforced concrete that was made to stand the test of time. But, because it wasn’t the “new hotness” like the Ypsi plant they’re advertising in the brochure, it was just an old building. I’m surprised they even named the street it was on. Anyway…
After Detroit Transmission moved to Livonia, eventually GM moved some Cadillac production in there for quite a few years and then used it as a warehouse until the early 80’s. And GM people know what happened then; plants started closing left and right. My parents were both working at a plant that closed in 1987, I believe, and I think that one lasted longer than some others. (Luckily my parents both ended up at the Warren Tech Center and were able to leave GM on their terms, not GM’s.)
It’s a little fuzzy at this point, as far as this plant but I know a food wholesaler was in there for a while in the 90’s – Total Foods, until the people I work for bought it in 1998. They used it as a rental property (it was 500,000 sq ft) but used a big chunk of the space as a warehouse, shipping facility and eventually an assembly operation was added, for the manufacturing company they also own, Palmer Distrib. This is the company I actually work for but I am basically the admin asst for both of the owners, but in their main office in St Clair Shores. I never worked in the warehouse.
This building did stand the test of time, for a total of just under 100 years. I’m sure it would have lasted a lot longer too but a nasty, nasty fire took it on February 5th, 2014. It started on the 3rd floor and swept thru the building like it was a house of cards. It burned for 5 days, smoldered for 2 days and then burned for 2 more days. Someone close to the Detroit Fire Dept put together the video at the link below, if you’re interested in seeing the first couple days in 7.5 minutes: [youtu. be/wKe63ZdRsec]
That happened at the beginning of February and the last part of what was left of the building was knocked down at the end of May. I guess they’re still cleaning up the rubble and rest of the property. I think I heard mid-July is when they expect to be done with that.
So, now I’m helping someone do some research on the history of the building for a paper they’re writing. This is what brought me here. A lot of the early history I shared came from that other person’s research. I am obviously more familiar with the more recent uses of it. I have one more link to share, in case any one is still skeptical that this building I’m referring to is in fact part of the Detroit Transmission Division of GM….
[www.emporis. com/ building/detroit-transmission-division- general-motors-factory-detroit-mi-usa]
Christine,
Thanks for the information! I will have to check this out.
Aaron,
I’m glad I could contribute to the conversation. My friend is supposed to be posting his paper about this building on his website within the next couple of days. I will post the link to it once it’s up. It should include pictures from after the fire but before the demo – inside and out.
Thanks!
Christine
Thank you so much for all of your effort here, and with the article!
In my years( 83-03) as a New Car Manager & U/C Mgr in a multi-point GM store, I guess I have driven everything G.M. ever produced domestically, say from 1970 and newer, and many a lot older. (just a guess) I never thought to bring this up to the Factory rep, who may have been able to comment on my question. MY mother has driven Cadillacs going back to 1966 through today all except one of them being Sedan Deville’s, the exception was an ’82 or ’83 Coupe powder gray beauty, I was not permitted to drive them, of course, but now at 53 she may allow me to gas it up for her.. Loves her Cadillacs!
*
Why does a Cadillac tranny shift feel differently to me, it is just a unique feeling of smoothness & power? Now, that’s me. Have you ever heard this or experienced for yourself?
Well, even where different division share the same basic transmission, each division’s engineers have to tune the transmission to suit the torque characteristics of the division’s engines and their cars’ identity. With planetary gearset automatics, the transmission’s performance depends a lot the timing and firmness of clutch engagements and so forth; slower, slurred shifts tend to be smoother but less efficient, while firm, fast shifts are the reverse.
Finding the right balance for street cars is as much an art as a science because it tends to involve a lot of compromises. GM divisions tuned the Turbo Hydramatic (which most Cadillacs between 1965 and 1981 used) in various ways; some were very smooth, others produced back-slapping shifts (intentionally).
Cadillac had two advantages during much of that time: first, their engines generally had lots of torque, and second, Cadillac customers weren’t looking to extract maximum quarter mile performance. As a result, their engineers could strike a more relaxed balance on shift quality without sacrificing too much performance. They could also more thoroughly isolate the transmission mounts from the car in ways that Supercar drivers would’ve complained about endlessly.
I should add that even within the individual divisions, there were multiple variations of the Hydra-Matic for different models or applications. For example, in the early fifties, Cadillac had three or four variations. The ratios were the same. I assume differences were chiefly in the clutches and control valves to handle different load requirements, although toward the end of the single-coupling era, some applications also added an oil cooler.
This is an interesting site; I just discovered it. It is loaded with good historical information. Obviously a lot of work and study has gone into it.
Some decades ago, I rode in a 1953 Cadillac with Hydramatic. The shifts were VERY firm; it was impossible not to feel them. The owner, who had bought the car as part of an estate, stated that it was intended to have firm shifts. On the other hand, I remember riding in a 1950 Pontiac convertible. The shifts were so smooth that you absolutely could not feel them although you could hear the change in engine speed. The same was true with, I believe, a 1949 Lincoln, which a landlord owned.
I’ve ridden in many cars made from 1950 through the early 1960s with Hydramatic. Even before the 1956 Jetaway fiasco, there was considerable difference in how they operated. I tend to favor somewhat firm shifts the theory being that firm shifts result in less wear and longer life of the friction surfaces.
It would be interesting to take a 1955 Hydramatic, remove the hydraulic “brain”, and replace it with electronically controlled solenoids.
The firmness of shifts in the early Hydra-Matic had a great deal to do with how precisely it was adjusted, which was something of an art. Compared to later transmissions like the TH400, there’s just a lot going on mechanically with each shift, so slight timing variations can really add up.
I’m very pleased you acknowledge the part the English Daimler Company played in the birth of Hydramatic. What Kettering & co aimed to do was give the Daimler transmission a brain, and the hydraulic servo system, controlled according to speed and load, was that brain.
But the basis remained the same – four speeds, fluid coupling and steering column lever control, so allowing three to sit abreast in front.
The Rolls-Royce made hydramatic was licenced for 5,000 units per year, I believe, so they sold hydramatics to Jensen, Armstrong-Siddley and I think a couple of other posh low-volume British marques.
I’ve enjoyed reading several articles on here today and they’re mostly very good, but I have to take some issue with parts of this – mostly detail and tone… possibly from a need to brutally economise to get it within 4 pages despite having an obvious wealth of information to include, IDK…!
First up… yep, “sliding-gear” manual transmissions ARE desperately out of date, obsolete, irrelevant, etc… and indeed, they were so well before WW2. One of the later cars to continue using actual “crash” gears, with straight-cut gears moving around in relation to one another, was actually the Ford Model A with its 3-speed manual version (not all old Fords were 2-speed epicyclic!). Even before synchromesh took hold, manufacturers had gotten the idea about constant-mesh, helical gears and dog clutches – the kind of setup still used in most motorcycles to this day, in racing cars (with straight-cut gears), and for a surprisingly long time in certain cars like the Fiat 500 (which never received synchro as far as I know – downshifts generally needed a blip of the throttle, but upshifts were merely a bit rough rather than grinding).
The main similarity between an old sliding-gear transmission and a modern 5-speed manual is that you operate it yourself with a pivoting stick, and that it uses gears acting directly on each other. The actual mechanical actions that take place when you change gear bear much less resemblance to each other, and the modern day iteration is perfectly fine to and relatively idiot-proof. The newest 3-shaft versions even lend themselves to dual-clutch semi-automation with the minimum of actual alterations… drill a couple holes, extend the shafts out, split the clutch into two, and shove servos on the release and selector arms…
(OK, that’s not 100% technically correct, but you get the idea)
In said more modern designs, the gears stay static and never go out of engagement, and so don’t need slamming back INTO engagement, which causes the grinding, chipping and balking. What moves are the selector/engagement/dog clutch rings (call them what you will), sliding along and spinning in engagement with the splined parts of the input shaft (the input-side gears themselves being freely-rotating on a smooth part of it). When you select a gear, the gearstick moves the selector arm which moves a selector fork on its own shaft, driving one of the rings towards one or other of the gears facing it. Large, rounded-off “dogs” protruding from the face of it then mate with similar outcrops (or indents) on the face of the gear and form a positive engagement for transmitting drive. Thanks to their size and shape, and the lego-brick mode of action (like jamming a stick in a bicycle wheel, rather than trying to get two bike wheels to line up and mesh nicely with each other… whilst both are spinning), this happens rather more easily, smoothly and quietly than in a crash box – though it still often requires the two sides to be turning at least roughly at the same speed. It’s easier for upshifts as the engine speed naturally falls as you take your foot off the gas (and the input shaft speed falls relative to that of the gear assembly even with the clutch depressed) and it’s just a matter of getting into the rhythm, whilst downshifts demand a deliberate raising of engine speed… particularly an issue if you’re trying to brake (doubly so on a steep hill) at the same time!
(For some reason this isn’t too much of a problem with motorbike transmissions… even though they’re sequential, without even any neutral gaps between gears other than 1st & 2nd for you to blip the throttle in with an engaged clutch, downshifts are usually a perfectly simple press of the gear pedal… I suspect there’s some additional trickery in how they’re made, maybe with a little more play engineered in – clutchless upshifts usually come with just the slightest of jolts suggesting that; additionally, despite their higher engine speeds, there’s usually a more-than-compensating reduction gear between the crankshaft and gearbox input (with then a generally smaller step-down from output to the also-larger wheel), so everything turns rather slower, which together with the overall much lower weight (overall and per-gear) leads to rather less inertia and probably a minor auto-synchronising effect due to the whirling oil bath everything’s soaked in. All I know is I’ve tried to figure it out with mine, with the help of an exploded diagram from the workshop manual that shows how there definitely aren’t any synchros in there (vs the similar diagram for my car), and ended up just giving in and consigning it to the “life’s little mysteries” cupboard.)
Early synchro boxes tended not to have it on 1st gear for several reasons … first being one of strength. Early synchros weren’t as strong as all that, and drivers were used to giving the lever quite a bit of welly, especially if it balked (which is to be expected when doing a difficult change with a weak synchro – just hold pressure on the lever for half a second more and it should go in…). This, along with the greater torque and rpms that would get shoved through them (even when fully engaged) thanks to the extreme gearing in 1st and the multiplying effect of a suddenly-engaged clutch meant the potential for them to break was rather higher – but plain dog-clutch engagement is very strong, which is why it’s used for race cars and for all but the most modern auto-shifting trucks (and this is why truckers have a gear-jammer reputation… they don’t have synchro and so have to become skilled to avoid dying on long downhills!). It’s also why you can identify when an old car is in 1st gear from the whining noise… the same stresses applied for the gears themselves as well as the synchros, and the one benefit helical-cut gears don’t impart, despite being smooth, quiet and efficient, is strength. Straight-cut gears are rather stronger – and noisier.
On top of all that, in such transmissions, Reverse and 1st usually shared a selector shaft/ring, being directly opposite each other on the H-pattern, especially with 3-speeds. And you really don’t want to run the risk of the accidents or mechanical mayhem that could happen from accidentally and far-too-easily shifting directly from one to the other whilst moving at some speed… versus that, a bit of a grinding, rattling noise telling you “don’t do that, idiot!” is positively welcome. (This is why reverse is still rarely a synchro gear even in modern cars – in fact, some manufacturers even deliberately engineer it with an otherwise primitive sliding-gear arrangement, specifically so you can only engage it with any ease when stopped or moving very slowly, even if already moving backwards. This is the source of the “crunch” noise you get if you try to slam into reverse too quickly after braking fairly hard from speed… even though the clutch is dipped, the input shaft hasn’t stopped spinning yet, and the reverse idler gear is rather unhappy about being shoved into engagement with it vs the stationary output… the sound has a rapidly dropping frequency as output actually acts as a brake on the input shaft, with the idler gear as its proxy, and it all slips into place only once the input has slowed quite a bit).
Another concern was one of cost. Synchro was both mechanically complicated, and had to be licensed thanks to the patents and copyrights. The less synchronisers you could get away with using (e.g. a single shared one on the 2nd/3rd selector ring in a 3-speed-plus-reverse box), presumably the less royalties you had to pay, as well as the fewer expensive, complicated and fiddly to produce and install parts you had to fit.
Don’t forget that a lot of old cars were relatively slow, and also relatively low- (and wide-) geared compared to present day ones, which is why people came up with overdrives. First was usually only for starting off, low speed manoeuvring and hill climbing, and might be all over and done with by 15, MAYBE 20mph (why do you think the old automatic manufacturers thought little of habitually starting in 2nd?). General motoring was almost all achieved in 2nd and 3rd, and quite a few cars even in the 4-speed, all-synchro era were strangely proud of being able to boast that they could run as low as 10 (or even 5!) mph in top gear and still pull back to high speed without needing a downshift (even if it took a whole minute to get up to 50mph again). 2nd gear was plenty flexible enough for most low speed situations. Then – as now, actually, thanks to much better low-end torque and stronger clutches – there would be little call for downshifting into 1st on the move, rather than after having slowed to walking pace (which would allow a smooth re-engagement at idle) or a complete halt, because unless you were approaching a hill that you knew would require (a throttle-blipping double-declutched downshift into) 1st to surmount, there wasn’t that much real-world performance benefit to be had, and it would only be useful for descending the very steepest of hills, the ones that would cause you to stop briefly at the top of them and take a deep breath. Thus, there was little actual need to fit synchro, and again, it could even be regarded as a safety measure to stop gung-ho motorists from slamming the shift into 1st at far too high a speed and blowing the engine – rather more likely with 2nd-1st (or 3rd-1st!) than 3rd-2nd. If you couldn’t rev high enough in neutral to enable the speed-matched downshift into 1st, then you probably had no business being in that gear.
Now compare all that to the behaviour of the contemporary automatics … there’s not that much of a world of difference, is there. There’s a lot of the same problems, but what the autos do is let the machinery take on all the dirty work, even if it does it about as badly as you do ;)
These days, manuals are somewhat more developed and refined … though I’m personally not sure what to make of the rather flimsy, cable-change affair in my most recent car. It seems like a step back from the solid, positive-feeling rod change one I had to give up. Makes it harder to do a quick, smooth shift thanks to how notchy and uncertain it is. But, that’s a minor gripe really, on the whole it works fine, and demands nothing more than “realise when it’s time to shift… dip clutch… move lever… lift clutch”.
Moving on from that, we also seem to have skimmed over the whole “epicyclic manual and semi-automatic” thing other than the brief mention of preselectors. If you consider the quite common* addition of a 2-speed rear axle to the Ford Model T, and the pedal rearrangement mods that went along with, these 4-speed variants could be sort of considered such. The “clutch” action was integral to the gear selection movement, whether by footpedal or hand lever; basically there were four different clutches (five including reverse) you could choose to engage by operating the gearshift mechanism, and going into low or reverse from a standstill involved careful take-up of engine drive so as not to stall it… but after that, it was pretty much the same as various other semiautos that came after, like the saxomat. Move the lever/pedal to the next position, and you’re in gear. With the simple addition of, say, a centrifugal clutch that would allow you to come to a halt with one of the gears still engaged, maybe a rotary-sequential selector device to co-ordinate the 2-part shifts (particularly between High-Under and Low-Over), and some device that would move the selector in response to the difference between driveshaft speed and throttle setting, it would have been a “proper” automatic with no hydraulic parts at all, and the world could have been rather different. As it was, it stayed effectively “manual” in all its guises despite strongly resembling an “AST” where the driver had to deliberately regulate the uptake of drive from rest, and command the in-range (1/2 and 3/4) shifts as well as the between-range (2/3) one.
(* Damn it, I can’t remember the general name for them right now – as a single company made a decent majority of all those fitted – but I’ve probably seen more in-game models and example videos on youtube of T’s and TT’s with 4-speed twin-epicyclic transmissions fitted than I have ones without. Generally it was in the form of a “splitter” type gear rather than an overdrive, thanks to the T’s rather wide ratio gap between low and high; the axle would offer something in the region of a 1.66/1.00 gearing (relative to whatever the actual rear diff ratio was) to the T’s default 2.73/1.00, with a choice of either multiplication or division relative to the norm. Meaning you could have a roadster with a multiplier one giving 2.73 – 1.66 – 1.00 – 0.60, for better top speed and improved lower-mid-range acceleration and hill climbing, or a truck with a divider giving 4.53 – 2.73 – 1.66 – 1.00 for better load hauling on steep grades and less tendancy to rapidly slow to the original low-gear speeds on shallower ones… Oh, and of course, two reverse gears. Rather pointless on the roadster unless you liked scaring the crap out of your passengers, but pretty handy for the truck. Naturally, if you couldn’t be bothered with the more complex twin-shift arrangement but couldn’t afford the time/money/care to modify it into an H-gate setup, you could just choose to leave one of the transmissions in its high or low position, and only shift the other one, reverting to either a 2.73 or 1.66-to-one 2-speed modus depending which you picked (…probably safest to shift the original transmission, I would guess, as the axle would be intended to experience at least some of its changes whilst the original was in neutral going between Hi and Lo at the combined shift point?).
Interestingly, some drag racers still use a similar sort of system, when uprated Powerglides and such no longer suffice – a bunch of daisy chained 2-speed epicyclics, with a single clutch that locks them into “forward reduction” when engaged and “direct” when released, and a regular plate clutch (or heavy duty “dumb” TC) at the head of it all. Bunch of little levers in a row forming the control panel. Shove them all (say, 2, 3, or 4 of them depending on your power and expected top speed) forwards, roll to the start line, do your burnout… rev up… lights go green, dump the clutch, then over the next three seconds or so rapidly slam each lever down in turn, jumping up through the ratios quickly, seamlessly – and completely manually – as each reduction clutch is released. Usually a uniform sequence is used, but there’s no reason a roadable dragster couldn’t have a back-and-forth type splitter gear in there which could be rocked one way and the other whilst the other, wider gears were being mostly changed in just one direction.
And, well…
All this sort of begs the question…
Given how the gear selectors in my bike box are quite happily synchronised in their operation by a single up/down footpedal despite the box internally more resembling some notional 5-speed version of the Fiat 500 (three different selectors, whose forks move back and forth just as if they were being pushed and pulled by a regular lever in an H-gate)… and operators of the Ts and TTs with 2-speed axles presumably didn’t suffer badly from jolty shifts even when having to personally synchronise a pedal and a hand lever… and even just now I manage to envisage controlling and tightly synchronising the latter using the former’s cam-like peg-in-groove shift barrel and some simple adaptions to make the output act on brake bands instead of dog-clutch selector forks…
…how come the makers of the AST and the 4-speed Hydramatic couldn’t dream it up? Heck, the former was almost there, they just needed something to hook up that 2nd-3rd change.
No complex hydros, no strange lag, anything like that. The barrel turns, the pegs located in the matching grooves are driven one way or the other, and the attached mechanical parts are forced to move in perfect synch. In fact, the shifter itself will balk and refuse to move unless they do, as unsynched movements will jam it before it can turn any further (an actual safety feature on my bike uses a torque-sensing ratchet pawl to do that on purpose to prevent you downshifting more than two ratios in one go without letting the clutch back out, as a 3-gear jump would generally represent a difference in rpms that would cause engine damage, dangerous loss of rear wheel traction, or both; soon as you let the engine hook up for a split second, it frees right back up). Very easy to control with an automated system because all it needs to know is whether there’s a need to go “up” or “down” from it’s current condition (say each shift takes half a second, and it needs to jump to 1st from 4th; “need to go down”… it shifts 4>3 … “need to go down” (still)… shifts 3>2 … and there’s still that pesky “need to go down” signal, well ok let’s go 2>1… and hey, that was “only” 1.5 seconds, probably quicker than a human would have figured it out!), and what the limits are (hi/lo ranges, or e.g. “S” or “2” position – and of course, the point at which 1st becomes Neutral becomes Reverse). The only outputs from whatever control logic is used – mechanical, electrical, hydraulic – would be “up” and “down” from the current position (and, if it’s an actively clutched setup rather than a fluid/centrifugal/TC one, some kind of release bearing servo). Could even add a tiptronic-style sequential manual shift option (either with positions 1-4, or a ratcheting push-up/push-down motorbike/flappy paddle setup), and have Park as something triggered by pulling the lever sideways from N.
But then, I am looking at this with 20/20 hindsight and the eyes of someone who’s already versed in everything that has come along since. I have a feeling bike manufacturers themselves didn’t hit onto the sequential/barrel shifter idea until after WW2 anyway. And although it could have been achieved just as easily with cams on a rotating shaft (produced using the same tooling as the engine camshafts!) actuating other parts in synch, and it’s almost certain the Victorians did similar things in order to make steam engines work properly, both of those would probably have been viewed as primitive, old tech solutions that should be left in the coal-burning age rather than as practical options for the modern early-20th-century gasoline-burning automobile maker. Hydraulics are where it’s at, don’t you know! So much better than cams and cables and all your old rot. It’s alright, we’ll work this out sooner or later…
I ride motorcycles and generally do the upshifts without the clutch. I can also do downshifts without the clutch but feel that that is more likely to cause problems which I prefer to avoid.
Motorcycles with chain drive have a cushioned sprocket on the rear wheel. I’m sure that that helps to smooth out clutches shifts. Also, the dog clutches have considerable play which is obvious if you try rolling the bike forward and backward while in gear with the clutch engaged. That play no doubt helps the dog clutches to engage even when there is a significant speed mis-match. The lighter rotational inertia of motorcycle engines also helps, at least on sport bikes or similar bikes. On a bike with a big heavy engine, such as a Hardly Davison, clutches shifting might not work so well.
On cars, there seems to be very little play within the manual transmission so speeds must be pretty well matched before engagement can occur. The synchronizers will block engagement attempts until they have forced speeds to match.
My father’s first car with automatic was a ’53 Pontiac–with Hydramatic, of course. He was a buttoned-up guy and usually not prone to getting openly excited about things. For weeks after getting the Poncho, he was telling people, “You just step on the gas, and it goes! You step on the brake, and it stops!”
Automatic transmissins in Indian traffic conditions have not proved effective due to high congestion oftraffic, unruly traffic and slow moving traffic too. Frequent shifting of gear ration result in fast wearing out of brake bands and aslo clutch plates of interlocking clutches. Slow speeds cause slippage in fluid couplings leading to poor fuel economy. Thus Indians mostly prefer manual shift gear boxes. Earlier we had ceilo a korean car with auto transmission that failed in indian conditions. Many customers got manual shift gear boxes fitted.Ofcourse some woman driversprefer auto gears and but are limited.
Automatic transmissions have generally been slower to catch on in markets where fuel prices, tax considerations, and available income favor cars with smaller engines, in part because automatics have traditionally been a serious handicap for smaller engines in performance and fuel economy.
I’m not terribly familiar with the Daewoo Cielo, but if I understand it correctly, it was essentially a restyled version of a car sold in the U.S. as the Pontiac LeMans, fitted with a smaller 1,498 cc engine and offering a rather elderly three-speed automatic. With only 94 lb-ft of torque, I imagine the automatic didn’t do much for either acceleration or fuel consumption.
I had a 64 Olds 88 in the early 80s. The HM was worn and 1st gear overran. I was annoyed to learn the HM had no exxternal means to adjust the bands. One trans mechanic said they basically self -destruct.
The ’64 Eighty-Eight did not use the original Hydra-Matic, but the later three-speed torque converter version, introduced in 1961. (That transmission is described in Part 2.) One of the things the later Hydra-Matics attempted to address about the original four-speed H-M was that it was very sensitive to proper adjustment of the bands — if they weren’t set up just right, it would throw off the shift timing and make the shift jerky, which was the cause of many complaints.
Very interesting article and comments.
Does anyone know where I might find a new Speedometer Pinion for a 1948 Cadillac Fleetwood 60 Special Hydra-Matic Drive?
Thanks for any information anyone can provide.
Mike
I don’t sell parts, so I’m afraid I can’t help with that. Sorry!
W. G. Wilson, did I miss mention of him?
Even worse a complete omission of that genius of the 20th century Frederick Lanchester!
http://www.lanchesters.com
Wilson is mentioned, albeit briefly, Lanchester is not. This article is about the origins of GM’s Hydra-Matic; it is not and does not purport to be a complete history of the automatic transmission or the fluid coupling, which would be an account several times the present length. The Wilson and Cotal preselector transmissions represent a separate (if not unrelated) and complicated subject in and of themselves.
I have a 1954 Rolls Royce Silver Dawn that uses the “Hydra-Matic” built by RR under a license from GM. The transmission was replaced under warranty in 1964, at about 75,000 miles, and the second transmission is still doing fine at 150,000 miles. An interesting feature is a servo-power takeoff on the side of the transmission that provides mechanical power assist to the braking mechanism. A very complicated set of levers and rods tying that all together but I’ll have to say it works very well.
The 1958 Pontiac would have had the second-generation Hydra-Matic, which Pontiac called Jetaway. It was similar to the original Hydra-Matic in some respects and quite different than others; it was intended to provide smoother shifts than the original. There’s more about that transmission in the second part of this story.
Wow, what a great discussion of this great Automatic transmission. I had 3 cars with the Hydra-Matic, a 1950 Oldsmobile, a 1955 Chevrolet with a 394 Cu. In. Olds engine & another 1955 Chevrolet with a 427 Chevrolet Engine. I was totally fasinated with the Hydra-Matic! I used to study the Hydra-Matic’s power paths shown in the motor’s manual. That really fascinated me! After studying those power paths a bit, it became obvious to me, that the 2-3 shift was VERY complicated & required exceptional timing & coordination control to make that shift feel smooth! Jim Geiger’s post above explains this very well. Some of the timing trickery used by the Hydra-Matic was the 2½ turn front band & a drilled hole in the side of the front band’s apply piston bore to give a timing signal to the valve body to aid the timing of the other 3 frictional elements that were ether applying or releasing at that moment! A mis-adjustment of that band will cause 2-3 shift problems. Has anyone seen the Hydra-Matic’s used in some WW2 tanks? They had a deep, cast iron oil pan with cooling water ports in it.
I’ve only seen the AFV version in photos, but I’m still amazed they made it work. Not at all an obvious idea, although I can see obvious benefit to not having to manually change gears while driving a tank with people shooting at you.
Hydra-Matic is one of those devices that belies the idea that mechanical analog systems are simpler than electronic ones. There’s a tendency to romanticize the purely mechanical as always meaning simple, rustic, rugged devices that even the local blacksmith could fix for you, but with stuff like the early Hydra-Matic (or most mechanical fuel injection systems for street cars), that’s obviously not the case!
It would have made things a little easier, I suspect, if the original Hydra-Matic had used one-way clutches rather than bands, but Oscar Banker’s patents in that area would likely have made that difficult, which may be why GM didn’t go that direction until the fifties, after the applicable patent terms had run out.
GM still builds Automatic transmissions with potential timing problems. Their popular 700R4 (82-93, Hydraulic control) & the newer 4L60 (basically the same transmission with computer control), if the 2-4 band releases before the 3-4 clutch applies, it feels like neutral & if the reverse happens, it goes into 4th. (overdrive) momentarily.
Must be some powerful reasons to build a transmission that way. Economics? Compactness? Efficiency?
Bands do have their advantages. The limitation of using a one-way clutch as a brake is that unless you do something clever and complicated, its behavior is completely dependent on load conditions. Brake bands allow positive control over when the brake engages or disengages, which allows more flexibility of transmission programming (whether analog/mechanical or electronic). Of course, the trade-off is that you then HAVE to control the engagement and disengagement and manage the timing and things can get wonky if the timing is off.
Also — and this is a point I’ve only recently come to grasp, working on revamping the article about later GM automatics — is that one-way clutches will automatically disengage if the driveshaft overruns the engine, which is a big problem in mountain driving, particularly with vintage drum brakes. Later Hydra-Matics (both Controlled Coupling Hydra-Matic and the three-speed Roto Hydra-Matic) dealt with this by also providing bands and/or disc brakes that worked in the lower gear ranges (for example, in a dual-coupling Hydra-Matic, selecting D3 or S range will cause the rear band to engage in second gear), but that’s an expensive way to go.
I had a 55 Holiday Super 88 with a strong 4bbl rochester. The old car had 80k but ran stronger than anybody else. The trans shifted and was done before friends had their foot on and off the clutch. Solid off the line feel was impressive. Most of us teenage drivers loved the strong. shift feel. That detent downshift was brutal and should be illegal but really raised friends admiration. Yes, that trans is for the teens as far as I am concerned. I even snapped a driveshaft companion flange with the kickdown.
If the start battery was low the ole car could roll down the grocery store parking lot slight incline then start at 6 mph when taken from neutral. What an impressive machine for that ere thanks to Highly-Dramatic. Nobody said things like “Slip and slide with Powerglide” or Dyna-Slow. Heavy as it was I felt like I had a real working machine under me as I drove that car. Never had another like it.
NOSTALGIA from 1964. Rich kids had a new GTO, but I had the Olds HYDRAMATIC 4 speed with 2 clutches and 2 planatary gear boxes to make 4 cleverly arranged gear results. I knew the guts of that car! 324.2 Cubic Inches with Rochester 4GC and Four speed Hydramatic solid shifter that is good enough for trucks.
So much for my feelings about automatics.
Thanks for listening!
Aaron,
Great article! But just a question about synchromesh gearboxes. You cover the design of the original synchromesh gearboxes, but there was a later (and very clever)improvement called the Baulk Ring Synchromesh, which is what is used in all manual transmissions (except motorcycle trannys) to this day. As I understand it, this was originally designed by Porsche in the 1950s. Is this correct?
Yes — this point is briefly mentioned in the article. I didn’t look up the patents for the balk-ring system — I didn’t want to get yet more sidetracked — so I don’t know the numbers or in whose name they may have been filed (although I assume the assignee was Porsche AG). However, it was written up in some detail in the automotive press around 1951 and you could probably look up the relevant patents relatively easily if you’re curious about it.
Thanks Aaron….. I must have missed the mention of it in the original article.
That point was added with the recent revision, so it was a new addition. If it makes you feel any better, with articles this long and this much revised, I sometimes end up having to search on a particular article to see if I actually included some detail or not!
Powerglide trans was designed by Bill Wallace head of hydromatic Toledo Ohio with a a converter with variable stator vains in the converter where my dad worked with Bill in the 50 to late 80s
Powerglide went through several distinct iterations, but it never had a variable-pitch stator. I suspect you’re thinking of the two-speed automatic introduced on the A-body intermediates for 1964, which is most commonly known as Super Turbine (or ST) 300, Buick’s name for it. The versions used by Buick and Oldsmobile from 1964 to 1967 did have a two-position variable stator. (Pontiac used the ST-300 for the Tempest/Le Mans and Firebird, but without the switch-pitch converter.) The ST-300 was similar to the late Powerglide, but not identical; they were not the same transmission.
Buick DynaFlow’s were probably the first AT to use a variable pitch stator in the torque converter..a few years previous to all the turbo variations of Hydramatics.
Yes, those are discussed in some detail in the Dynaflow article. However, I think the confusion in this case was conflating the two-speed Super Turbine 300 with the admittedly very similar Powerglide.
Buick engineers seem to have been much more taken with the variable-pitch stator concept, probably because they spent so much time trying to create a truly stepless automatic transmission. Something I’ve often wondered about is how that fell chronologically with the use of variable-pitch stators in jet engines (including some jet engines Buick Division manufactured in the fifties), which ended up being quite extensive. I’m still not sure which application came first or if one inspired the other.
question, i got a hydro-matic out of a 80-85 chevrolet delievery van, got it installed and driving off and on for 2 months, it shifts to second around 20 mph, fluid level good , all hoses intact. is that normal? ive drove for 30 mins to hour trip, hasnt overheated. any help would be appreciated. nathan
I’m sorry, this is an automotive history website — I’m not able or qualified to provide repair or troubleshooting advice. Also, this article is about the original Hydra-Matic transmission, which has very little in common with the one in early eighties GM products other than the name!
Congratulations for this great article.
I live in Brazil (where transmissions are mostly manuals). Anyway, here, “hidramatico” is synonymous of automatic transmition, obviously due to the GM hidra-matic!
99% of the people here don’t have a clue that it was a kind of auto box!
Aaron, I enjoyed reading through your comprehensive discussion of the development of the Hydra-Matic. My introduction to these boxes came in 1950, when, as a teenager, I bought a rather clean 1941 Olds 98 with an inoperative H-M for about 1/4 of what it was worth working properly.
I had a 1949 Motor’s manual, which had a step-by-step guide for a 1948 Pontiac to work with. The failure point was quite obvious—the bronze cross-drive gear that drove the front pump was stripped. There were a lot of people who thought this project was doomed to failure, and my success led to a job with Packard in 1954, dealing with the gear-start/twin Ultramatic problems. That’s all a long story, but the important part was that I learned some things about the H-M from Forrest McFarland and a few others, and later was involved in transmission shops where our bread and butter was H-M’s and Dynaflows. I’ll toss out a few comments on the 1937-39 semi-automatic and 1940-56 Hydra-Matics.
On the semi-automatic, “neutral” was in the head-end gearset, along with “forward” and “reverse.” With the car stationary, the single oil pump did not move. Both servos were spring-applied, so the gearbox was in 1st gear, and could not upshift until the car was in motion.
The most comprehensive repair manual for these units I know of was in the 1939 Olds shop manual.
Starting off in Hi range, which gave a shift pattern of 1-3-4, was hard on the box, and at low speed, oil pressure would drop, causing the unit to hunt between 1 and 3. The spec oil was “the same as is used in the engine,” which made shift quality quite sensitive to oil temperature and viscosity. Also, with no torus to cushion shifts, they were pretty abrupt, although with only 3 pistons instead of the 6 used in pre-1944 H-M’s, clutch application was slower and had less clamping force. As to the number produced, a 1942 Motor’s Flat Rate manual, which also lists parts, indicates that the third (and last) pump was introduced in 1938 at s/n 26482.
Going on to the 1940 H-M, one reference I have that you did not list is “1940 Oldsmobile Hydra-Matic Drive Service Instructions,” Oldsmobile Division, General Motors Sales Corporation (no date). Provenance is “Library, General Motors Research Laboratory.” This is not the shop manual, but an introductory 144-page manual that also introduces some other changes between 1939 and 1940 cars. It cites “30,000 (1937-39) automatic transmissions…” A rather salient point in the text is that while the H-M is different from the semiautomatic, it is being introduced as a second generation unit. It is also specific that dealer shops will need to have at least one person prepared to service the units, and presumes that each shop will have someone already familiar with the semiautomatic. Explicitly stated is that Olds will not provide exchange new/rebuilt units, as they had done with the semiautomatic.
Now, to a few points that need some amplification:
The prewar units (both the semiautomatic and the H-M) used alternating bronze and steel clutch plates. The steels were coned, to provide some cushioning on make-up; but, more importantly, to provide release pressure forcing the “sandwich” to open up. A major service issue with this was that the steels would flatten out after a while, causing the clutch to drag when clutch pressure was released. This generally affected the front clutch, so the upshift pattern would be 1-2-4/3-4, which jerked the bands and U-joints and broke parts. Raising the throttle pressure a bit generally gave a better 2-3 release of the front clutch.
Prewar units did not have an annular clutch piston. Instead, there were six round pistons that slid into pockets acting on a cast pressure plate. There was no “check ball” provision for draining clutch actuation oil—it all had to go back through the oil delivery sleeve. Thus the need for a real “kick” to force the plates apart.
I’ll note here that the (Raybestos-developed) lined plates that were introduced (with the annular clutch pistons) in the 1944 revision were bonded to “wavy” spring-steel plates that gave a much more positive and permanent release force, and were specified for service in parts books.
The governor valves were hydraulically balanced against line pressure (fixed 80 psi), rather than spring balanced. Thus, if one of the governor valves stuck, hydraulic pressure would force it to move.
Before 1952, downshift was 3-1, skipping second speed. A forced downshift at 10-15 mph, when the throttle was opened, was a real lurch. Keep in mind that the rear servo had an accumulator with a check valve that delayed application of a released rear band by a couple of seconds, so that a coasting downshift did not generally give a sudden drag. However, this made the 3-1 downshift more of a lurch, and required a pause in a forward range when shifting from neutral to reverse.
Prewar units used a cross-driven front gear pump in the front servo, with limited capacity. That, coupled with a spool valve regulator, both near the oil pan bottom, produced a pronounced whine until around 20 mph, when the rear pump pressure came up. Oldsmobiles were particularly loud. The Cadillac units had a 3-gear set, with an idler on the drive gear, which gave a different lower-pitched whine. With any wear, the transmission would “lock up on reverse coast” when the spring-applied rear servo applied the band, and allow the front band to release at idle as well.
The shims under the reverse pawl bracket are not there for mechanical clearance, but for timing.
To get to reverse from neutral, the rear band had to apply to stop the drum from spinning, but to get the reverse pawl to engage, the band had to start to release just as the pawl was entering the reverse unit teeth.
The loose steel plug at the back of the governor sleeve is there to balance the force of line pressure at the front of the sleeve. It pushed on the side pan to reduce the fore-aft force on the sleeve.
Band release was by application of clutch pressure to the servo. The front servo was “always on” except in neutral, and released by applying clutch pressure to a larger area than on the apply side. This prevented runaway (“flare”) as the front band wore—the band would not release until the clutch was actually engaging. the rear band/clutch operation was similar, except that the band was spring-applied.
I’ve noted a spring 1944 date for introduction of major changes in the H-M. As has been discussed, H-M production continued through WWII for tanks and personnel carriers, and there were plenty of developments that were incorporated into what was a “second generation” H-M. Notable are:
1. Front pump—a much larger crescent pump and simplified pressure regulator now mounted in the front cover, concentric with the shafts.
2. Case—one common case for all H-M’s. Reverse pawl now mounted on a separate bracket that would break free if overstressed, rather than cracking the case.
3. Reverse blocker—incorporation of a piston pressured by the rear servo accumulator that prevented movement of the shifter beyond Lo until the rear band was fully applied.
4. Clutch redesign. Annular clutch pistons with soft lip seals replacing the six-piston/pressure plate setup. Plates were now flat steels and lined spring steel wavy plates.
Going into postwar H-M’s, there were two standard drive trains, “big” and “small” which fit in the same case. The “small” units had three planets in the planetaries; the “big” had four. Also, the “big” units had more clutch plates. “Big” units went into larger-engine cars (Cadillac, Olds V-8, Lincoln, Hudson Hornet). GM published configuration guides outlining these differences.
Indeed, GM published a lot of material targeted to independent shops to encourage them to do automatics for profit—in the 1950’s they were not at all secretive about servicing these units outside the dealer network. Much of the actual “tuning” of the units came in the torus member configurations. 1949 Olds V-8’s had full-vaned torus members, and were notoriously heavy shifters. For 1950, they reduced the torus member vane areas, making for much softer shifting. 1950-51 Olds H-M’s were built with a second gear start in Dr, obtained by changing the springs in the 1-2 shifter valve. That caused a problem because with the specified 375 RPM idle, the torus now spun at 375 RPM instead of 258. They went so far as to offer a service bulletin and kits to return the units to the normal 1st gear start. 1950 also saw the introduction of modulated main line pressure.
In 1951, saw the introduction of the hydraulic cone clutch reverse brake. The old pawl setup was retained, but it now had a windup spring and positive blocking so that the pawl would only engage as a parking lock. 1952 changed the valving—not only providing a 3rd gear “Drive” range, but adding valves to allow a 3-2 downshift.
One other change between prewar and postwar was to move the detent spring-loaded “clicks” from the shifter control head in the car to the valve body in the transmission. That was important if you were going to retrofit a postwar box into a prewar car. In general, that was a fairly straightforward swap particularly with Oldsmobiles.
As to oil to use in an H-M, all of the old boxes that ran on Type A—use Dexron.
As a side comment, when it comes to automatics and efficiency, the Ultramatics (and the Studebaker first type B-W automatics) were not inefficient gas hog slush boxes. Both had converter lock-up clutches and were direct mechanical drive, just like a standard transmission, for most driving.
Hank,
Thanks for the info. Some of this I had found (like the alternating bronze and steel clutch discs) and omitted both in the interests of simplification and in the vain hope of discouraging folks from thinking I know how to fix or rebuild these transmissions, which I definitely don’t! The 3–1 downshift I had not been clear about, since a lot of Hydra-Matic references talk mostly about Dual-Range and don’t address the differences in the earliest units.
Regarding the introductory instruction manual for 1940 units, I would have loved to see that in revising this article! I spent a LOT of time looking for specifically for a 1940 or 1941 manual, without any luck. (The closest I might have found was that the library had a 1944 Cadillac training manual, which I decided not to go through the hassles involved in accessing, since it wasn’t going to answer most of the questions I had about the early Oldsmobile units.)
I’ve done some work on a followup item about split-torque and lockup clutches, referencing the Ultramatic and Borg-Warner DG, although I keep getting sidetracked.
I’m working on two Hydramatics from a 1945 M24 Chaffee tank, which is powered by two flat-head V8 Cadillac engines. The Chaffee was considered a light tank, weighing in at around 20 tons. It was fast and agile for its day.
Doing research on the HM, I came across this excellent information. My contribution is to list the unusual features I have found on the model 256T HM’s that were built specifically for the Chaffee:
– they have no reverse gear. Reverse is in the transfer case that connects the two drivetrains to a common driveshaft. This makes the transmission very simple and very short.
– they don’t use the rather finicky linkage from the carb to the valve body to provide an engine load signal. Instead intake manifold vacuum is supplied to a “throttle vacuum control piston” that moves the throttle valve in the valve body, similar in concept to the vacuum modulator used on much later transmission types. I wonder why this design did not catch on post-war? It seems much simpler.
– the manual control rod from the shifter comes in at the back of the side cover and moves the manual valve linearly, unlike the rotary, detented linkage entering the middle of the side cover. This may have been because the drivetrain is back-to-front; the engine is at the back of the tank, with the HM in front of it.
– they have a massive flat-tubed cooler in a very deep, cast iron pan. The cooler is located in its own sealed-off compartment. Coolant flows over the outside of the cooler and oil flows through the tubes. This requires very reliable sealing between the coolant- and oil-filled compartments in the pan.
– all the internal drive train components are standard parts, such as were fitted in cars of this period, but I did read an article that said GM expended much effort improving these transmissions in subtle ways, for example machining a generous radius on parts susceptible to fatigue, using high quality steels, and hardening areas subject to wear. I assume post-war cars benefited from this.
Malcolm Towrie
Malcolm,
That’s fascinating — I’ve looked at the technical manuals for the M5, but not the M24, so I hadn’t realized it differed so much. On the latter point, that’s probably a safe assumption. Cadillac engineers have said that with production-spec equipment repurposed for military use, they had to make a variety of refinements to satisfy Army inspectors, and the postwar Hydra-Matics were more reliable than the early units.
GM did start taking an interest in vacuum modulation for transmission control after the war, although not on Hydra-Matic. The first 1950–1952 Powerglide transmissions (the dual-impeller variety) used manifold vacuum to adjust line pressure based on load, a feature that gradually spread to other transmissions. I wonder if Detroit Transmission was wary of it for Hydra-Matic for reliability reasons. The mechanical linkage was fussy, but I imagine its failure tolerance was higher (it would work, albeit badly, if out of adjustment, and could be fixed more easily in the field than a vacuum leak).
I’ve recently, through YouTube, become aware of an interesting variation found in some Hydramatic equipped, 1953 Kaisers, but not all of them. Some of them evidently had not only a Dual range Dr position on the shift quadrant, but also a dual range L. This is not just a mistake on my part. The car demonstrator in the YouTube video also mentioned it in his description of the car. This prompted me to verify it visually. I surmise that one of the tick marked L positions allowed a 1-2 shift, while the other started the car off in second gear, as was normal for the low range in Hydramatic equipped cars I’ve driven from ’52 through ’55 model years.
Another interesting anomaly: Some factory fresh 1956 Pontiacs and Olds I drove back in the day, seemed to be equipped with truck Hydramatics. These particular cars did not have the new dual coupling transmission, but had harsh, unpleasant shifting through all the gears, exactly like truck Hydramatics but not typical in cars through 1955.
Interesting — I’ve never seen a dual-low-range Hydra-Matic, although it would be a logical development. You also reminded me that I didn’t mention the second-gear start in Low, which was a new feature of the Dual-Range Hydra-Matic. (I added that point to the text.) With the regular Dual-Range transmissions, full throttle would still give you a first-gear start, which seems like it would occasionally be counterproductive, given that the whole point of the second-gear start was as sort of a poor-man’s traction control. A mechanism to let you decide definitely whether you wanted to start in first or second would seem useful.
Regarding the 1956s, the older single-coupling Hydra-Matic continued to be available in Oldsmobiles and Pontiac into the 1956 model year. I don’t think it was a truck Hydra-Matic, just the existing Dual-Range unit. As for it feeling harsher, I can see several possible explanations. One of the rationales for the dual-coupling transmission was that it had greater torque capacity, so the final single-coupling transmissions may have been firmed up to deal with the more powerful ’56 engines. Another possibility is that the transmission mounts of the ’56 cars were tuned for the smoother new transmission, leaving the older transmission feeling jerkier than ever. (My assumption is that the only reasons for the overlap in single-coupling and dual-coupling transmissions were to clear out stocks of the older unit and perhaps to cover any shortfalls while production of the new transmission ramped up.)
“(My assumption is that the only reasons for the overlap in single-coupling and dual-coupling transmissions were to clear out stocks of the older unit and perhaps to cover any shortfalls while production of the new transmission ramped up.)”
This might be true. Every 1956 Oldsmobile that I have heard about or 1956 Olds HM transmission that I owned, the transmissions were name-plated R55, for the 1955 production year, I never heard or saw one labeled R56, for the 1956 production year. These R55’s were used in just the 1956 Olds 88’s, not the 98’s. But there is a possibly that you could special order the DR HM in the 98’s as they were still available. I personally have never seen a 1956 98 with a DR HM trans.
I agree. The ’56 brochures I’ve seen indicate that the new Jetaway Hydra-Matic was standard on the 98, which would suggest you’d really have to twist someone’s arm to get a 1956 ’98 with the D-R unit. I suppose it’s possible — if I ran into one like that, I wouldn’t necessarily disbelieve it was authentic — but I can’t see any reason why someone would go to that trouble, so it may well be that there just weren’t any 98s built like that.
I have been reading about the original Hydramatics here, and I would like to know how prior to the “dual range” hydramatics introduced in 1952 could a hydramatic car be downshifted from fourth to third gear in order to get engine braking going down a long, steep, hill? My Dad owned a 1954 Buick Special with Dynaflow transmission, which was super smooth and at speeds under 55 mph you could shift the gear lever from Drive to Low to downshift and get engine braking going down a long, steep hill to save the brakes. Didn’t the original Hydramatic cars have a problem not being able to manually downshift into third gear going down a long, steep hill to get engine braking? Low position was only first and second gears and you couldn’t go faster than 40 mph or so. So what if you were going about 55 mph down a hill, you had no way to get engine braking with the original hydramatics?
I don’t think so — you could kick down from fourth to third, but I don’t think there was any way to manually hold third. So, the Dual-Range Hydra-Matic was a great advance in that respect. (I’m somewhat surprised they didn’t do that earlier; the idea of a third-gear hold is something Earl Thompson had originally proposed for the Automatic Safety Transmission back in the thirties!) There was some engine braking in direct drive because of the split torque arrangement, although being able to hold third would obviously have been preferable.
Thanks very much for your reply, Aaron. I always thought the Cadillac engineers made a huge mistake not finding a way to shift into third gear and hold it for going down a steep hill somewhere in the Colorado mountains. I noticed Rolls Royce when they acquired Hydramatic immediately made all four gears available to the driver with the shift lever. Speaking of Dynaflow, that 1954 Buick was in our family for 13 years and we never had any transmission trouble. I think that the torque converter in Dynaflow was a superior design that later ended up in all automatic transmissions by the end of the 1960’s.
The Rolls-Royce version of Hydra-Matic was the 1952 Dual-Range unit. Rolls-Royce manufactured it themselves (except at the very beginning), but apart from generally higher manufacturing tolerances, it was functionally the same as the ones used by GM and GM customers from 1952 on. The Dual-Range Hydra-Matic didn’t QUITE give you complete control over all four forward gears, although it was close enough for most purposes.
Hydra-Matic’s later development was mostly the responsibility of Detroit Transmission Division (taking some concepts from the Engineering Staff transmission group); Cadillac was essentially just one of their customers. My guess is that Detroit Transmission probably started thinking about revising the control layout around the time the Livonia plant was being built or perhaps a little afterward. Although Thompson’s group had proposed a third-gear hold years earlier, the original design team had been under some pressure to get the initial version ready for production, so I imagine a number of “nice to have” features were shelved at that stage for expediency. By 1949–1950, Detroit Transmission had lots of service experience and feedback from various Hydra-Matic users, including GMC, to guide future development.
The torque converters used in the Dynaflow series — there were four major versions, as discussed in the sequel to this article — have very little resemblance to the ones in later automatics like Turbo Hydra-Matic. (When the latter was introduced in 1964, some Buick press and marketing materials implied that it was basically a Hydra-Matic transmission with a Dynaflow/Turbine Drive torque converter, which is really not true at all and falls apart immediately if you actually look at the comparative schematics.) As it turned out, none of the major early GM transmission families ended up representing the direction of automatics of the sixties and later. The prescient models were the Borg-Warner/Fordomatic transmissions (three-speed Ravigneaux gearset with torque converter) and TorqueFlite/Ford Cruise-o-Matic (three-speed Simpson gearset with torque converter), all of which used a straightforward three-element torque converter. (The original Dynaflow had a five-element converter with a two-stage impeller and dual stators; the early Twin Turbine unit in your family’s car had a four-element converter with a single fixed-pitch stator with twin turbines driving an integral planetary gearset; and late units had a five-element version of that with both fixed-pitch and variable-pitch stators.)
By the way, if your comment doesn’t post right away, you don’t usually need to repost. I have comment moderation turned on to avoid besieging you all with the mountains of stupid spam I periodically get, so I have to manually approve comments before they appear.
Thanks again for all your info Aaron, I’ve learned a lot from reading your comments. You are right that Rolls Royce did indeed use the Dual-Range Hydramatic. In a picture of the gear shift on a Silver Cloud steering wheel, I can see “N 4 3 2 and R”. Obviously, 4 gave you all four gears, 3 gave you just third gear for engine braking on a hill, and 2 gave you the first two gears, just like on Cadillacs and Oldsmobiles of 1952. I’m also glad you explained that modern torque converter transmissions are based on the Borg Warner design, which was the basis for the Fordomatic and Mercomatic three speed automatic transmissions of the early fifties. I had always erroneously presumed that Buick’s Dynaflow was the basis for modern Turbo-Hydramatic transmissions. But I still think Buick deserves credit for having one of the first successful torque converter transmissions, with Chevrolet’s Powerglide coming shortly afterwards. Powerglide transmissions, incidentally, are still widely used by drag racers because of their tremendous durability.
The Dual-Range unit actually still gives you all four gears in D3 (or 3, or however it’s labeled on the quadrant) — it just raises the shift point much higher. The way Hydra-Matic (and most pre-electronic automatics) shift gears is with a series of shift valves, which are held closed by spring loading and throttle valve pressure (hydraulic pressure metered proportionately to the position of the throttle linkage) and forced to the open position by pressure from a centrifugal governor on the tail shaft; the wider the throttle is open, the more pressure the governor pressure it takes to execute the shift. In D3, pressure is applied to the “keep closed” side of the 3–4 shift valve equivalent to the throttle being wide open. The transmission will still shift to 4th (keeping the engine from overrevving), but it will do it at a much higher speed (generally in the region of 65 mph, depending on axle ratio). Low or “2,” surprisingly, would actually sometimes let you hold 2nd. If you started in Low, the Dual-Range transmission would start in 2nd gear and stay there, although if you were going slowly enough, you could kick down to 1st.
I don’t mean to imply that all modern transmissions are based on the Borg-Warner model (of which there were a number of significant variations) so much as that and TorqueFlite were much closer to the typical automatic of the seventies or early eighties than either Hydra-Matic or Dynaflow were. Hydra-Matic and Dynaflow are very odd ducks that got a series of belt-and-braces upgrades over the years to make up for their inherent limitations. (You might be surprised to know that the people who developed the concepts behind Dynaflow did not consider the fact that you could and frequently needed to manually select Low either advantageous or desirable! They didn’t want to need any driver control other than forward, reverse, neutral, and park or to have any perceptible shift points at all.)
According to what I’ve read, if the Dynaflow is often started in L, problems will result because the Dynaflow was never designed to be started in L. L was supposed to be used as sort of an emergency gear rather than regularly. The same was true with the early Ultramatics. Exactly what problems would result from frequently starting in L I don’t know.
At one time I owned two antique Packards with Ultramatic: a 1953 Cavalier, and a 1955 Caribbean convertible. The 1955 Ultramatic was much better because it provided the option of automatic low gear starts.
It’s not that starting in Low was intrinsically bad, but that the designers initially assumed that you would normally operate in Drive, with the high clutch always engaged, and stressed the clutch pack and low band accordingly. Transmission line pressure in Drive was also about half that in Low or Reverse, so pressure would drop off as you moved to Drive.
Thanks again, Aaron. You really know your stuff. I’ve learned more from reading your comments about Hydramatic and Dynaflow
than in several years of reading about them from other sources. It’s a subject that I’m very interested in because my Grandfather had a 1949 Pontiac Chieftain Six with a Hydramatic transmission that I actually drove (it’s still the oldest car I’ve ever driven).
I can remember him and my father arguing about what was the better transmission, Hydramatic or Dynaflow. My grandfather felt that since Cadillac had Hydramatic, it must have been superior to Dynaflow. But from driving the two vehicles, I noticed that under full throttle from a complete stop, the Pontiac had a very unsmooth first to second shift, whereas my Dad’s Buick was always very smooth under hard acceleration, plus you could get engine braking by downshifting to Low.
That was the tradeoff: Dynaflow was extremely smooth because in Drive, it was functionally a continuously variable transmission. (In the twin-turbine units, there was a planetary gearset within the converter, but it didn’t create any perceptible shift points.) However, maximum gear multiplication in Drive was, on a 1954 Twin Turbine unit like your father’s, 2.45:1, which was (numerically) less than 2nd in Hydra-Matic. In a 1949 Hydra-Matic, 1st gear is 3.82:1, which gives you much better off-the-line acceleration, but also means a big ratio drop from 1st to 2nd, which emphasizes the mechanical shift shock. (The 1–2 shift on a single-coupling Hydra-Matic engages the front clutch, which locks the intermediate shaft to the front sun gear drum.) Hydra-Matic was probably also a little thumpier because the transmission (or more specifically the front ring gear) was actually bolted to the engine flywheel. In Dynaflow or most other fluid clutch transmissions, the fluid coupling or torque converter isolates the planetary gearsets from the engine, which provides an additional cushioning effect. The Hydra-Matic way is much more mechanically efficient, but you don’t have quite that same isolation.
The trick to getting decent pickup in the 1954 Buick Special was to start out in Low and then upshift to Drive at 55, which I did quite often on my way to Fairfield University. You got about 14 or 15 seconds zero to sixty, which wasn’t too bad for a 3800 lb. car. My grandfather’s 49 Pontiac Chieftain could cruise easily at 60 but top speed was only about 75 with the 93 HP six. But that car was built like a tank and it outlived my grandfather and my uncle kept it until well into the 1970’s with no rust.
If you started in Low, the early Twin Turbine Dynaflow would give you a maximum starting ratio of 4.46:1, which would then fade to 1.82:1 at higher speeds. So, you can see why that would give a lot more snap. Of course, then you had to shift by hand, and Dynaflow’s clutch pack and low band weren’t really designed for routine use.
What was the second-to-third shift like? That was the one that had the reputation of being abrupt.
The full throttle second to third shift in the 49 Hydramatic Pontiac Chieftain wasn’t too bad, it happened at about 30 mph, and then it got up to 50 mph in third gear but wouldn’t go any higher than that. I finally lifted the accelerator and it shifted up to fourth and I got up to sixty with no trouble. What I noticed most about driving the 49 Pontiac compared to my Dad’s 54 Buick was the very high driving position. The steering was very precise in the Pontiac, no play at all.
Concerning comment by Hank Van Cleef re: 1. The shims under the reverse pawl bracket are not there for mechanical clearance, but for timing.
To get to reverse from neutral, the rear band had to apply to stop the drum from spinning, but to get the reverse pawl to engage, the band had to start to release just as the pawl was entering the reverse unit teeth.
I have 1948 Pontiac 8 that has correct Reverse gear backlash but exhibits high gear lever resistance to engage Rev from low. It also is very stiff to release from Rev back to low.
When engine is off and no pressure the lever is smooth both directions. Does anyone know if this was normal in -48 or a fix relating to this complaint?
Love reading your posts, which I found trying to find info on the above query.
Aaron I started by removing 1 of 3 reverse blocker shims to see what would happen. Surprisingly the problem was worse. I then made an additional shim and assembled with 4 total. The result was an improvement from original(3)
This surprised me, but does confirm what Hank states earlier that the shims are for timing. Additional to this the line pressure is 95lbs @ 1000rpm, 80@ 600rpm. I also lowered the idle speed after the changes were made & this cleaned up the problem altogether.
Note: this trans is rebuilt and in excellent internal condition.
On another matter, I rebuilt one in a 1962 Rolls Royce 20 yrs ago (Silver Shadow if memory serves, with mechanical brake servo). It had absolutely no shift feel after rebuild and worried me that it would not last too long.
The original throttle valve spring had been changed to a hair spring which I changed to something that looked like one shown in Motor Manual disassemble view.
The resultant shift was beautiful, just discernible and nothing else.
The owner reported back after he had the car in a concourse show that the judges knocked the car because the shift was not as per RollRoyce “feel” (none)
Thanks again for your reply
regards Bob
Hi Aarron,
Do you know which product is the first commercial automatic transmission that utilizes the combination of one-way clutch and low and reverse clutch? I tried to search for it on the internet but failed. Thank you!
Do you mean like in a Ford C-6, where you have a Simpson gearset with a clutch instead of a band to hold the rear drum in both directions (while the one-way clutch prevents reverse rotation)? I’m not sure. (The first production Simpson gear automatic was TorqueFlite, but TorqueFlite and early Ford Cruise-O-Matic have a low/reverse band, not a clutch.) You might want to look into Philip Gott’s book Changing Gears, published in 1991, which is an exhaustive survey of automotive transmission development to that point.
I drove a ’56 Buick Century for years, and in that time had to have transmission repairs twice. I habitually started off in L, then shifted to D at about 25 mph in normal street driving. The shift thump was not pleasant, but acceleration was much more satisfactory. I learned that the result of that habit was accelerated wear of the Drive piston lip seal. From standstill in Drive with such wear, the front pump couldn’t supply enough pressure to firmly engage the drive clutch, so slippage was terrible. Starting off in low became mandatory. It engaged firmly, and after the car got under way, the rear pump provided enough extra oil flow to overcome the leakage and raise the pressure sufficiently to keep Drive firmly engaged at street speeds.
I also learned in the transmission shop that habitual drag racing (not me) using the Low range in a Dynaflow, often resulted in mechanical breakage. I suppose the mechanic was referring to the planetary gear cage.
A friend once told me of a situation in a ’50s single coupling Hydramatic, where the car couldn’t get moving from a standstill, but with a push from another vehicle, would propel itself strongly after it was moving fast enough that the front planetary would have shifted out. As long as the car didn’t come to a standstill, this worked, the car could run at highway speed, and got the family out of Mexico where the failure occurred. I found in later years that this also worked with a dual coupling Hydramatic that had the same symptom (failed sprag clutch in that case, I suppose).
As a final thought, I think the drivabiity of the single coupling Hydramatic could have been improved a great deal if they had incorporated part-throttle downshift. Regardless of wishful thinking, they’re still my favorite. Can’t help loving their low slippage.
Aside from whatever seal wear you experienced, Dynaflow actually had variable pump pressure depending on manual gear selection. Line pressure was basically doubled in Low and Reverse to provide additional holding power for the bands.
Thanks Aaron for this great work you do. The old Dual Coupling Hydro was my favorite all time transmission. Back in 1962 Pontiac had a performance reputation both in drag racing and stock car racing. Back then Fireball Roberts was still driving Pontiacs so as a teenager I was really enthralled. One guy’s dad had a ’59 Star Chief and two others had 61s, one a Star Chief and one a Catalina and they would all three fly. That summer my dad bought his first air conditioned car. He bought . 1962 Catalina. But when I took it out my buddies all ran off and left me. I ran it through the quarter at a whopping 80 MPH. My buddies were getting more like 90 MPH. What the heck I’m thinking? This thing’s a dog. It would roll top end wise and damn near hide the needle on the Speedometer. But for acceleration? My sister’s Renault Dalphine would give it a run. At least off the line. Through the next couple months I did a little research, stone age style, by reading everything from Hot Rod Magazine to various motor manuals. Then I stumbled on something. I was reading in a Chilton’s motor manual where for 1962 Pontiac’s Roto Hydramatic was in the Catalina and Grand Prix but the Star Chief and Bonneville still had the old 4 speed Hydro.
Then when I took the car out I confirmed it only had three forward gears. I went riding with my buddy Marvin in his dad’s 61 Catlina. It had 4 forward speeds. Also, the Roto Hydro didn’t have a real positive lock up. It wouldn’t turn a tire on ice. Yet Marvin’s 61 would absolutely smoke the tires. When I made all this discovery I told my dad he needed to trade that POS for a Star Chief. But he wasn’t into speed.
Also, I had a friend name Jack whose mother ordered her a new Cadillac Convertible every two years. His dad was the officer in charge of the local Coast Guard Station at that time. His mother had a 1963 Red Convertible with white top. That thing would fly. It would do an amazing 95 MPH in the quarter (Speedometer indication only, but still) and blow the doors on some of the fastest cars in town. It had the Dual Coupling 4 Speed Hydro, of course. Then the next year, which was late ’64, his mom ordered a new 1965 model with Turbo 400 Trans. It was nice and I think it had a few more horse power, but the ’63 would blow its doors off. Those two Cadillacs were the same model, even color, and close in weight and the 65 had more power but the 63 was faster through the quarter. Why? I believe it was the Transmission. Detroit doesn’t always change things for the better. I think the old Dual Coupling Hydramatic was the best transmission GM ever made.
Thanks, Mike
I read your very fine article because I told a friend of mine that I thought Cadillac had Hyromatic drive in 1941. He said no. I thought your article could give me the answer.Your article “sorta” mentioned a ’41 with Hyromatic in passing. So is it definite that ’41 Cadillac had Hydromatic drive?
P.S I have a ’48 Caddy with Hyromatic and have had a ’41 Cadillac with standard shift. But I THINK I remember seeing a ’41 with Hyromatic.
Hydra-Matic was an extra-cost option on 1941 Cadillacs, so some buyers got it and some did not. Roughly one-third of ’41 Cadillacs had Hydra-Matic and the rest had standard shift.
What are the proper front and rear clutch pack clearances? I can’t find a specification anywhere.
I’m afraid I can’t give technical advice except to strongly recommend that you look for a Hydra-Matic shop manual that covers the specific model and year of your car. (I wouldn’t assume that the clearances would be identical from model to model, since the clutches themselves aren’t necessarily the same.)
Speaking of pre-war GM or in this case Chevrolet, does any information exist on an experimental overhead value V8 designed by British born engineer Alex Taub in the early-1930s that was (perhaps too hastily) rejected by Chevrolet?
Well, a patent search finds some interesting work by Taub on related areas, including exhaust valve cooling (US1727621A) and combustion chamber design (US1757399A, which describes a four-cylinder iteration, and US2214941A, which describes an intake-valve cooling scheme). None of them is specifically for a V-8, but I imagine that in these, you can see some of what he may have been thinking in design ethos.
Interestingly, after he left GM, he designed a horizontally opposed engine layout building on some of those concepts (US2506250A). That patent describes an F-head layout that he thought could be cast in aluminum and applied to H-4, H-6, H-8, or H-12 engines!
I know there was a reworked Hydra-Matic, the B&M Hydro Stick, that was popular with drag racers. I understand that it could be held in first gear. The photo of the shift quadrant of the ’52 Pontiac gives the impression that putting the lever in Lo would hold first gear. So I take it the Hydro Stick offered something above and beyond?
While there were variations depending on which version of Hydra-Matic you were looking at, Lo generally did not actually hold first gear, but rather kept the transmission from automatically shifting beyond second; it would still shift automatically between first and second. With the Dual-Range Hydra-Matic, Lo would also give you a second-gear start unless you pushed the throttle all or nearly to the floor (a trick also used by some later Mercedes-Benz automatics). In both respects, the reasoning was that the high numerical ratio of first gear gave first short legs and could easily result in excessive wheel spin on slippery surfaces. This was true, but not exactly a boon to drag racers.
I’m not familiar with the workings of the Hydro Stick controls, but the area where the standard Hydra-Matic control system most needed some help, from a performance standpoint, was actually in delaying the shift from second to third. With standard gear ratios, you wouldn’t necessarily want or need to hold first longer than the stock Hydra-Matic normally would, although a drag racer would probably NOT want to start in second gear, and in some non-drag-racing competition (e.g., rallying), being able to positively hold first gear might sometimes be helpful.
The post-war Hydra-Matic would shift from 1-2 at a maximum of about 12-15 mph with the throttle held to the floor, or the transmission selector engaged in the Lo position.
While adequate for road use, high power, high revving racing engines would not reach peak performance in 1st before the transmission shifted.
B&M’s modification added a valve between the governor and the selector valve block, engaged by the selector cam. An additional detent had to be filed in the cam between Lo and R.
The valve prevented the speed-related oil pressure from the governor engaging the 1-2 piston. Shifting to Lo would cause the transmission to shift when the driver instructed.
Lo and Dr continued to operate as per the regular design (increased hold point 1-2 in Lo, and regular 1-2-3-4 in Dr).
Thanks for the additional information. It may be worth mentioning that the low 1–2 shift point was probably a reasonable choice for street use because the four-speed Hydra-Matic had a quite low (high numerical) first gear — typically 3.82:1 on Dual-Range Hydra-Matic units. Since most American street engines were not what we’d consider high-revving, pushing them to high RPM was generally neither necessary nor an especially good idea. O course, race-modified engines were a different matter.
How come there is not a single mention to José Braz Araripe and Fernando Lehly Lemos?
They invented the first functional hidraulic transmission and sold it to GM.
The development of Hydra-Matic within GM is relatively well-documented and is evidenced by a whole series of contemporary patent disclosures. There is no mention of Araripe or Lemos in connection with the Automatic Safety Transmission or Hydra-Matic in any contemporary sources I’ve been able to find. I found one snippet of a 1943 article describing José Braz de Araripe as a hydraulic engineer and a 1957 corporate tax record identifying Fernando Lehly de Lemos as a director of what I gather was a commercial publisher in Rio de Janeiro (Companhia Editora Comercial F. Lemos), but I see no patent records, in the U.S. or otherwise, and no SAE papers under either name. Indeed, the only references I see identifying them as inventors of a hydraulic transmission are relatively recent online posts that cite no sources.
The field of transmission development, and of automatic transmission development, is broad, and if Araripe was a hydraulic engineer, it’s plausible that he worked on developing some kind of hydraulic transmission controls in the thirties. It’s not even out of the question that he might have sold some invention to GM; Earl Thompson obviously did with what became Synchro-Mesh, and GM has a Brazilian subsidiary, General Motors do Brasil, which was established in 1925. However, even if we assume that was the case — and I would need to see some more credible evidence of that than unsourced blog posts — there is a substantive difference between “invented a related piece of technology” (which is plausible) and “independently developed Hydra-Matic and sold it to GM” (which is not, and would contradict a variety of well-substantiated evidence).
If you have some kind of sources for that assertion, I will certainly investigate them, but I’m not seeing any.
Actually it was never necessary to stop before shifting into low gear, even before low gear was synchronized.
I learned to drive on my mother’s 1950 Chevrolet. I also learned how to double clutch to shift into low gear without stopping.
Well, yes, and if you were sufficiently adept, you could also shift without using the clutch. However, these were not skills that were widely taught, and the grinding that results from doing it wrong tend to discourage experimentation.
Aaron, I love your articles about transmissions! One little question, though, about the Olds Automatic Safety Transmission. You mention several times that the governor was driven off the OUTPUT shaft. However, according to Philip G. Gott’s “Changing Gears: The Development of the Automotive Transmission”, the AST governor is driven off the ENGINE. This is also shown in his cross section (Figure 6.3) and his sketch of the control system (Figure 6.4). Also, the governor being on the engine seems to make sense to me, as the the shift from Lo to High Range (rear planetary) was done by the shift lever (so no need for a governor), while the front planetary was automatically shifted so it needed a governor. What I’m missing?
This is what we in the trade call a mistake! It took me more than an hour of digging through the various patents and such to figure out by whom; the Gott book, while very useful, does contain a number of significant errors, many of which are probably attributable to having been written in an era when it was much harder to just pull up copies of patents and the like. In this case, the error was likely mine, so I’ve revised the text.
This was a surprisingly difficult point because the trajectory of the various patent filings suggest that there was a lively debate among Thompson and his team about whether the governor should be driven by the engine or the output shaft. Some of the early patents (e.g., the 2,195,605 patent) proposed the latter, some the former, and of course the eventual Hydra-Matic transmission finally opted for the latter. I unfortunately don’t have anything resembling an actual service manual for the Automatic Safety Transmission, nor do I know for sure if such a thing was ever published or not (Oldsmobile was not keen to have dealer technicians disassembling the thing), so it was a matter of trying to weigh Gott’s account (which I value but don’t unconditionally trust), a small diagram with labels and no key, and at least five different patents that were all close to but not necessarily exactly the same as the production transmission. Also, I recently discovered that Gott had coauthored a 1979 report on the history of automobile transmissions for the Department of Transportation (which I didn’t have in 2016, but discovered by accident in 2021), which also asserted that the Automatic Safety Transmission shifted based on road speed! So, you can perhaps understand the degree of confusion involved.
On the plus side, reviewing the latter report finally enabled me to locate a period article containing a detailed technical analysis of the actual production Automatic Safety Transmission (as opposed to the patent disclosures, which as mentioned are not necessarily fully reflective of the production transmission), which confirms that the centrifugal governor ran off the same shaft as the oil pump. That source also has a three-dimensional illustration of the transmission’s inner workings, including the pump/governor drive, which was significantly easier for me to get my head around than the patent illustrations. I also found and reviewed an additional related Thompson patent I was aware of but hadn’t previously looked at (2,193,524, divided from the 2,362,418 patent originally filed in 1937)
In any case, I think I’ve now rectified that error in the text and made a number of related corrections and clarifications.
(If you’re curious, the 1979 report is entitled Study and Test to Confirm Automobile Drivetrain Components to Improve Fuel Economy, Volume 1: History of the Automobile Transmission in the United States, prepared for the U.S. Department of Transportation by Donald A. Hurter, Philip G. Gott, and Carl A. Gottesman of Arthur D. Little, Inc. and released to the public by the National Technical Information Service. Since it was prepared for a federal agency, it’s freely available online; you can find it at, inter alia, the National Transportation Library Repository and Open Science Access Portal: https://rosap.ntl.bts.gov/view/dot/10723.)
Thank you very much for taking the time to investigate the Automatic Safety Transmission’s governor being driven off the engine (not the output shaft). I fully appreciate that good info can be hard to find! And I’ll admit that I find patents to be notoriously difficult to figure out, which is likely done on purpose! So I also congratulate your perseverence.
Also, as you say, Gott’s book is very good but not infallible. For example, his schematic for the original Hydra-matic 180 (Figure A.2) shows a cone clutch on the Reverse ring gear. This appeared to be reinforced by the Weber State U videos showing the disassembly of what was purportedly a 1949 Hydramatic. And Professor Kelly, indeed, showed a cone clutch, saying that Reverse was done sort of like a synchronizer, whereby the cone clutch gets the ring gear stopped and then the Reverse Anchor latches into the ring gear like a park pawl. However, he later explained that his disassembled unit was actually a 1953 Hydra-matic. I believe that, in fact, the cone clutch only came out in MY 1951, just as you say in your Hydramatic article. And even the Figure 6.5 in Gott’s book shows an original Hydra-matic 180 cross section without a cone clutch on the Reverse ring gear. Further, a Hudson Service Manual refers to the 1951 Hydra-matic as having a “friction” Reverse, as opposed to the Reverse Anchor latch. So even knowledgable and well-intentioned experts can get these things confused. But in my humble opinion, I think you have correctly sussed out the AST governor drive.
Thanks again for your help. I share your love for this early automotive technology, including its impact upon society in general. And your articles are absolutely excellent!
I think I may further rewrite the AST section to better explain how the mechanism worked and possibly also take a shot at making a color diagram of the Automatic Safety Transmission, which I didn’t previously feel confident in doing for the reasons I mentioned. (I’m actually really inordinately pleased that I was able to locate the contemporary references I eventually found — I was specifically LOOKING for something like this in 2016, but never managed to find it because it was like looking for a needle in a haystack.)
Thank you! Your proposed further explanations of the AST sound worthwhile and interesting.
I have also done some writing about early car transmissions, so I can fully relate to your satisfaction at finding those “needle in a haystack” references!
Interestingly, from 1941-1953 Chrysler offered 4-speed semi-automatic transmissions. Pre-war they were vacuum controlled (Vacamatic is one name) and post-war they were hydraulically controlled (Prestomatic is one name). They initially used a fluid coupling to ease the launch, later offering a 4-element Torque Converter. Behind that was a conventional pedal-operated clutch and then a 4-speed countershaft transmission with synchronizers and a one-way clutch. Like the AST and then Hydramatic, the Chrysler used a 2×2 configuration for the forward speeds, whereby each gear-set had both reduction & direct drive, alternating to get a total of 4-speeds. And, just like the GM units, the Chryslers had a “double-transition” shift between 2nd and 3rd.
The rear Chrysler gear-set was shifted by a conventional manual shift lever, but the front gear-set shifted sort of automatically. And the governor for auto-shifting the front gear-set was essentially driven off the ENGINE. (Well, actually, the countershaft headset.) So there are similarities to the AST. Which is one reason I suspected the AST also used an engine governor to shift the front gear-set.
Also, just like the AST and pre-war Hydramatics, the Chrysler gearshift positions were N Hi Lo R, reflecting the 2×2 configuration of all of them. The order was different, as the Chrysler used a modified “3-on-the-tree” column linkage, and the AST was different from the Hydramatic. (After the war, Chrysler used Dr instead of Hi, just as Hydramatic did.) Whatever, I find the similarities interesting. (When the amazing O.K. Kelley came along with his Dynaflow in 1948, it was a totally different animal!)
Anyway, good luck with your AST update. I’m looking forward to it!
Aaron, I was struggling to get the Gott (et al) paper and got an Error message.
But I was just now able to get a pdf file.
Thank you! I will read it.
Okay, that’s good. I’ve found a lot of federal databases can be weirdly buggy (especially on the weekend, when there’s no one around to troubleshoot), often work only with one particular set of web browsers, and lag behind in areas like human-readable URLs. In this case, it appears they’re having problems with the site’s Drupal content management system.
You’ll find that the first two-thirds of the report reads like a very early draft of Gott’s later book, albeit with a narrower scope (the report focuses specifically on transmissions used in the U.S.) and quite a few errors of various magnitudes. In that respect, it’s kind of an historical curiosity more than a useful resource, although the fact that it actually credits where the Automotive Industries illustrations came from was enormously valuable. (Automotive Industries was a weekly trade publication, and even if you’re reasonably sure it covered a particular topic, actually finding the applicable article(s) without knowing the specific issue date is really down to luck.)
The Chrysler semiautomatic transmissions (and Fluid Drive, with which they’re often — incorrectly — conflated) are an intimidating subject because they went through so many iterations that the “version control” issues you noted yesterday become a serious problem: A statement that may be correct about one version isn’t necessarily correct about all versions, and you may have to look very carefully to know which one you’re looking at.
Thanks, Aaron. You’re right! Chrysler Corp’s offerings of “Fluid Drive” semi-automatics from MY 1939-1954 were a real dog’s breakfast of confusing technologies and terminologies. And reliable technical data is hard to find.
However, my uncle’s new 1951 Chrysler New Yorker with Presto-Matic TC + 4-speed made a big impression upon my nerdy little self, especially when contrasted with the Ford-O-Matic in my father’s new ’51 Ford. So I later investigated it as a teenager, mostly using Chilton’s and Motor’s Manuals from the library. I also drove a number of cars with these trannies. And lately I’ve gotten copies of some Chrysler service reference literature. So I think I have a pretty decent idea of how these particular transmissions work. But I was only really offering that because, again, I find the similarities with the Automatic Safety Transmission and Hydramatic to be interesting.
Speaking of the AST, I have been reading that 1979 DOT tech paper by Gott and others. So far, I think you’re also spot on with “the report reads like a very early draft of Gott’s later book.” Yes; a very early and very rough draft. The report asserts that the AST shifted automatically from 1st thru 4th, using nearly the same hydraulic controls as the later Hydramatic, including a hydraulic throttle valve and output-speed-driven hydraulic governor. I don’t think any of this is true. (I also found point about other transmissions that were conflicting with other sections of the same report.)
But the silver lining is that it appears you discovered in the DOT report some very useful references which explain the AST. So I’m very much looking forward to your AST update!
No, but in this case, I think I can make a pretty good surmise where Gott (and his coauthors; I have no idea of the division of labor and don’t necessarily want to blame him for a collaborator’s mistake) went awry, which is also illustrative of the problem I was describing.
I believe that the authors conflated the production Automatic Safety Transmission with Thompson’s April 1938 patent (2,204,872), which mechanically is very much like the AST, but proposed a mostly new hydraulic control system, an early iteration of the ideas underlying the Hydra-Matic controls (albeit still with at least one big conceptual difference, also described in O.K. Kelley’s 2,211,233 patent). That transmission DID operate more or less as the text of the 1979 paper described, although so far as I’m aware, it did not enter production in that form. (I would not be surprised if there were working prototypes of it, but that isn’t the same thing as being offered to the public through Buick and Oldsmobile showrooms.)
Here you see the problem: The ‘872 patent was filed not that long after the Automatic Safety Transmission was launched, and because the mechanical layout it describes is very similar, anyone familiar with diagrams of the AST would probably recognize it right away. The dilemma is that that ‘872 patent was one of at least five related patents filed by the same person over about six years as part of the same project, reflecting the ongoing evolution of a work in progress. With the production AST, I think Oldsmobile said basically, “Look, we need to build SOMETHING — freeze the design at what you think is a workable point and let’s go forward with that while you guys continue tinkering,” and the point at which that happened was actually an earlier iteration of the design than the subsequent ‘872 patent reflects. That’s an understandable mistake, and not an easy one to sort out 40 (or 80!) years after the fact.
Thanks, Aaron! Very interesting!
I guess I should cut Gott a lot more slack.
I have to be sympathetic to the difficulty of managing what by modern standards are minor factual points in the pre-computer era, where people were working with handwritten notes, probably on notepads or note cards, with little ability to just quickly double-check online how something was spelled or the date something was filed. If it’s a survey covering a lot of ground, like that report, I imagine it was even more challenging, and errors large and small were probably inevitable.
Still, there are statements in that document that don’t really make any sense even in their context. For example, their description of the Packard Ultramatic (which they repeatedly misspell) includes a critical comment about the low efficiency of the torque converter, which is silly on its face: Ultramatic had a mechanical lockup clutch, so the converter never reached, or needed to reach, a high-efficiency coupling stage because whenever it approached that point, the clutch would engage and the converter would be locked out completely. (The only time the torque converter’s coupling efficiency might have become an issue would have been in extended cruising at speeds too low for the governor to engage the clutch, like crawling along in a parking lot at less than 15 mph.) So, the statement isn’t an error in the sense of some of the other actual mistakes, but it’s a remark that makes clear that an important point had been missed. Stuff like that ends up limiting the value of the document as a reference.
Thanks, Aaron. I have finished reading the 1979 DOT paper, and I also have some detail issues. However, my bigger concern is that, UNLIKE YOU, the DOT paper seriously underestimates some key aspects of transmission development:
1. The difficulty of shifting early unsynchronized transmissions.
The key problem of rpm matching was complicated by early carburetors, with their hand throttle and lack of accelerator pump. (Tough to “blip” the throttle for a downshift, for one.)
My grandparents’ generation maintained their long-held reluctant to shift a manual transmission, even after synchronizers became common.
2. The importance of the Model T Ford to the creation of the entire car industry.
3. The importance of the Model T’s planetary transmission to its market success.
It was not ideal, but it didn’t demand the difficult rpm-matching.
Yet the DOT paper says the big advantage of a planetary back then was quieter operation.
Not even a cross section of a Model T transmission, even though 15 million were sold.
4. The rapid acceptance of Thompson-type synchronizers and their importance to car sales.
Rather, the DOT paper talked up the development difficulties, hinting slow acceptance.
Most of the above shortcomings are still evident in Gott’s 1991 book, “Changing Gears…”
He does admit, however, that Thompson-type synchros gained market acceptance pretty quickly.
The Model T now seems to represent a different sort of problem for historians: when contemporary sources don’t bother detailing things that are considered ubiquitous until the details start to be forgotten. When that report was issued, the Model T had been out of production for more than 50 years, but that was no further back than a 1970 Chevrolet Impala is now, so it was well within living memory, and there were still many of them around. Of course, one could still drive an Impala in modern traffic with no great problem other than maybe gas mileage, where a Model T was not up to even late ’70s freeway driving and would have required patience around town, but the Model T was so common and so comparatively crude that it was a pretty easy hobby car. Now, the T is old enough and odd enough to seem truly antique and exotic to modern eyes, and not discussing it feels like a bigger omission than it might have in the seventies.
Thanks, Aaron. I’m grateful that your Hydramatic section fully acknowledges the difficulty of shifting those early manual transmissions. As you wrote, even Henry Ford — who clearly had mechanical aptitude — struggled with them. And, although he was ruthless in seeking simplification in order to lower costs, Ford chose the more complex and expensive planetary gearbox for his game-changing Model T because it didn’t require rpm-matching. So it certainly belonged in the DOT survey of significant transmissions in automotive history.
You also gave appropriate credit to Earl Thompson and his synchronizer. Again, I’m grateful to you.
BTW: There was a generally excellent British TV detective series called “Foyle’s War” that was set in WW2 England. Amazingly, one episode revolved around the transmission synchronizer. However, the TV show “explained” that a British guy actually invented it, but a ruthless American businessman stole the idea and made a fortune with it, cutting out the Brit totally. Then, to avoid the Brit broadcasting the true story, this greedy and odious American came over to England and killed the Brit inventor! I very much object to this completely bogus history and its tacit “diss” of the great Earl Thompson.
That reminds me of the 1952 movie Breaking the Sound Barrier, directed by David Lean, which gives the strong impression that the titular event was a purely British exercise. There is no place in the plot, narratively or thematically, for someone like Chuck Yeager, a hillbilly from West Virginia who definitely didn’t have an old school tie. That kind of chauvinism is very common when it comes to automotive subjects, and it drives me up the wall. It’s often applied to the Japanese auto industry, either insisting that everything of merit from Japanese automakers was an inferior copy of some British idea or that an entire national industry was created for the sake of the American market, neither of which is true.
Thanks, Aaron, for the interesting and very apt analogy of the “…Sound Barrier” movie to the “Foyle’s War” episode about the invention of synchronizer. And the flying movie was just as bogus. Chuck Yeager was amazed that, years afterwards, even US Air Force generals would ask him if he had really used counterintuitive control movements to survive getting through the sound barrier!
But film is a very powerful medium. And if it happened on the screen, then it happened…period. So, as an example to cut some slack for the Brits, recent research indicates that Captain Bligh was actually a humane ship’s master (at least by the harsh standards of the day) and very competent. Set adrift by the mutineers to die in a small open boat with 17 loyal crew members, he piloted his way over 4,000 miles and rescued all hands. It is still regarded as the greatest feat of seamanship in the history of the British Navy, and Bligh later reached the rank of Vice Admiral. But Hollywood and Charles Laughton’s superb “bad guy” acting trashed that reputation in just 132 minutes!
Nice work on enhancing your material about the Automatic Safety Transmission’s governor!
I have actually spent quite a bit of time in the past couple of days rewriting that section (which is not yet reflected in the version that’s online) to make a variety of minor corrections and clarifications. Hopefully I can post the updated version in the next few days, although that may depend on whether I can do a diagram. (I’m still mulling how to do a diagram comparable to the others that reflects the structure of the forward-reverse unit and its sliding gears.)
I’m aware that people get frustrated when I spend time tinkering with old articles rather than adding new content. However, I’m also (painfully) aware that there’s very little readily available information on these transmissions, and I’d rather not promulgate incorrect information about them!
I posted the updated text, although I have not yet made a diagram, which will take some time.
Thanks, Aaron. I do NOT get frustrated with your “tinkering.” As you say, this material is difficult to research and find reliable information about. So I look forward to your additional enhancements, even though there will likely be no end to them. Your articles don’t lend themselves to a, “There; that’s done and dusted” approach. And a good diagram of the AST would be helpful to anyone who reads your article. So please keep up the good work!
One of the most enjoyable SAE events I ever attended was on Sept 20, 1989; a panel discussion on the occasion of the 50th anniversary of the automatic transmission. The panelists were 9 retired engineers from all the US companies who had developed automatics in the 30s, 40s. & 50s. And for well over 2 hours they just talked about how things happened. Fascinating. Because this section is about the Hydramatic, I’ll just give a couple of tidbits from Maurice (Rosey) Rosenberger:
1.) When Alfred Sloan took his first ride in a Hydramatic prototype in 1938, at 60 mph the transmission made an unscheduled downshift from 4th to 1st gear (due to a stripped gear on the governor/oil pump drive). The engine over-revved so badly it put the engine fan through the hood and the car was dead in the water! Sloan was ready to cancel the whole program, but Charles Kettering came to the rescue, telling Sloan it was just one of those things that give engineers sleepless nights, yet they get sorted out in the end. Kettering, who also reminded Sloan how difficult driving a manual transmission could be, convinced him to continue with the Hydramatic. (A few weeks later, I called Rosey about some elements of this. He turned out to be a very nice person who willingly talked to me.)
2.) Rosey explained how the team resolved the problem of getting hysteresis in shift points (to avoid shift hunting/cycling) via a back-of-the-envelope sketch of a dual-diameter shift valve. (Refer to SAE 470247 1947-01-01 Automatic Transmission Control System by Kelley & Rosenberger.)
Whatever, every subsequent automatic transmission at any OEM with hydraulic shift control logic has used this solution.
As Hegel famously wrote, “Nothing great was every accomplished in this world without passion.” And all the panelists were clearly passionate. But this doesn’t necessarily bring logic & order. There were big differences of opinion as to the “right” solution, often within the same company. Buickians never forgave Earl Thompson for his bad review of their beloved Roller CVT, and they never put a “Hydra-Jerk” into any Buick. (At least not until later, when the THM 400 came out.)
And the panelists were highly creative people. But creativity is usually a messy process.
Sorry for the ramble, but I’m just trying to support your point about how difficult it is to get reliable information. The SAE panel discussion was wonderful, and I still have my original notes and internal company report. But it was “Oral Only.” And the only press report I could find was an Oct 1989 Ward’s, which included a separate interview with Frank Winchell of GM, who quipped, “I have a hell of a time remembering what happened yesterday, let alone 50 years ago.” So, even though the panelists were giving “eyewitness testimony,” it was recollection of events which happened up to 60 years before. As Mark Twain wrote, “When I was younger, I could remember anything, whether it had happened or not.” Which means the panel discussion is a little shaky as source of hard information today. But it sure was fun!
Thanks for these recollections! I actually have a copy of that 1947 SAE paper (here in front of me, in fact), but I hadn’t thought to note the point about the snap-over action to discourage “hunting.” I’ll add that to the text. I strongly suspect that this was also a major reason why they finally moved the centrifugal governor from the input shaft to the output shaft; thinking about it, it occurred to me that having the governor be engine-driven meant that governor pressures would drop off immediately after each upshift, which meant a fair likelihood that the transmission would immediately downshift after completing an upshift, especially if the driver backed off the throttle.
I wrestled a bit with how much detail to try to add about the workings of the hydraulic control system, which is really quite complex and became more so over the years. My assumption is that most readers will be struggling enough to grasp the basic principle of opposing pressures, and that some of the other nuances risk making it incomprehensible (possibly also to me!). However, I think that part is worth mentioning because it also probably addresses why they changed how the governor was driven, apparently at nearly the last minute, production-wise.
I have been toying for the past two and a half years with doing an article about the principles of hydraulic shift control, but it’s kept stalling in part because I know it would demand illustrations that I don’t feel confident I could create and for which I don’t have a good alternative source. (I’m not a illustrator and especially not a technical illustrator!) The dilemma, as I tried to express in the text of this article, is that even a comparatively simple hydraulic control layout is like a computer circuit board, and indeed functions as an analog computer system.
Most of the sources I’ve seen also indicate that Buick engineers were resentful of the Automatic Safety Transmission and Hydra-Matic projects, which I can well understand. The Roller project was developed back in the era when the divisions were mostly responsible for their own engineering R&D (the Engineering Staff had only recently been established at that point), so when it was canceled, it meant writing off a pretty substantial investment of time and money and labor with nothing to show for it, which was bad for the bottom line as well as a sting to divisional pride. There is a certain irony, however, in the fact that some of the same engineers who developed the Automatic Safety Transmission and Hydra-Matic (in particularly O.K. Kelley) were then the driving force behind the many evolutions of the Buick torque converter automatics.
Aaron, I fully understand your dilemma of trying to explain the complexity of hydraulic controls, including the need for good, yet simple, diagrams of a hydraulics circuit.
(BTW: Although you claim not to be an expert at technical illustration, you certainly do a nice job with your gearbox schematics!)
If you are interested, I wrote a paper during COVID lockdown about the transition from hydraulic to electronic transmission controls. It includes hydraulic and electronic circuit diagrams, but in a very simplified format. It also includes simplified explanation of hydraulic sensors like governors, rpm sensors, load and throttle position sensors, as well as their electronic equivalents. You are very welcome to this information, as it might give you some ideas. But I do not want to “publish” it by posting it on your website. So if there is some way I can send you a PDF file on the side, let me know. If it turns out to be of interest, I could dig up and send you the original art for you to use/modify. Just let me know if you want anything.
I will separately address the other very interesting topics you touched upon in your recent post.
Hi, Aaron. Now for your topic of engine- vs. ouput-rpm governor…
Yes, Rosenberger felt the “snap-over” action of the shift valves was crucial to getting good automatic shifts. However, I’m not so sure this key to Hydramatic going to an output-rpm governor.
Rather, the Hydramatic used a double-governor (2 governors in one body), and the SAE paper gives a pretty good explanation of why. However, the AST did not need a “high-mph” governor for the rear planetary because it was shifted manually by driver (using his own “built-in” governor).
So the AST needed a governor only for the 1-2 and 3-4 shifts of the front planetary.
But the AST didn’t really “care” if it was the 1-2 or 3-4 shift. It was the same front planetary — with the same ratio step — that was shifting in either case. So engine rpm could be the basis of the shift, and modulation for driver demand/load via throttle position could be used.
This permitted a simpler 1-stage engine-driven governor, which was relatively simple because of the AST’s countershaft input section. I have seen this on other transmissions:
— Chrysler Prestomatic. As previously explained, it had the same 2×2 transmission layout, and the rear gearset was shifted manually. So it was only the front gearset which needed a governor, and Chrysler used an engine-rpm governor.
— Allison lock-up shifts: Some older 4- and 6-speed commercial-duty automatics shifted from TC to LU in each range. These LU shifts were controlled via a pitot tube in the input section. (I think it measured turbine rpm, not engine rpm, but I’m old.) Whatever, it wasn’t necessary to worry about road speed, as the LU shift didn’t “care” what range the transmission was in. And its “step” was the ratio of engine-rpm/turbine-rpm.
to be continued
Continuation of engine- vs. output-speed governor…
(I had previously gotten an error message, maybe because of length?)
On the other hand, there were early automatics with Lock-Up which used road-speed governors:
— Packard Ultramatic: No need for a governor for basic functions as it was pretty much like a Buick Dynaflow. But its LU clutch needed one. So, perhaps because a turbine speed governor would have been difficult with planetary gears, and in “D” the gearbox had a 1:1 ratio, anyway. So Packard used an ouput-shaft-rpm governor. (Which was useful years later, when they automated the 1-2 shift.) But I don’t know how they handled the LU shift when the driver selected Low range. In any case, the Ultramatic was known to have a rough LU shift.
— Studebaker/Borg-Warner automatic: It had automatic gear shifting, so it had a road-speed governor. But because of conflicting descriptions I’ve found, I’m not sure if LU was engaged independently of gear range or if LU was engaged only in 3rd gear. Or maybe even together with 3rd gear? In any case, the LU shift could be pretty rough.
Whatever, I don’t think LU in passenger car automatics really got acceptable LU shift quality until electronics came along at the end of the 1970s. That permitted fairly simple measurement of engine and turbine rpm, which enabled better control of the LU shift. But again, I’m not sure of that history.
Bottom Line: I don’t think the snap-over shift valves influenced the switch to a road-speed governor on the Hydramatic. On the other hand, you have docs showing that Thompson was considering a road-speed governor for the AST. So perhaps I’m missing something?
I’m not sure about the error message, but it may have been triggered by length. (This can be difficult to judge visually because the limit is based on total characters, including spaces; last time I checked, it was 4,000 characters, although I don’t know if subsequent WordPress updates have changed that.) Sorry for the inconvenience, at any rate.
At least one of the early Automatic Safety Transmissions patents (2,195,605, filed in 1934) arranges the governor so it’s driven off the output shaft. This is distinct from the production Automatic Safety Transmission, although the way the governor works is similar. Unlike Hydra-Matic, AST governor rotation doesn’t control metering valves, it actually moves a series of rods and lever arms. The automatic gearset is actuated hydraulically, but the control valve operation is purely mechanical: The governor levers move the linkage one way, the throttle levers try to crank it back the other way, which increases the distance the governor linkage has to move.
Arranged like that, which end the governor was driven off of probably didn’t make a great difference operationally, and it’s possible running it off an extension of the pump shaft was for packaging reasons. (The pump drive on the AST was a worm gear kind of tucked in the back of the forward-reverse unit; the pump and governor assemblies added width, but not length.)
The dual governor weight strategy was introduced in Thompson’s 2,204,872 patent, filed in 1938, which describes the first iteration of the fully automatic shift schedule, albeit with the same mechanical layout as the production AST. The rationale for the dual weights was the same as in Hydra-Matic; as the 1947 SAE paper explains, combining the G1 and G2 pressures flattened the net pressure curve. However, it may be worthwhile to note that the combined pressure curve still was not actually linear, just MORE linear; Figure 12 of the ‘872 patent is a graph showing how the two curves added up. So, even with the dual weights, a relatively small decrease in speed resulted in a proportionally greater decrease in governor pressure. The ‘872 patent doesn’t suggest that as a problem in terms of shift hunting, but it’s possible that subsequent experimental work indicated that it was more of an issue than initially expected.
I freely admit that I’m speculating, but within 18 months, they had moved the governor from the input-driven pump shaft to the output shaft, combined with the (new) rear pump, and I have to assume there was a reason for that. Rival automatic and semiautomatic transmissions did not necessarily use the dual governor strategy — perhaps because of its patent encumbrance — and thus had to wrestle with the square law issue the dual governors were intended to mitigate. In that regard, it might be significant that many of them put the governor on the tail shaft. On the other hand, I may be reading too much into it.
I don’t remember off the top of my head if the Ultramatic lockup clutch would engage in Low (I’d have to look it up to be sure), but I think it was probably set up not to operate in Low or Reverse, since those ranges needed the extra multiplication of the converter and weren’t intended for cruising. (In automobile applications, even in the eighties, lockup converters generally didn’t engage in first gear, with the exception of the Ford centrifugal lockup clutch, which worked on all forward gears because its engagement was entirely centrifugal, not hydraulic.) I’d need to look at the patents and maybe a service manual to see how Packard handled the control layout in Low, although it would have been straightforward to arrange the hydraulic controls to cut off the line to the lockup valve with the selector in Low or Reverse.
The Borg-Warner DG lockup clutch was an integral part of how the DG transmission obtained top gear, which was a purely mechanical direct drive. Consequently, the high clutch could not operate in any other range, and you couldn’t have direct drive gearing without the locked-up converter. (I have diagrams of it in the split torque article, although the diagrams are purely of the mechanical layout, not the controls.) It was loosely comparable to the much later Ford AOD, where the fourth-OD range was also purely mechanical. This was efficient, but meant you couldn’t have the benefits of converter multiplication without a kickdown, which in pre-CAFE days was considered a pretty substantial drawback.
Thanks, Aaron. But my face is now VERY red. I now believe that you were were right all along. That is, even the production-version AST had the governor driven off the OUTPUT shaft rpm!
I came to this conclusion because yesterday I finally got a copy of Earl’s 2,195,605 patent , and I started comparing it with the production AST as explained in the Automotive Industries May 29, 1937 article. And AI’s Fig 4 shows that the 2-stage governor (and smaller oil pump unit) are actually driven off shaft W, which rotates with the first intermediate shaft, hence whenever the rear wheels are turning!
I had been deceived by the Fig 1 cross-section where the governor appears to be driven by shaft F of the input countershaft section. But, as Fig 4 shows, the drive is more complicated, with inner and outer shafts. And the inner shaft is driven by shaft W, while the outer one for the larger oil pump is driven by the engine (via a countershaft gear).
I was also deceived by a controls actuation sketch (Figure 6.4) in Gott’s book, which shows the governor being geared directly off the engine via a countershaft gear.
I’m still wondering where the main oil pump is located with Earl’s 2,195,605 patent, but these things are tough to muddle through, and it will take me more time. But I wanted to point out the governor drive situation right away.
I’m very sorry for this confusion.
(I will address your other points separately.)
In the 2,195,605 patent, the transmission oil pump and governor drive are roughly in the middle of the transmission. If I may be presumptuous, I suspect the difficulty you’re having in identifying it in the patent diagrams (which I had as well) is that the pump and governor assembly is mostly in a different plane than the other components shown in longitudinal section. Take a look at the side view cross-section in Fig. 20, and look for a star-shaped gear labeled “204.” That’s the pump drive gear. If you then refer to Fig. 8, which shows a cross-section of the pump assembly from the rear, you’ll see the same gear 204 and how it relates to the pump and governor drive. Note that “32a” in that figure is a section of the output shaft.
Now, the 2,195,605 patent represents what I assume is an early prototype stage of the Automatic Safety Transmission, and has some major differences from the production transmission, so it’s most useful as a way of illustrating the evolution of the design rather than its production form. The patents that are closest to the production Automatic Safety Transmission are 2,193,524 and 2,362,418, originally filed as a single application in March 1937. (There are still some points that differ from the production transmission, so one must be cautious about that.) Fig. 1 in both those patents shows the forward-reverse unit, with the engine presumed to be at the right; the shaft numbered “5,” at the far right, is the driven shaft of the main clutch. In those Figs. 1, note the star-shaped gear labeled “175”: That’s the drive gear for the pump and governor shafts.
This gear is also illustrated in Fig. 1 of the Automotive Industries article, although there, it’s depicted as a series of concentric circles rather than having the star shape. A point that may be throwing you here is that compared to the patent illustrations, the diagram in Fig. 1 of the Automotive Industries article is oriented in the opposite direction: In the latter figure, the clutch driven shaft, labeled “A,” is at the left of the diagram. If you look for those labels in both figures, it will help to visually orient you so you can better tell what you’re looking at. The Automotive Industries cross-section assumes the engine is on the left, with the output shaft on the far right; the patent cross-sections do the reverse.
Because the pump/governor shaft that gear drives is again in a different plane (it’s perpendicular to the clutch driven shaft), the pump and governor assemblies are not shown in the longitudinal cross-sections. If you examine Fig. 3 of the patents and look for gear 175 (which in those illustrations is roughly at the center), you’ll see it from a different angle that better illustrates its construction. The comparable illustration in the Automotive Industries article is Fig. 4, where the drive gear isn’t specifically labeled, but can be identified by looking at the gear teeth of the two shafts that converge approximately in the center of the illustration.
I must admit that I still had difficulty visualizing where these components were located until I studied the perspective view in Fig. 6 of the Automotive Industries article. Note that the pump drive gears (still not labeled) are just below the word “MAIN” in the label for the main shaft. That illustration is “exploded” somewhat for visual clarity (I think the pump and governor shafts are shown to be longer than they actually are), but it makes their relative positions a good deal easier to understand.
(I hope this doesn’t sound unduly lecture-y. I also had quite a struggle getting my head around this layout, which wasn’t helped by the fact that all the times I had previously seen reproductions of Fig. 1 of the Automotive Industries article, they a) were very small, making it hard to determine at a glance which end was which; b) did not include the key, making the labels useless; and c) did not cite the specific issue in which the diagram originally appeared, leaving me unable to find the actual article. Sorting it out was kind of a project.)
Aaron, it looks as if the “missing” pump for the 2,195,605 AST is actually the ENGINE’s oil pump! So engine oil pressure is sent to the hydraulic control of the transmission.
(I can see why this was not implemented in production, especially when it was later found that sperm oil was useful for transmission life. Also, I wonder what would happen when/if engine engine oil leaked into the transmission oil. Did they make a kind of common sump?)
According to the Automotive Industries May 1937 article, the production AST was self-contained, oil pumps, oil sump, and all.
Aaron, I had intended to respond to your other points but, after my governor flub, it’s probably best if I lie low. If there were any way I could atone by helping you with corrections, I’d be glad to. But my best approach is likely do do no harm; i.e., keep my hands off the keyboard!
Don’t sweat it — there is no shame in being confused by patent illustrations!
Thanks, Aaron.
If nothing else, this exercise enhanced my already high opinion of Earl Thompson.
Looking at those old patents showed how very creative and clever his work on the AST was.
And his 2-stage output-driven governor was not really needed for the AST, as the rear planetary was shifted by hand. But it fully convinced me that the AST was only an intermediate step, and — even back when he “only” automated the rpm-matching of a manual transmission — Earl had always intended to build a fully-automatic transmission.
And he did…with spectacular success.
To be clear (if only for the benefit of other readers trying to follow this conversation), the Automatic Safety Transmission governor was driven, like the pump, off the input shaft at engine speed. The earlier 2,195,605 patent you looked at (which DID specify driving the pump and governor off the output shaft) represents an earlier iteration of the design; although there may have been working prototypes of it, the transmission wasn’t offered to the public in that form, and the design described in that patent differs quite a bit from the transmission Oldsmobile and Buick actually built and sold. Neither the ‘605 patent nor the production AST had the dual-stage governor — that was introduced on the 1938 2,204,872 patent, which was also not sold to the public in that form, although it obviously presaged a lot of the concepts used for Hydra-Matic.
This is not to undersell the significance of the achievement. What all the iterations of the AST (production or not) clearly illustrate is that Thompson understood the fundamental challenge involved in creating a truly automatic transmission: the need for shift decisions to be based on at least two different parameters, speed (either engine speed or road speed) and load.
By the thirties, the idea of executing automatic shifts based on speed was pretty well-established, and there had been attempts to do that going back at least 30 years, usually via some kind of centrifugally controlled clutch(es). The problem (aside from various mechanical issues that fell more in the realm of quality control problems than actual conceptual flaws) was that a shift mechanism that responds only to speed can only respond to increased load through a loss of momentum. Transmissions like that would only automatically downshift if the engine didn’t have enough torque to maintain its speed, or by the driver executing some kind of manual downshift, neither of which was desirable.
Thompson recognized that each shift point actually needed to be a range of speeds that could be automatically adjusted upward or downward based on throttle position. The way the AST did that was rudimentary, using rods and levers that the governor and throttle linkage cranked back and forth, and it only worked on one gearset, but it was a very important conceptual advance, and it worked well enough that Thompson’s team was able to ask, “Okay, how can we do that for all forward speeds rather just two?” They concluded that trying to accomplish that with mechanical linkages was going to be too complicated — an AST-style control linkage capable of handling all four speeds would have ended up being like the innards of a player piano — and so they devised the opposing pressures hydraulic system.
I said this in the text, but the Hydra-Matic controls were in a very real sense computerized. It was an analog computer operating via hydraulic pressures rather than an electronic computer using variations in electrical voltages, but it was still a computer, and a pretty remarkable breakthrough for the late thirties. Because of the Automatic Safety Transmission and the unproduced variations of it that are reflected in the various patent disclosures, we can see how that system evolved from the simpler and more rudimentary mechanical linkage, but it was still a tremendous achievement.
That is also the answer to the various questions to the effect of “Wasn’t Hydra-Matic just a Model T planetary transmission/Wilson preselector/Fluid Flywheel with some extra hydraulic controls?” Yes, there were other planetary transmissions, and other transmissions with fluid clutches; it was the controls, more than the mechanical layout, that made Hydra-Matic remarkable. People had been trying to make self-shifting transmissions for almost as long as there had been automobiles, but achieving reliable self-shifting based on even a single parameter (much less two) had been a real bear, and the results hadn’t ever really been satisfactory even when they weren’t accidentally flying to pieces. Hydra-Matic could not only balance two input parameters, it opened up various possibilities for automatically tailoring shift points and even shift firmness based on a variety of different inputs.
Gee, Aaron, I took the information showing the output-driven governor out of the article from the Automotive Industries of May 29, 1937. I was under the impression that it reflects the production version. If so, it clearly has the secondary pump and governor driven from the drive wheels, both in illustration and in text. Also, mentioned in the text was “a two-stage flyball governor.” What am I missing now?
I think you may be assuming the longitudinal cross-section in Fig. 1 of that article is oriented in the opposite direction; the forward-reverse unit is at the FRONT of the transmission, not the rear. Take a look at Fig. 1 (p. 806, the first page of that article). In this cross-section, the engine and main clutch are presumed to be at the left; the clutch input shaft is labeled A, at the extreme left. To the right of that is the forward-reverse unit, with the input gear B and the sliding gear C next to it. To the right of the sliding gear, between the main shaft and the countershaft, you’ll see a set of concentric circles, above the worm gear indicated as “F.” That is the pump/generator drive gear; as the body text on that page says, “a worm F through which the oil pump of the control system is driven.” Flip over to page 808 and look at Fig. 4, which is a through section of the pump and generator drive. The gear teeth in the middle, below gear W and above gear F, is the same gear indicated by the aforementioned concentric circles in Fig. 1.
That gear is driven off what the article calls the “intermediary shaft D.” The trailing end of that shaft is integral with the ring gear G of the first planetary unit. With the forward-reverse unit in forward drive — with the sliding gear C meshed with the input gear B — shaft D and ring gear G rotate forward at engine speed. Since the pump and generator drive is geared off that shaft, their speed is thus also proportional to engine speed, NOT to the speed of the output shaft, which is on the opposite side of the transmission. (In that Fig. 1, the driveshaft flange is on the far right of the diagram, on the other side of the two planetary units.) You can also see this if you look at the relative positions of the pump and governor in Fig. 6 on p. 809. They’re driven off what that figure calls the main shaft, which is integral with the ring gear of the front planetary unit.
The text of the article does indeed describe the governor as “a two-stage flyball governor.” However, it’s important to understand how the governor of the Automatic Safety Transmission actually works. Its rotation doesn’t operate hydraulic metering valves, as on Hydra-Matic; rather, it cranks a lever connected through rods and pivots to the automatic shift valve. If this linkage moves far enough toward the rear of the transmission (toward the right, in Fig. 1), it pushes the automatic shift valve to the upshift position. That motion appears to be what the Automotive Industries writer is describing as the first “stage.” There is also a second mechanical linkage, connected to the throttle, which serves to change the effective length of the automatic shift linkage. The more the throttle is advanced, the farther the governor has to crank the shift valve linkage to push the automatic shift valve into the upshift position, and thus the higher the speed of the shift point. That motion is what the text is calling the second “stage.” It’s not ideal wording, in my estimation, because describing this arrangement as “two-stage” doesn’t really convey what’s intended to be a progressive motion of the throttle linkage. It definitely has nothing to do with the dual governor weight arrangement found on Hydra-Matic, where the governor is essentially two governors in a common housing.
To your earlier point, this arrangement WAS really needed on the Automatic Safety Transmission, because it allowed the automatic shift point of the front planetary unit to vary based on how far the throttle was advanced. Without it, the automatic unit would always shift at a predetermined engine rpm, regardless of load, and would not downshift without a significant drop in engine speed. The throttle-controlled linkage made the automatic unit capable of what would later be called “kickdown,” and of holding the lower gear for longer with the throttle advanced, which was the important conceptual breakthrough I was talking about earlier. To quote Thompson’s 2,193,524 patent, this “yields a selective effect of engine braking and acceleration … and permits the driver to establish a fixed reduction ratio for gradient work where torque rather than fuel economy is desired.”
Hi, Aaron. In response to your June 19th post at 5:56 PM:
I do not agree that “the AST governor was driven, like the pump, off the input shaft at engine speed.”
I generally agree with all your terminology, though, with the “concentric circles”, front & rear, input & output, etc. But I’d like to focus on the AI paper of May 1937, as I assume it reflects the production version of the AST.
There are actually two transverse shafts (a hollow outer one and a 2-piece inner one) which drive the primary oil pump and the secondary oil pump + governor, respectively.
As you say, the engine-driven countershaft gear F drives the hollow OUTER shaft.
So the primary oil pump rotates in direct proportion to engine rpm. No argument.
BUT the secondary oil pump and the governor are driven by the INNER shaft which,
in turn, is driven by gear W via a splined-sleeve-with-gear, just below gear W.
(And the splined-sleeve-with-gear also locks the two inner shafts together as one.)
Significantly, though, gear W does NOT rotate with the engine!
Rather, it rotates with what Fig 6 calls the “main shaft”.
As per the text, the main shaft (aka intermediate shaft) “is operating whenever the rear wheels are turning.” This provides a ground-driven secondary oil pump and a ground-driven governor. So the governor is NOT rotating with the engine.
Rather, it is rotating in relation to the speed of the wheels.
Admittedly, the governor is NOT rotating in direct proportion to output shaft rpm.
Rather, I suspect it rotates either at output shaft rpm or by the ratio of the rear planetary gear-set. (I can’t be 100% sure because I can’t quite translate the cross section to a gear schematic.) In any case, this would appear to make it complicated to automatically shift the input planetary via signals from governor (and throttle).
But it worked.
BTW: For these discussions of governor drive location, it doesn’t really seem to matter much if the logic of modulating the governor signal via the throttle signal is done mechanically (like the AST) or hydraulically (like the Hydramatic).
BTW2: I can fully appreciate the difficulty of sorting out how the AST works!
Before responding, I had to go back to look at the patents, which have a more detailed description of the pump and governor drive, and whose illustrations of it (Fig. 2 in both the 2,193,524 and 2,362,418 patents; the illustrations are identical) better depict the relationship of the driving gears than do the illustrations in the Automotive Industries article.
I think part of the confusion here comes from some odd and inconsistent wording in the Automotive Industries article. The “main shaft” in Fig. 6 is what the primary text describes as the “first intermediary shaft” or “intermediary shaft D,” which the patent disclosures and illustrations as “shaft 8.” That’s the shaft to which the sliding gear is splined, which is integral with the ring gear of the front planetary unit — it’s the input member of the planetary gears with the transmission in forward or reverse drive.
You’re right that that shaft doesn’t necessarily move at engine speed. With the sliding gear in the neutral position, it might not move at all, and with the sliding gear meshed with the reverse idler, the shaft turns backward, which also drives the secondary pump backward, opposite the primary pump. However, as I said yesterday, with the sliding gear meshed with the input gear, the forward-reverse unit is in direct drive, and shaft D/shaft 8 therefore turns forward at engine speed. Neither the Automotive Industries article nor the patents note the actual ratios of the pump drive gears (which might be 1:1, although they don’t specify one way or the other), but at that point, the speeds of both pump shafts and the governor will be directly proportional to engine speed if not equal to it.
It may be worthwhile to note that because the Automatic Safety Transmission had no fluid coupling, the car must be in motion if the forward-reverse unit is in either forward or reverse and the main clutch is engaged — otherwise, the engine will stall. So, in that sense, the speed of shaft D/shaft 8 and thus the governor and secondary pump will also be proportional to the speed of the output shaft, since all are in mesh with the same gear train. I think it’s more accurate to describe the planetary gearsets and output shaft as the cart rather than the horse in that particular train, although I suppose that’s again more a philosophical distinction than a technical one.
There is one scenario in which the pump shafts (and by extension the governor) may be driven by the output shaft rather than the engine, which is when push- or tow-starting a stalled car. The Automatic Safety Transmission servos were arranged so that both planetary gearsets would be in direct drive (clutches engaged, brakes released) with the hand selector in neutral, so that pushing or towing the car would rotate the main shaft (shaft D/shaft 8). Under those conditions, the output shaft is driving the governor and secondary pump even though the engine isn’t (yet) running at all.
I should note before I forget that this arrangement only applies to the 1937 and 1938 versions of the Automatic Safety Transmission, which had the unusual dual pump drive. The 1939 version had a simpler single pump, with only one set of internal gears. I unfortunately have not been able to find any illustrations or detailed description of the revised layout.
Thanks, Aaaron, for your June 20th reply at 7:15 PM. I appreciate your taking the time and effort to respond. I now realize that, in order to satisfy myself, I will have to take the time to slog through the drawings and draw a schematic of the AST gear system so I can better understand what rotates with what. The AST is a deceptively complex design, and I can now better understand why I can’t find any authoritative schematic of it!
And because our recent discussion revolved around the governor, I can also better appreciate how significant it was that Thompson incorporated governor modulation via a throttle signal into the AST controls. Most of my career was spent with commercial vehicles around the world. And even in Europe and Japan, many people struggled with the idea of getting something as “touchy-feely” as the driver’s wishes (via throttle position) into the function of a non-living mechanical device such as an automatic transmission. So Earl’s AST was seminal in many ways, and it is worthwhile to document and preserve its history.
Hi, Aaron. Toward an eventual goal of making a clear and simple AST gear schematic, including the pump and governor drives, I’ve been slogging through your materials, the Automotive Industries article of May 29, 1937 (AI), and Gott’s schematic in his “Changing Gears …” book.
I now accept my original contention that, with the AST selector lever in L or H and the pedal clutch fully engaged (i.e., “locked up”), the governor turns (along with the secondary oil pump) in direct proportion to ENGINE rpm, with the proportionality constant being the ratio of the worm-drive gear-set off the “first intermediate shaft”, aka “Mainshaft” in the AI article, even though there’s a “second intermediate shaft” that AI labels “M”. You agreed with my original contention and changed your AUwM text accordingly. So we are now in full agreement.
(I had subsequently retracted my contention but, as you suggested, some of my confusion was based upon AI’s inconsistency in shaft terminology and labeling. AI’s text also asserted that the governor was driven by the rear wheels. But I now fully agree with you that, in forward motion with pedal clutch engaged, the governor is driven as a linear function of engine rpm.)
So, anyway, I am now fairly comfortable with how the AST gearing works, including pump and governor drives.
HOWEVER, I am now hoping you can help me with understanding the hook-up of the AST’S Rear Planetary Clutch. It is clear that the OUTER teeth on the clutch plates are splined to the assembly which holds the rear planetary’s front ring gear (aka annulus).
My difficulty lies with the INTERIOR teeth on the matching clutch plates, which are splined to what I’ll call a “clutch hub.”
There are two conflicting viewpoints:
— Gott’s schematic (Figure A.1) shows the clutch hub is attached directly to the shaft which AI calls “second intermediate shaft M”.
— Your AUwM text says “With the band released and the rear multi-disc clutch engaged, the front annulus was locked to the second planet carrier and the unit was in direct drive.” By “second planet carrier”, I assume you mean the carrier in the 2nd planetary from the front of the transmission (not the 2nd — or rear — carrier in the compound rear planetary unit). This suggests that the clutch hub is directly attached to the “second planet carrier.”
The effect of both the above viewpoints would bring the same result when the rear clutch is applied — a locked rear compound planetary rotating as a unit. However, in order to make an accurate schematic, I would sure like to know what the clutch hub is actually attached to. But when I look at the AI cross-section in Fig. 1, I just can’t figure it out! (And Fig. 6 is no help, either.) CAN YOU HELP ME?
I went back to study the patent text, and I think I made an error in my revision of this article text last week, having been confused in the same way you are.
I agree with you that the illustrations both in the Automotive Industries article and in the patent are not especially illuminating in this regard, and the text of the Automotive Industries article doesn’t have much to say about the operation or arrangement of the rear unit except that it has a bigger clutch because of its greater net reduction ratio. (I suspect the writer didn’t want to have to try to describe the way the rear gearset is compounded.)
So, this leaves us with the actual text of the 2,193,524 and 2,362,418 patents. Some of that text is quite opaque, but in its description of the rear planetary unit, the text says, “Clutch drum 59 is keyed to rotate with shaft 21, and is splined for clutch plates 60. Springs 89 are release biasing means for clutch 55—60, the plates 55 being keyed to rotate with drum 39, extended at 56, in which portion clutch cylinders 75 mount pistons 76, piston pins T bearing against presser plate 78.” In the patent disclosures, “shaft 21” refers to the intermediate shaft, which the Automotive Industries article calls the “second intermediary shaft M.” That shaft is splined to the planet carrier of the front unit, and the two sun gears of the rear planetary unit are integral with it.
Studying Fig. 1a in both patents, I agree with you that what the patent text is calling “clutch drum 59” is more accurately described as the clutch hub — the pointer line identifying item 59 is pointing to the central hub of the clutch rather than the outer drum, which the text identifies as “drum 39” and later as “reaction drum 39.” Fig. 1a suggests that clutch “drum” 59 is splined or keyed to a large gear or hub on the intermediate shaft just ahead of the two sun gears, which, perplexingly, does not get an identifying number and is not directly described in the text. (It also isn’t indicated in the Automotive Industries illustrations that I can discern.)
Nonetheless, the patent text subsequently states, “Clutch 55—60 as in Figure 1a couples reaction drum 39 to shaft 21 to establish direct drive in the rear unit, and brake 90 prevents rotation of drum 39 and annulus gear 42 to establish geared drive.” So, it appears that with the clutch engaged, the driving plates 60 on hub 59 are pressed against the driven plates 55, which causes the intermediate shaft to drive the whole drum assembly. Since the drum is affixed to the first (reaction) ring gear, engaging that clutch causes the first ring gear to rotate forward with the sun gears at the same speed. With no speed difference, the first carrier and second ring gear also rotate together and the whole rear unit is in direct drive.
What I meant by “second planet carrier” in the text of my article was the second planet carrier of the compound rear unit (what the Automotive Industries text describes as “the second-reduction train”). However — crucially — I think I was wrong. Thus, the text that said:
… should actually read:
Thank you for pointing out this confusion. I started working on a diagram last week and haven’t finished it because I’ve had a bunch of other things to do, but if I had, it would likely have ended up being wrong!
Incidentally, I noted that there’s a significant typographical error in the Automotive Industries article. On p. 808, it says, “Carrier U is formed integral with the main drive shaft of the transmission,” but the preceding sentence had just stated that the carrier of what the text describes as the “second-reduction train” is identified as V, with U identifying the second internal gear (ring gear). This is obviously just an editorial glitch, but since you’re looking at that text, I thought I should point it out so it doesn’t trip you up.
As I understand it, the Wilson and Cotal preselector transmissions did not use a clutch. Is there some technical reason why they couldn’t have been semiautomatics in the mold of the Saxomat and Sportomatic? If there was no reason why they couldn’t been, is there a reason why they weren’t?
A while back I found a web page (whose URL escapes me now) for Daimler buffs that said that if you’re a gearhead, driving a preselector-equipped Daimler should be on your bucket list. So presumably a preselector had some perceived advantage.
I must preface this by saying that I haven’t studied the Wilson and Cotal preselectors in enough detail to talk about them with any confidence; they are complicated, there were several different versions, and are surprisingly poorly documented (at least in sources I can reasonably access). I will say that they did have clutches — the Cotal transmissions, as I understand them (which is to say only a bit), had as many as four, electromagnetically operated — although they didn’t necessarily have a main clutch the way the Automatic Safety Transmission did. (Some iterations of the Wilson preselector were combined with a fluid coupling, which of course is a clutch of a kind.)
“Semiautomatic” is a troublesome word because it can mean several different things. The Automatic Safety Transmission or Reo Self-Shifter were semiautomatic in that they shifted automatically sometimes and other times required some manual gear selection. A preselector transmission was semiautomatic in that each shift was executed automatically at the driver’s direction; unlike a manual gearbox, the driver didn’t actually execute any of the mechanical actions necessary to complete the gear change, although the driver did still have to decide which gear to select and initiate the shift. A Saxomat or Autostick wasn’t exactly either: The driver still had to both initiate and execute the shift, with only the declutching part automated. (A Chevrolet Torque Drive transmission would qualify, though.)
Of course, that’s somewhat obfuscated on the Saxomat, Sportomatic, and Autostick by the use of a torque converter, which meant that you could at least sometimes just rely on the converter without actually changing gear; whether one regards that as automatic or semiautomatic is more of a philosophical question than a technical one.
Hi, Aaron. Toward an eventual goal of making a clear and simple AST gear schematic, including the pump and governor drives, I’ve been slogging through your materials, the Automotive Industries article of May 29, 1937 (AI), and Gott’s schematic in his “Changing Gears …” book.
I now accept my original contention that, with the AST selector lever in L or H and the pedal clutch fully engaged (i.e., “locked up”), the governor turns (along with the secondary oil pump) in direct proportion to ENGINE rpm, with the proportionality constant being the ratio of the worm-drive gear-set off the “first intermediate shaft”, aka “Mainshaft” in the AI article, even though there’s a “second intermediate shaft” that AI labels “M”. You agreed with my original contention and changed your AUwM text accordingly. So we are now in full agreement.
(I had subsequently retracted my contention but, as you suggested, some of my confusion was based upon AI’s inconsistency in shaft terminology and labeling. AI’s text also asserted that the governor was driven by the rear wheels. But I now fully agree with you that, in forward motion with pedal clutch engaged, the governor is driven as a linear function of engine rpm.)
So, anyway, I am now fairly comfortable with how the AST gearing works, including pump and governor drives.
HOWEVER, I am now hoping you can help me with understanding the hook-up of the AST’S Rear Planetary Clutch. It is clear that the OUTER teeth on the clutch plates are splined to the assembly which holds the rear planetary’s front ring gear (aka annulus).
My difficulty lies with the INTERIOR teeth on the matching clutch plates, which are splined to what I’ll call a “clutch hub.”
There are two conflicting viewpoints:
— Gott’s schematic (Figure A.1) shows the clutch hub is attached directly to the shaft which AI calls “second intermediate shaft M”.
— Your AUwM text says “With the band released and the rear multi-disc clutch engaged, the front annulus was locked to the second planet carrier and the unit was in direct drive.” By “second planet carrier”, I assume you mean the carrier in the 2nd planetary from the front of the transmission (not the 2nd — or rear — carrier in the compound rear planetary unit). This suggests that the clutch hub is directly attached to the “second planet carrier.”
The effect of both the above viewpoints would bring the same result when the rear clutch is applied — a locked rear compound planetary rotating as a unit. However, in order to make an accurate schematic, I would sure like to know what the clutch hub is actually attached to. But when I look at the AI cross-section in Fig. 1, I just can’t figure it out! (And Fig. 6 is no help, either.) CAN YOU HELP ME?
I’m confused — this appears to be the same comment you left yesterday afternoon, which I replied to previously. Was this submitted in error?
Here’s another attempt to respond to your post of June 24, 2023 at 5:21 PM:
Thanks, Aaron for looking into this! It can sure be tricky figuring out the AST.
If I understand you correctly, the “clutch hub” is splined to AI’s second intermediate shaft, M. If so, this means Gott’s schematic is correct (at least for the planetary section of AST). So I agree with your rewrite.
Thanks also for telling me about that typo on AI page 808.
BTW: Do you know the count of gear teeth on the 3 AST planetaries?
That’s what the patent text indicates. Based on Figure 5 of those two patents (which as previously noted have identical illustrations, having originally been submitted as a single application), each of the individual clutch discs has teeth around its inner circumference that mesh with splines on the intermediate shaft (called shaft 21 in the patent disclosures, intermediate shaft M by Automotive Industries). Whether that’s true in the production transmissions I’m less certain, but however it was executed in production, the patent text indicates that the hub of the rear clutch is arranged to turn with the intermediate shaft 21. In this way, when the clutch engages, it allows the intermediate shaft to drive the rear clutch reaction drum and the first of the rear planetary unit’s two ring gears.
(The reason I’m hedging about details in the patent disclosures is that the patent disclosures describe a number of features that are not present in the production transmission, in particular an additional drive range labeled “3” whose purpose is essentially the same as D3 in the later Dual-Range Hydra-Matic, albeit executed differently in keeping with the mechanical nature of the semiautomatic transmission’s governor and throttle linkage controls. Where the details described in the patent text and illustrations are obviously the same as the ones illustrated by Automotive Industries, I think they can be safely assumed to represent the production version, but this is a gray area because the article doesn’t really talk much about how the rear planetary unit is arranged.)
I do not — only for early Oldsmobile Hydra-Matic transmissions. The AST ratios are quite similar, though not exactly identical, to the gearing in early Cadillac Hydra-Matic units, which also went into tanks, so if you were very determined, you might be able to make a reasonable guess by figuring out the tooth counts for the latter, perhaps by studying the photos in the wartime technical manual (TM 9-1727C). (I didn’t have the patience to actually count gear teeth in the photos of the PDF scan I have of that manual, whose photos aren’t really crisp enough for that sort of thing, but you might have better luck.)
As a point of reference, the 1940 Oldsmobile Hydra-Matic had a front unit with 24 teeth in the sun gear and 54 teeth in the ring gear, giving a ratio of 1.444:1, and a rear unit with 45 teeth in the sun gear and 69 teeth in the ring gear, giving a ratio of 2.533:1. However, the Oldsmobile units didn’t have the compound rear planetary set used by Cadillac.
The published ratio of the front planetary unit of the AST is 1.42:1. Since there are three planet pinions in the front unit, for them to be evenly spaced, the teeth counts for the sun gear and ring gear should each be evenly divisible by 3. If there are 24 sun gear teeth and 57 ring gear teeth, both of which are multiples of 3, that would give a ratio of 1.421:1, so that would seem a plausible combination. (Keep in mind this is a guess, not a fact!)
Since the rear unit sun gears of the compound gearsets have to be identical, and the teeth count of both ring gears is likely divisible by 3, one could probably calculate a reasonable estimate of the number of gear teeth based on the final ratios (2.23 for the Automatic Safety Transmission, 2.26 for Cadillac and tank units), but that’s more algebra than I have the presence of mind for at this hour.
Thanks, Aaron, for looking into this. Regarding the clutch hub, I understand that we’ll probably never know definitively, but I think your evidence is strong enough that I’m going to use it as my basis of attempting an AST schematic.
Thanks also for your subsequent post about Hydramatic ratios. Good stuff!
But I’m also not sure if I have the nerves or the time to calculate things backwards.
My surmise is that the Automatic Safety Transmission rear clutch hub is probably similar to the early Hydra-Matic. If you take a look at the wartime Hydra-Matic technical manual, there are photos of assembly and disassembly procedures that show how the clutch plates are installed, which look to be similar to what Thompson described in the AST patents. (Disassembly procedures for the rear clutch are on pp. 112–115 of TM 9-1727C, assembly procedures are on pp. 164–166.) On an early Hydra-Matic, the rear clutch hub looks sort of like an external gear: a cylindrical metal donut with external teeth that’s held on the intermediate shaft with a snap ring. The actual clutch plates, which have teeth around their inner circumferences, slide over the hub, with their internal teeth meshing with the hub’s external teeth.
The big difference in this regard is that in Hydra-Matic, the intermediate shaft is hollow, and essentially terminates at the clutch hub rather than also driving the sun gear(s), which are instead driven by the fluid coupling turbine (driven torus) through the main shaft. In the Automatic Safety Transmission, the intermediate shaft continues behind the clutch hub to also carry the rear unit sun gears.
I took a look through the TM 9-1727C technical manual, and there’s not really any good unobscured photo of the rear sun gears or ring gears. In the best photo it does provide of the rear reaction drum (on p. 111), someone’s hand is covering parts of the first ring gear, but looking at the uncovered half suggests a total teeth count in the same ballpark as the non-compounded Hydra-Matic, whose rear ring gear had 69 teeth. (It might be a bit more than that — perhaps as many as 75 — but it appears to be in that realm.) Depending on your patience for doing algebra, that would probably be enough for calculating a reasonable estimate.
With the rear band engaged, the speed of the first planet carrier would be the intermediate shaft speed divided by (1 + R1T / S1T), where R1T is the number of teeth in the first ring gear and S1T is the number of teeth in the first sun gear. That reduced speed then becomes the speed of the second ring gear. The second sun gear turns at intermediate shaft speed, so the speed of the second carrier and output shaft is equal to: (R2T * the speed of the second ring gear + S2T * intermediate shaft speed) / (R2T + S2T), where R2T is the number of teeth in the second ring gear and S2T is the number of teeth in the second sun gear. We know the speed of the second carrier and the output shaft (intermediate shaft speed / 2.23 for the Automatic Safety Transmission, intermediate shaft speed / 2.26 for early Cadillac Hydra-Matic units); S1T needs to equal S2T; R1T is approximately 70 teeth (give or take); and the numbers of teeth in the sun gears and ring gears are probably multiples of 3.
I don’t have time right now to simplify that any further or to make a spreadsheet to calculate the values, so I leave that as an exercise for the interested reader, but you get the idea.
(With regard to TM 9-1727C, I should point out for reference that the page numbers I noted above are the page numbers shown on the original document and may not reflect the pagination of the various scanned copies of the document online, which are somewhat variable.)
Hi, Aaron. In working on a schematic for the Automatic Safety Transmission, I came across a possible typo in the AUwM color-coded table for the AST Gearing Sequence.
That is, for the Rear Planetary Gearset in 3rd gear, you show the Band Servo ENGAGED, yet the Band is ON. But shouldn’t the Band be OFF in 3rd gear?
(If both Band and Clutch were applied, wouldn’t shaft M be locked to the transmission housing?)
Ah! You’re absolutely right. I should have looked more closely at the table when I was editing that section.
(I do most of the composing and editing of articles in HTML — I don’t trust the WordPress WSYIWYG text editor — and one of the unfortunate consequences is that it makes managing tables a real pain. A raw HTML table is ugly and hard to parse visually.)
Thanks, Aaron. I realize how truly difficult it is to proofread one’s own material!
So I’m developing a lot of tolerance for “errors” in articles and text.
Especially when sussing out these old car technologies, it is extremely difficult to get things straight, particularly when the source data is not necessarily accurate.
Especially with the AST. There are a few remaining AST cars out there, but VERY few.
And I’d bet most of the owners wouldn’t be up to someone tearing into their AST, just to figure out if the literature is correct!
With all this in mind, your articles are TRULY excellent.
In this case, it was mostly a mechanical error resulting from my not having thought to look at the preview of the table to recheck its contents, which I should have done.
A raw HTML table first lists all of the columns, and THEN all of the contents of each row, cluttered with various formatting parameters, including, critically, the table header tags that indicate whether a row or column should be considered a heading or not (which the WordPress editor does not have any way to add, a serious accessibility flaw). This is all quite important, but it makes actually reviewing the contents of the table to make sure they’re correct several steps more complicated than it would be in something like a modern word processing document. (There is of course not necessarily any guarantee that I would notice an error like that in the composed table, for the reasons you mention, but I would at least be able to SEE it without taking several additional steps to view it!)
The table issue is one of the reasons I still don’t trust the WordPress text editor, whose limitations continue to drive me right up a wall. (It will generate a basic table, but not an accessible one, and may strip out my attempts to create a properly formatted HTML table in the interests of code sanitization; there are third-party plugins that provide additional options for managing tables in WordPress, but I don’t know if their output meets accessibility guidelines.) And this is without using the awful block editor they’ve been trying to force onto users since 2019.
Gee, Aaron, that WordPress stuff sounds awful!
Most of the papers and presentation slides I did were for work and were written with MS Office. So, for example, I’d just create a table in Excel, where I could directly see what I was generating. Then I’d paste it into a Word doc or Powerpoint slide.
When submitting a paper or slides to some tech organization, I’d sometimes have to adhere to their proprietary format, which could be a bit of a pain. But nothing like your nightmare with tables.
Then again, you are publishing on a much greater scale and distribution than my stuff, with lots of running interaction with readers. So you need to get some system to do it. I just don’t know enough about it or the the available options. Whatever, you have my sympathies. I’m now even more amazed at what you’re doing!
The WordPress content management system acts sort of like a template, where you create a layout with menus and footers and the like with a blank area where the text and images for a specific article are loaded. The CMS itself provides the means for managing various administrative functions, like security for administrative logins, automated backups, menus, comments, and things like that, which is useful. Text editors for this kind of CMS — the area where you actually input the content — still lag behind word processing programs in many respects; it’s like creating a document in a very old version of Microsoft Word or WordPerfect. Part of the dilemma in this regard is that the content needs to be formatted in HTML so that web browsers can read and display it in a reasonably consistent way. HTML has its own peculiarities, and the ways it deals with things like formatting are somewhat different from how modern word processing formats like a DOCX file handle them. (Symbol characters may display differently too; I’ve learned it’s often safer to use the HTML codes for specific symbols than trust that they’ll display properly when pasted from a Word document.) In theory, you can save a Word document in HTML format, but the code it creates will often be very, very messy and full of potential pitfalls. A traditional CMS text editor tries to sort of bridge the gap by giving you an interface that looks like a word processor, or at least a rich text editor in an email client, but that will produce relatively clean HTML, and that will limit the kind of coding errors you can create if you write HTML code by hand.
The WordPress CMS a couple of years ago implemented a new editing system called a block editor, where each section of content is a “block” that can hold different types of content, like an embedded video player, a picture, or a block of text. In practice, this makes composing all but the simplest posts like trying to write a term paper in a PowerPoint presentation, or with every paragraph in a separate text box; it’s infuriating, and I’ve had to install a third-party plugin that turns it off and makes the editor work the way it used to work. That still isn’t ideal because the “classic” text editor will still do fluky things: It will try to “strip” (remove or sanitize) certain types of code, and the way the WYSIWYG display appears to be handling things isn’t necessarily how it will actually look to a website visitor. I find this exasperating, so I will compose and edit posts in an open source text editing app called Notepad++, which highlights the HTML “tags” so I can more easily see if they’re mismatched, mistyped, etc. However, this still isn’t super-convenient for more complex structures like multi-level numbered lists or tables, where it’s necessary to check to see if it displays in a browser the way it’s supposed to.
With tables, the other issue is accessibility. Some people use screen readers or other assistive technologies to browse the web, and it’s important to arrange the code in ways that will facilitate that. (Accessibility is also an issue with Word and Excel documents, although what needs to be done about it sometimes differs quite a bit.) For instance, with an HTML table, it is important to include code that says, in effect, “This cell is part of a header row or a header column,” so that screen readers are better able to present the content to users in ways that make sense. The WordPress text editor doesn’t do that well, either the classic editor or the block editor, and so I have to do it by hand. (Originally, I presented tables as images, which I later discovered is absolutely the worst way to go about it from an accessibility standpoint because it means people using assistive technology won’t have any idea what’s in the image other than whatever is in the ALT tag and the image caption. This is also true of Word documents, and recent versions of Word will at least bug you to add ALT text.)
There are a couple of third-party WordPress plugins that are supposed to give you more options for managing tables, including the possibility of pasting in an Excel spreadsheet the way you might in a Word document. That might be handy, but I don’t know how the output does in terms of accessibility, which is daunting; before implementing such a thing, I’d have to do a fair amount of experimentation to see what its HTML output looks like. It might be better, or it might be a whole lot worse, because plugin developers are not necessarily thinking about accessibility issues.
Wow! I’m amazed your nerves can cope with all that.
The original reason I started composing things in HTML in a text editor was that the Joomla! CMS I used before was worse. It had a problem where the tokens used to log into the administrative back end would expire after a while (I think it was either 20 or 30 minutes) without any warning if the system thought you were idle. Unfortunately, it considered typing or editing text in the text editor to be “idle,” so if I went to write or edit a post and took a little too long, my login cookie would expire and if I tried to save my work, it would tell me I wasn’t logged in and go back to the login screen, dumping whatever I was doing without saving a draft, leaving me with no recourse other than vainly yelling, “Noooooo!” as all my work disappeared. The WordPress CMS has an auto-save feature (as Word does), but it only takes being burned a few times to make one eternally mistrustful of any such thing. I grant that the Joomla! version I was using was older and newer versions would probably have been better, but upgrading or even reinstalling anything would have been almost as difficult as starting from scratch, and WordPress worked better in every particular, even if it has since found new and different ways to try my patience and waste my time.
It does make some routine administrative stuff vastly easier, and it limits my need to have to code stuff myself, which is not my field and not something I’m good at or enjoy. Unfortunately, there are times when certain fairly simple things don’t have out-of-the-box solutions other than, “Oh, ha ha, you’ll have to code that yourself, good luck!” and it means that certain routine tasks end up having a lot of moving pieces I have to just remember to do manually.
Aaron, more on the AST schematic: It looks as if driving the secondary oil pump off the first intermediate shaft has the advantage of higher oil pump rpm when push-starting the car. Unlike the Hydramatic, which drives the rear pump off the output shaft, the AST will spin the smaller secondary pump as a function of output shaft rpm x 3.17 (1st gear ratio), even if the main pedal-operated clutch is not engaged.
Then again, even with no oil pressure, both bands will be on, so there will be a geared ground connection with the first intermediate shaft. So, engaging the pedal clutch would still turn the engine over. Still, it’s probably nice to have some oil pressure going through the AST gizzards while you’re trying to push-start the car.
Any thoughts about this? Maybe it was addressed in one of the patents?
Yes, push-starting (or tow-starting) is definitely one of the explicit design goals of the transmission, as it was for many automatic transmissions until really the early sixties, and maintaining pump action for that purpose was deliberate.
As the text of the 2,362,418 patent explains:
(By “jaw clutches,” Thompson just means the engagement of the forward-reverse unit’s sliding gear with the internal teeth of the input gear when the hand lever is in either forward drive position, so that the shaft 8/shaft M will rotate with the main clutch driven shaft.)
Theoretically, push-starting in neutral should have
both planetary gearsets in direct drivethe rear planetary gearset, but when starting from rest, there’s not enough pressure to engage the servo and clutc (which is also the reason starting in H actually gave you a first gear start). So, when push-starting, I think you’d get first gear initially, overdriving the secondary pump, as you note. Then, as pump pressure increased, you’d pretty quickly get a “shift” to third, still with the transmission driving the secondary pump (and governor, although it wouldn’t do anything in this regime). This shift would not involve any movement of the automatic shift valve; it would strictly be a function of developed pump pressure. The patent text explains:(I initially said this would cause a shift to direct drive fourth, but I realized that’s probably not correct because of the engagement bias of the front brake band. With the hand lever in neutral, I don’t think there’s any means for oil pressure to engage the front servo, and so as soon as the intermediate shaft started to rotate due to car motion, the band would probably hold the front sun gear stationary. So, when push- or tow-starting, the front gearset would still be overdriving the pump even after the rear gearset automatically switched to direct drive.)
I’m confused, too! I sent a totally different message yesterday, aimed at responding to your June 24 message at 5:24 PM. Not sure what happened. I have to resurrect it from memory, I guess. Sorry about the confusion.
Thanks, Aaron, for your June 28th post at 6:03 PM. Re your 1st paragraph, I was aware that that a ground-driven oil pump is needed for emergency push (or tow) starting of the engine. However, my real question (although I didn’t express it very well) was:
“Did the AST designer purposely drive the secondary oil pump off shaft M to take advantage of the 3.17 first gear ratio to make it spin faster than driving it off the output shaft?”
The citation introduced in your 2nd paragraph only says it’s good to drive the secondary oil pump by some mechanism that turns when the drive wheels are moving (i.e., ground-driven). But this would also be true if the pump were driven off the output shaft.
I agree with your parenthetical 3rd paragraph regarding what Thompson meant by “jaw clutches.”
However, I have trouble with the opening of your 4th paragraph: “Theoretically, push starting in neutral should have both planetary gearsets in direct drive.”
Rather, I believe that, with the car at rest in neutral, there is no oil pressure from either pump. This means that both bands must be firmly ON. (It takes oil pressure to turn the bands OFF.)
Therefore, I also believe there’s another typo in the AST Gearing Sequence table. That is, in neutral (with the front servo released), the front band should be ON (not OFF). If so, it also means that the asterisked comment below the table should be amended accordingly.
This is why, as you correctly wrote, starting in H actually gives 1st gear because the rear band is ON (i.e., rear gearset in reduction) due to a lack of oil pressure at launch. But, as the engine rpm and vehicle mph increase, oil pressure increases, and the rear band comes off while the rear clutch comes on to eventually give a 1:1 ratio of the rear planetary (i.e., 3rd gear). However, this overlap of band release and clutch apply usually results in a “tie-up” which is very tough on clutches. So starting in H turned out to not a good practice for durability. (I suspect that the 1st gear start in H was an unintended and unforeseen “feature.”)
Now, back to my question, “Why is the secondary pump driven by shaft M, not output?”
I believe that both the secondary pump and the governor should be driven by the same shaft (one that always rotates with the wheels). But, as I’ve mentioned before, the AST’s manual shift of the rear planetary means that it’s easier and better to connect the governor to shaft M, because only the front planetary shift needs a governor as it’s the only only automatic shift. In other words, the front planetary shift doesn’t really care what state the rear planetary is in.
In like manner, I believe the fully-automatic 1-4 shifting of the Hydramatic was better off with a governor driven off the OUTPUT shaft. So that’s also why the secondary oil pump was driven off the output shaft. In short, the governor function dictates which shaft it should be driven off; the oil pump just follows the governor.
But this is a complicated topic, and there might easily be something I’m not seeing.
If so, please let me know.
Ah, I see now what you’re asking. So, it’s important to specify that the secondary oil pump gear only benefits from the overdrive effect of the planetary gears being in reduction in one specific scenario: pushing or towing a car whose engine is stalled or whose main clutch is disengaged, so that the momentum of the vehicle rather than the engine is rotating the driveshaft. In that scenario, the overdrive effect of the planetary units would appear to be beneficial, insofar as the transmission is only able to drive the secondary pump shaft.
Thompson’s 1937 patents note that the pump is supposed to build up pressure rapidly enough to be able to operate both servos at “medium low speeds.” With the engine running, the clutch engaged, and the hand lever in either drive range, both pump gears are operating with additive effect, but when push-starting, the secondary pump gear has to do all the work. So, in the latter regime, I assume that having the pump shaft spin faster gets it up to operating pressure more quickly. (There is no specific mention of this that I could find in the patent text, but that would appear a logical inference.) However, this again raises the question of how the pump drive was arranged on the 1939 Automatic Safety Transmission, which had only a single set of pump drive gears. I unfortunately don’t have any further details on that.
Yes and no. The opening sentence you quote is, I think, incorrect, so I’ll amend that; when push-starting in neutral, I think the front gearset remains in reduction. (I had initially assumed that the controls were arranged to also put the front gearset in direct drive in neutral along with the rear unit, but after studying the patent descriptions, I’m not seeing any means of doing that.) As for the rest, it depends. With the car at rest in neutral, the primary pump gears will be operating as long as the engine is running and the clutch is engaged, which probably provides enough line pressure to engage the rear clutch and servo. With the car at rest in neutral with the engine off and/or the main clutch disengaged, then neither pump gear will turn and there will be no oil pressure to operate the servos or clutches.
That’s not a typo so much as semantic issue, although I think you’re right that the front band is most correctly listed as ON, and I amended the table. I left the asterisked comment, because it gets to a particular nuance noted in the patent text regarding the bands. With the front servo off, the front band is on; that’s true. However, Thompson’s patents emphasize that the bands used with the planetary gearsets are designed to have as little self-energizing effect as possible, and that the actual engagement of the band against the outside of the drum depends on having a torque reaction that attempts to rotate the drum in either direction. In forward or reverse, the main shaft turns the front ring gear, at which point reaction torque on the sun gear will attempt to turn the drum and lock it against the band; similarly, if the car is being pushed in neutral, the rotation of the intermediate shaft caused by the car’s momentum will exert reaction torque on the sun gear and the band will hold the drum stationary. However, if the car isn’t moving and the forward-reverse unit is in neutral, there’s no torque being applied to the front gearset from either end, and no reaction torque on the drum. Like I said, this is kind of a semantic detail and doesn’t change the fact that the front servo appears to remain released in neutral.
The first part is how I’ve understood it: that starting in H was possible, but not ideal, and that starting in L was easier on the transmission as well as providing better start-off performance. As for the parenthetical, I think you’re probably right, and I had previously said something to that effect in the text, but I’m now less sure of its intentionality. The patent text does note the line pressure issue that results in the first gear start, but doesn’t really comment on its desirability one way or the other. The Automotive Industries articles mention that the transmission will start in first when in H and then short-shift to third, but don’t note any specific factory injunction against doing so, which one might expect if Oldsmobile were trying to discourage the practice; I assume the Automotive Industries writers were working from an Oldsmobile press kit. When Buick offered the transmission for 1938, their brochures also noted that starting in H would give a 1–3–4 shift pattern, without any qualifier about how you should start in L. So, it seems that both divisions were willing to accept the behavior as a feature, even if it was more of a quirk that wasn’t great for the transmission’s mechanical health.
Ultimately, I think this issue was primarily a side effect of the Automatic Safety Transmission using a conventional main clutch. With the main clutch disengaged and the car stationary, the oil pump gears can’t turn at all. (If the clutch is engaged and the hand lever is in neutral, the primary pump shaft DOES turn, as it’s driven off the countershaft, but you have to declutch to put the car in forward or reverse, which still momentarily cuts off drive to the pump.) Thompson’s 1937 patent disclosures suggest that this was by design, making it possible to warm up the engine without uselessly running up line pressure in the transmission. However, it meant that when starting from rest, e.g., at a stoplight, the pump gears had to accelerate from zero or close to zero, rather than from idle, to reach operating pressure.
Hydra-Matic didn’t have that issue because the use of a fluid coupling rather than a main plate clutch made it possible to drive the front oil pump off the same input shaft that drove the ring gear of the front planetary gearset, which was driven by the torus cover and thus always rotated at engine speed even at idle. However, this made it difficult to provide operating pressure during push-starting without the use of a second rear pump.
Well, since Hydra-Matic and many subsequent GM automatics did that until the early sixties, it seems their engineers agreed with you! With the Automatic Safety Transmission, I think Thompson was trying very hard to avoid the need for two separate oil pumps, most probably for cost reasons. (The 1937 patent application notes, “The novel nature of the double pump system is an essential feature of my invention … and of unusual commercial utility.”) Oldsmobile was losing money on the Automatic Safety Transmission, which they were selling below cost, and which I think they were really only offering to the public as a way of partly offsetting the development costs, which were substantial. There were a lot of compromises, and I suspect that was one of them.
More or less. The central design assumption of the Automatic Safety Transmission, the possibly errant first-gear start in H not withstanding, was that the transmission would provide two driver-selected ranges with automatic shifting between two ratios in each based on engine speed, adjusted up or down based on throttle movement, but arranged to prevent over-revving (or lugging in the higher range). The vehicle road ranges at which those ranges might be used overlapped, so it made sense for the governor to rotate with the engine rather than off the output shaft based on road speed. As for the front gearset being the automatic one, so far as I can see, that was only relevant insofar as it may have made it easier to position the automatic shift valve physically closer to the governor and throttle linkage and thus make their linkage marginally less cumbersome and convoluted. (The governor and throttle operated a series of links and levers to physically operate the automatic shift valve rather doing so by means of regulated oil pressure.)
No, but the reverse is not true, as the Automatic Safety Transmission adjusts the operating pressure of the manual unit’s servo and clutch when line pressure is applied to the front servo and clutch, via a differential valve.
Well, yes, because Hydra-Matic had a very different control scheme than the Automatic Safety Transmission. As mentioned, the AST automatic shift valve controls are purely mechanical; the motion of the linkage rocks the shift valve back and forth between two positions. Because of the way the two planetary gearsets are compounded to produce four forward ratios, changing automatically between second and third requires shifting both gearsets at once, so for fully automatic shifting, each gearset needed to be able to go from reduction to direct drive and back to reduction in overlapping ranges of road speed. It might have theoretically been possible to accomplish that with a system of rods and levers, but it would probably have approached the complexity of a 19th century automaton, and as Kelley and Rosenberger noted in that 1947 SAE paper, they pretty quickly decided using regulated oil pressure was going to be simpler.
I should emphasize that using the governor as a two-stage pressure regulator did not necessarily require driving the governor off the output shaft. In Thompson’s 1938 patent disclosure (2,204,872), which outlines a fully automatic version of the Automatic Safety Transmission with hydraulic rather than mechanical controls, the governor is still driven in the same way as the AST and, as the text puts it, “partakes of engine speed” when the forward-reverse unit is in forward drive. However, I suspect the limitation was that the ratio spreads of these transmissions were wide enough that each shift would result in a pretty sizable change in engine rpm. For instance, on a 2–3 upshift at 3,700 rpm, engine speed would drop to about 2,350 rpm. Based on Thompson’s sample pressure curve charts, that would cause regulated pressure to fall from 80 psi to about 35 psi, which doesn’t seem terribly desirable from the standpoint of limiting the “hunting” between ratios. (The 1938 disclosure seems to get around this mostly by setting the rpm thresholds for automatic downshifts quite low.) With a hydraulic governor, it would seem to me preferable for governor pressure to be the same at a given road speed rather than be dictated by engine rpm. Again, Thompson seemed to think the latter would work, but they went with the former approach for Hydra-Matic, and most competitors followed suit, which seems to me revealing.
I wouldn’t necessarily assume that. There’s no particular reason the governor has to be combined with a pump, and of course from the early sixties on, it generally wasn’t. The use of the rear pump, which was obviously more expensive, was primarily about wanting to facilitate push-starting. On the early Hydra-Matic transmissions, the rear pump was also arranged to act as a “cruising” pump that would consume less power than the main pump during low-demand operation, but adding a separate oil pump as a — slight — economy measure seems penny-wise, pound-foolish, so I suspect this was a matter of trying to add value to a component the designers were already determined to retain. I assume by the sixties, better electrical systems meant that most stalled-car situations were better addressed with a jump start than a push (and if it was something else, like a plugged fuel filter or vapor lock, push-starting probably wouldn’t work either), so the extra expense was no longer considered worthwhile, but it was a significant preoccupation of early automatic transmission designers.
Thanks, Aaron, for your thorough reply.
It looks as if we won’t find a definitive answer as to why the AST secondary oil pump is driven via ground, but at engine rpm.It was likely done to get higher pump rpm to reduce the vehicle speed needed to get a good engine cranking rpm.But, then again, the Hydramatic used a secondary oil pump driven by the output shaft for push starting, and it worked fine.I would still suggest that the secondary oil pump was driven at engine rpm so the governor would be, too. Simpler.(Although I’ll admit it is not absolutely necessary, as was seen in patent 2,195,605.)
I sent you a separate email regarding how the 1939 AST got by with only a single set of pump gears.From a 1939 Olds service manual I just obtained, there is only pump, ground-driven off shaft S.But I still don’t understand how oil pressure is developed in the AST at rest or in Reverse. (And there’s no sign in the Olds manual of the engine’s oil pressure being sent to the AST.)
I’m still a little shaky about the table footnote indicating that the band is not really on until there is a torque reaction.When I look at those BIG springs which apply the front and rear bands, I wonder why the front drum isn’t held firmly.Admittedly, the front band (and clutch) are smaller, but that’s only because they see less torque than the rear unit.And the springs that apply the ATS bands look much like those in the later Hydramatic, and its rear band hold was strong enough to get a firm “park lock” (in combo with the reverse anchor). But who am I to question Earl Thompson?!
Then again, maybe Thompson was simply saying that, with no rotational motion of the engine or rear wheels, there would be no torque to be held with the band? But as soon as you would try moving something, there would be torque through the bands?Yet, for example, I would expect that, when parking the car, the driver could put the shift lever in F or R, then shut the engine off. The ATS would be “in gear”, just as if it were an MT parked in gear (a common practice). And the bands would hold it.But maybe Earl is just saying there’d be no torque on the band unless the car tries to move?Then again, the ATS is a complex unit, so maybe I’m yet again missing something?
Regarding the 1-3-4 sequence when launching in H, I have read and heard various versions of what was recommended.But when I look at all the Oldsmobile ads for the AST wherein the driver is shown, he/she is invariably holding on to the steering wheel with both hands, and two fingers of the right hand are on the gearshift lever, ready to flick it between L and H. This would not be done if people were simply leaving the lever in H and driving in a “just stomp and steer” mode.
On the other hand, I couldn’t find a single Buick ad which pictured someone driving the AST. But I just watched a YouTube video (1938 Buick Special Self Shifter — Presentation & Restoration) about fixing up an old Buick with Self Shifter (aka AST). It included lots of good AST technical information, including tooth counts for each gear, partially-colored original drawings of critical bits (like the governor/pump drive), etc. Interestingly, it also included a Buick sketch of the speedometer, showing in which gears the AST would be in at various vehicle speeds, depending upon launch in L or H. And in H, it shows the 1-3-4 sequence. So it was clearly as acceptable to Buick as launching in L and then shifting to H.
I also found another YouTube video (Buick Special Self Shifter) about the same car, and it shows the owner driving his restored Buick. He follows the Olds advert procedure of driving with his right hand on the steering wheel, but with two fingers on the shift lever. And he starts in L and flicks down to H at higher speeds, and flicks down to L when he slows down or stops again. Shift quality appeared to be very good. This is interesting because I’ve read a number of posts where people said the Buick AST shifted very roughly — “with a bang” — but they attributed it to the Buick’s torque-tube propshaft. Whereas the AST in an Oldsmobile was said to shift better because of its open driveshaft.
However, I have heard that argument about torque tubes exacerbating shift harshness in many different cars, but I never found much credence in it. For example, I have driven 1-2 auto-shifting Powerglides in 1953-54 Chevys with a torque tube and in 1955-57 Chevys with an open propshaft, but I never felt any significant differences in shift harshness between the two. I also drove many Buick Dynaflows where I made power-shifts between L and D, yet I didn’t see much difference in shift harshness with powershifting other automatics. Even with stick shifts, I have driven many examples with both torque tube and open propshaft, but I didn’t see any big differences in harshness during gear shifting. And lots of old cars used to have torque tubes. Admittedly, torque tubes have largely disappeared, but I don’t think shift harshness was the reason.
IMHO, an AST launch in H is not good for shift quality or for transmission life. Those tied-up shifts can be very rough, and they are hard on clutches and bands. But many people did it that way, and for a variety of reasons:
— Launching and driving in H was easier than shifting between L & H.
— Dealer salesmen (Olds & Buick) liked to demo how easy it was to drive.
(And they didn’t have to fix it!)
— Buick officially and clearly OK’d a launch in H
Some people feel that Buick recommended launch in H because they were happy to see the AST fail. Why?: As you said, there was bad blood between Buick and Thompson. Earl and his team had given a negative review of Buick’s Roller CVT, and many Buickians blamed him for the program being cancelled by GM. Although this wasn’t really true, Earl had indeed called their beloved baby ugly and had even recommended euthanasia! Whatever, Buick never installed a Hydramatic — aka Jerkamatic — in any of their cars, ever. (The THM 400 had the Hydramatic name for marketing reasons, but it was actually the first true “GM” transmission.) Unfortunately, GM decided to build the AST at a Buick plant, which didn’t help build quality or cooperation on running changes.
I fully agree that the governor doesn’t HAVE to be connected with a pump.However, when there is a requirement for push-starting (which there clearly was back then),then you need a pump which is ground driven so it puts out oil pressure when the engine is not turning but the wheels are.And you would additionally want a pump that runs whenever the engine is running.So it seems easier to me to combine the governor with the engine when the 2nd planetary is manually shifted, and it’s easier to combine the governor with the output shaft when you have 1-4 automatic shifting.
When I said that the front planetary shift doesn’t really “care” what state the rear planetary is in, I was talking about WHEN the shift occurs, which is primarily influenced by governor pressure modulated by throttle position.But, as you say, there are interplays of line pressure to the rear planetary depending upon the status of the front planetary.
But again as you say, as soon as there’s no requirement for push starting(which, in my view, was made much more feasible by going from 6-Volt to 12-Volt systems), then it’s best to just provide an engine-driven oil pump…period. But for shifting through all gears in the tranny, I still think it’s best to have a governor driven by the output shaft. And that’s the way most passenger cars became (before full-tilt-boogie electronic controls came along).
Gee, the AST is deceptively complex!
Well, I think the immediate priorities were fairly straightforward, for the most part. The original dual-gear pump arrangement ensured that there would be pump pressure in all but one specific operating regime (engine running with main clutch disengaged and car stationary), and, as you note, shared a common drive gear with the engine-driven governor, without the added expense and space requirements of a second pump. Thompson described as “of unusual commercial utility,” by which I have to assume he meant, “It’s cheaper and more compact because it gets one pump to do the job of two.”
Where I see a certain amount of ambiguity is in the question of whether biasing the front planetary unit to be in reduction with the transmission in neutral was intended to facilitate push-starting by increasing pump speed or whether that was just a side benefit. It would have been possible to arrange the hydraulic system so that the front servo would engage in neutral as soon as there was enough hydraulic pressure to do so (which I originally assumed, incorrectly, that it did), but they didn’t do that. It’s possible that was to give more pump pressure in a push- or tow-start, but it’s also possible that it was just something they skipped for some practical reason, such as development time constraints.
If the pump gears are driven off the main shaft, I don’t see any way the pump would be running at rest, and I assume the pump drive gears would turn backward in reverse. In the first instance, they may have decided it wasn’t that important; even with the dual-gear pump, the pump wouldn’t operate with the car at rest until the main clutch started to engage anyway. As for reverse, maybe they concluded that running the pump drive gears backward (without the weird opposed-gear issue the dual-gear pumps created) provided adequate pressure for lubrication in reverse. (I’m wondering if perhaps the pressure regulator was modified with that in mind.) That is quite a bit different than Hydra-Matic, or the later Powerglide and Dynaflow transmissions (which greatly increased line pressure in reverse to hold the reverse band engaged firmly), but with the Automatic Safety Transmission, the hydraulic pressure requirements in reverse were more modest.
Yes, that’s what I took from it, at any rate. As I said in my earlier comment, I think that probably the most important point for the general reader is that the Automatic Safety Transmission relies on the sliding gear in the forward-reverse unit to provide neutral; the transmission doesn’t provide any means of neutralizing the planetary gearsets other than simply cutting off the flow of power to them. (Thompson took pains in the patent disclosures to emphasize that the planetary gearing doesn’t allow freewheeling, which was a fairly common feature in that era, easing shifting at the cost of engine braking.)
I wouldn’t read too much into the ad images, and in any case, the point Oldsmobile was hoping to make was that shifting between L and H once in motion was very easy: no declutching required, and scarcely any need to remove either hand from the steering wheel. I definitely wouldn’t assume that would necessarily indicate how owners would use it in the real world.
I dug up some pertinent excerpts of the 1938 Buick service manual supplement for this transmission, and Buick did recommend starting in Low range rather than High, although the wording is notably less emphatic than the Oldsmobile handbook. Buick essentially said you COULD start in High if the engine was warm and the car was on a level surface, but that you should avoid doing so because it was hard on the rear clutch. Of course, a dealer service manual is not the same as an owner’s manual, and I don’t know how the Buick owner’s manual framed it, even to the extent anyone read the manual. However, if the service manual supplement is any indication, the difference between the Oldsmobile and Buick recommendations was essentially one of emphasis. For instance, if a customer asked, “I accidentally started in High, will that hurt my transmission?” a service manager who’d read the Buick supplement would probably say something like, “Don’t make a habit of it, but if every now and again you forget to move the lever back to Low, it probably won’t hurt anything.”
(I amended the revised paragraph in the text to better reflect that distinction. I had previously hedged my original statements because I was concerned that they weren’t adequately sourced, but I now think I was being overly cautious about it.)
I think there’s a case to be made that the argument is one of received wisdom rather than observation. For instance, some cars with torque tube drive used the early single-coupling Hydra-Matic transmissions — Nash, for instance — and didn’t seem to cause any great scandal with their shift quality. On the other hand, it was also a question of degree. There was a discussion a while back on the Curbside Classic website of some period road tests of mid-sixties Ramblers, which still had torque tube drive and suffered pretty badly for it with the more powerful engine options. With the sixes (which one may assume most Rambler customers bought), it was no big deal, but Motor Trend found that the greater torque of the AMC 327 really exposed the limitations of the drivetrain, making it distressingly easy to get the axle hopping unless you drove very gently. Obviously, most people don’t drive like they’re a magazine tester on a dragstrip, but these are the kinds of things that show up in manufacturer proving ground testing. So, while Buick engineers didn’t like the original Hydra-Matic and may thus have been inclined to exaggerate its flaws, I wouldn’t assume that those complaints were wholly without foundation.
The Buick service manual warns that starting in H “causes excessive slippage in rear transmission clutch,” so even Buick engineers agreed on that point.
I think those were probably the main reasons. On the second point, salespeople are not necessarily very technically inclined (I doubt many salespeople read the service manuals — not their department — and who knows what the Buick sales briefing books said?). Also, the Automatic Safety Transmission put them in an awkward position: how to get customers to agree to pay an extra $100 (not a small amount of money in 1938) for an “automatic” transmission that still had a clutch and they still had to shift. You can see why they might say, “Oh, look, you can just leave it in the second drive range most of the time and not worry about it.” Even if the owner checked the manual, that would be after they’d bought the car!
I find that one a bit of a stretch. The division was not happy about the whole project, but they were still at least temporarily stuck with it, and sabotaging it in that way was not in their interests or the interests of their dealers, especially since swapping the transmission for a conventional Synchro-Mesh unit was a major endeavor. Admittedly, people do sometimes do self-sabotaging things out of spite, but in this one, I’m inclined to point the finger more at overzealous salespeople and owners not reading the manual.
Not really. As much as Oldsmobile had played an active role in the development of Hydra-Matic, it was a corporate transmission, built by the new Detroit Transmission Division and offered to various internal and external customers, as were the later Dual-Range Hydra-Matic, dual-coupling, and single-coupling three-speed units. Detroit Transmission Division was renamed Hydra-Matic Division in October 1963, around the time TH400 production began. So, TH400 was not called “Hydra-Matic” for marketing reasons: It WAS a Hydra-Matic transmission, produced by the division of that name.
Yes, as I said the other day, improved electrical systems pretty clearly reduced the prevalence of push-starting or tow-starting for a stalled car.
The other consideration for transmission oil pumps, as I mentioned before, is that with a fluid coupling or torque converter, it’s possible to have a front pump that’s driven by the engine at all times, even at idle, which eliminates the “zero line pressure at rest” issue the Automatic Safety Transmission faced.
Aaron, regarding the 1939 AST, perhaps the single oil pump works because, in 1st and Reverse, there is no need for oil pressure to activate any clutch or band. Front and rear bands are held tight, providing the 1st gear ratio; so no oil needed. But lube would be nice, too. And in 1st gear it’s not a problem as soon as the car gets up any speed. It wouldn’t get any lube in Reverse, but it doesn’t operate there very long or under heavy loads. So they can get away with it.
This further suggests that the springs are definitely capable of holding the bands.
It’s not whether the springs are capable of holding the bands engaged (they certainly are, and the patent text notes that they’re kind of overpowered to avoid band slippage, since the bands are designed not to really be self-energizing), but whether there’s torque applied and thus reaction torque to lock the drum against the bands.
I thought about this earlier and decided a) you have a point and b) I’m over-complicating the most pertinent detail for the casual reader, which is the question, “If the front band is engaged and the rear band is released in neutral, does that mean neutral is really the same as 3rd gear?” The answer is, “Generally no, because in neutral, the front gearset only receives any torque to multiply under certain conditions.” I decided that it would be better to revise the asterisked footnote to just say that, viz.:
I’ll reply to the rest of your comment later; I received but haven’t yet had the opportunity to finish reading your email.