The Hydra-Matic, GM’s first fully automatic transmission, was a great success, inspiring a host of rivals — including some within General Motors itself. This week, we look at the origins of Dynaflow and Powerglide, the ambitious but ill-fated Turboglide and Triple Turbine, the later controlled coupling Hydra-Matic and Roto Hydra-Matic, and more.
TORQUE CONVERTER DRIVE
As we saw in our first installment, the original Hydra-Matic, introduced in late 1939, was the world’s first really successful fully automatic transmission. By 1952, General Motors’ Detroit Transmission Division had produced more than 2 million Hydra-Matics, which were used by Oldsmobile, Cadillac, Pontiac, and a variety of outside automakers, ranging from Kaiser-Frazer to Muntz. Hydra-Matic was standard on all Cadillacs by the early fifties and went into most Oldsmobiles and more than 80% of Pontiacs.
Notably absent from the list of Hydra-Matic users were GM’s other automotive divisions, Buick and Chevrolet. Instead, between 1948 and 1963, those divisions fielded no fewer than seven distinctly different automatic transmissions, none of them related to the original Hydra-Matic or its successors (which we’ll discuss in more detail later in this article). Moreover, Buick and Chevrolet did not use the same transmissions, although their respective designs were conceptually similar in many respects.
This curious divergence may perplex the modern reader accustomed to a world of corporate engines and transmissions, even at GM. At almost any other automaker, then or now, Hydra-Matic (in various light-, medium-, and heavy-duty versions) would have been the automatic transmission until being phased out in favor of something newer and/or better. Even more surprising is the fact that the original impetus Buick and Chevrolet’s unique automatic transmissions came not from the engineering staffs of those divisions (which in that era still enjoyed considerable autonomy), but rather from one of the principal architects of Hydra-Matic.
Engineer Oliver K. Kelley (often known as “O.K.” Kelley) began his career as an engineer at Cadillac in the late twenties and later worked for GM’s Yellow Truck and Coach Manufacturing subsidiary before joining Earl Thompson’s transmission development group, which by then had become part of the central Engineering Staff. Although Hydra-Matic was a team effort building on ideas Thompson had been developing since 1932, the three patents that most closely reflect the early production versions of the Hydra-Matic transmission are actually in Kelley’s name. When preproduction of the initial Model 180 Hydra-Matic began in mid-1939, Kelley was among the corporate engineers reassigned to Detroit Transmission Division (of which Kelley’s colleague William L. Carnegie became the first chief engineer) to oversee the transition from prototype to mass production.
We may presume, therefore, that Kelley was as familiar as anyone was with the original’s Hydra-Matic’s strengths and various limitations. As we’ve previously discussed, Hydra-Matic was very clever in many respects, but it was by no means a light, compact, or mechanically elegant design and it can’t have been cheap to manufacture. Furthermore, its operation was far from seamless even under the best of conditions, something that would earn the transmission considerable criticism in the years to come. There was obvious room for improvement.
Nonetheless, considering how much money GM had invested in the project, proposing, as Kelley and his colleague George R. Smith did in the summer of 1939, that the corporation begin working on another new and completely different automatic transmission was a bold suggestion indeed — particularly since at that point Hydra-Matic had not yet gone on sale. The most compelling point of Kelley and Smith’s argument, and the likely reason their proposal was not dismissed out of hand, was Hydra-Matic’s substantial production costs. While those might be acceptable for the senior divisions, which could pass the cost along to the customer, Hydra-Matic was expensive enough to be a dicey proposition for Chevrolet. Chevrolet owners were as weary as anyone of shifting gears (as evidenced by Chevrolet’s decision to make a vacuum-assisted shift linkage standard equipment for 1940), but whether the buyer of an $800 Chevy would be willing or able to spend $100 or more for a self-shifting transmission was another matter. The demand was there, but to tap it, Chevrolet would need an automatic transmission that could be priced to sell.
We don’t know what higher-level discussions Kelley and Smith’s proposal may have prompted, but the gist is not hard to guess. Even during the Depression, Chevrolet’s total sales volume had only once fallen below 400,000 units per year and 1939 sales had been closer to 600,000. If Chevrolet could offer an automatic affordable enough to achieve a take-up of 50% or better, that would mean more than a quarter of a million transmissions a year. Since very few American drivers liked to shift, offering such a transmission would also give buyers a compelling reason to choose Chevrolet over low-priced rivals, so Chevrolet might even stand to increase its market share. With numbers like that, developing an automatic transmission for Chevrolet was likely to be a worthwhile investment even if it didn’t share a single bolt with Hydra-Matic.
The upshot was that Kelley and Smith’s rather daring proposal eventually paid off. In the summer of 1940, as first-year production of Hydra-Matic was winding down, they were transferred to the Engineering Staff as part of a reorganized transmission research team (known in contemporary GM vernacular as a product study group). This worked out particularly well for Kelley. Not only was he once again doing advanced research work — which we have to assume was more interesting than production engineering — he was now leading the team, Earl Thompson having left General Motors about three months earlier.
The initial focus of Kelley’s new group was on torque converters. As Kelley was undoubtedly aware, some Yellow Truck & Coach buses had recently begun offering a Spicer torque converter transmission, a licensed derivative of the Lysholm-Smith unit developed by engineer Alf Lysholm of the Swedish firm Ljungstroms Angturbin AB. Over the previous decade, that transmission and others like it had become increasingly common for bus and railroad use, although to our knowledge, there had not yet been any production automotive applications.
Today, we’re accustomed to thinking of torque converters primarily as clutches, but a torque converter is also a type of infinitely variable transmission. The bus and rail-car torque converter transmissions of the thirties used the converter primarily as a transmission, sometimes adding a separate clutch to connect the converter to the engine; conventional reduction gears were typically used only for reverse. Such transmissions were capable of providing torque multiplication comparable to Hydra-Matic with no perceptible steps and no need for a complicated hydraulic control mechanism, making them a potentially attractive Hydra-Matic alternative for Chevrolet.
Before Kelley and company had had the time for more than preliminary research, however, outside circumstances shifted their attention to a very different application.
In June 1940, about two months before the establishment of Kelley’s new product study group, GM president William S. Knudsen had been summoned to Washington, D.C., where he was asked to oversee the ramp-up of American military production. By then, Europe had been at war for months, a growing number of European nations had fallen to the Nazis, and Great Britain’s position was looking increasingly precarious. Knudsen’s assignment was to enlist domestic industry in the accelerating U.S. rearmament effort.
Late that year, Kelley’s group was asked to shift their attention from a potential Chevrolet automatic to the development of a transmission that could take the place of the conventional gearboxes then used in most U.S. armored fighting vehicles (AFVs). The idea of automatic transmissions for tanks may sound faintly ridiculous, but what is merely annoying in a car — like the need to shift gears — can be positively hazardous for combat vehicles, particularly lightly armored ones. While Cadillac would shortly adopt Hydra-Matic for use in light tanks (mated, as we explained in Part 1 of this article, to Cadillac V-8 engines), Hydra-Matic had neither the torque multiplication nor the torque capacity needed for heavier AFVs.
Kelley and his team responded to this request by devising a heavy-duty semiautomatic torque converter transmission, which was subsequently produced by Allison (then a GM division) under the trade name Torqmatic. The original Torqmatic 900T AFV transmission combined a six-element torque converter (with a single impeller, three turbines, and two stators) with two hydraulically controlled planetary gearsets, providing three forward speeds and one reverse. The transmission still had to be shifted manually, but there was no need to de-clutch and little danger of missing a shift. Moreover, the torque converter alone provided a stall ratio of 4.8:1, so a useful amount of torque multiplication was available even in the direct-drive third gear.
This transmission was selected for the Buick-developed T-70 tank destroyer, which entered service in 1943 as the M18 Hellcat. The 900T helped to keep the M18’s nine-cylinder air-cooled Continental radial engine within its narrow power band all the way up to the Hellcat’s 50+ mph (80+ km/h) top speed and had the torque capacity to withstand the 972 cu. in. (15,972 cc) engine’s monstrous 940 lb-ft (1,275 N-m) net torque output, which would have made an oily metal milkshake of the Hydra-Matic’s innards. The transmission performed well in the M18 and later in the derivative M39 armored utility vehicle and the M26 Pershing medium tank, both introduced in 1944.
It was obvious early on that the torque converter transmission would also be well-suited to heavy civilian vehicles and equipment. After the war, Allison developed Torqmatic into an extensive and long-running line of heavy-duty torque converter transmissions for different military, commercial, and industrial applications, including trucks, buses, and heavy machinery. (Today, Torqmatic remains a registered trademark of Allison Transmission, which is no longer owned by General Motors.)
While military work remained the top priority for Kelley and his team (and the auto industry in general) until late in the war, they had not forgotten their original objective. In late 1944, Kelley filed a patent application (U.S. Patent No. 2,606,460) for an automotive torque converter transmission combining a simple planetary gearset with a novel five-element torque converter featuring dual stators and dual impellers, one driven directly by the engine and the other connected to the first by an overrunning clutch. The blades of each impeller were of different pitch; essentially, the primary impeller was optimized for near-stall conditions (i.e., during the period of torque multiplication) while the secondary impeller was optimized for cruising speeds. An overrunning clutch allowed the secondary impeller to spin freely until turbine speed approached engine speed, after which both impellers would turn together.
By mid-1945, Kelley’s group had installed working prototypes of this transmission in several test mules. Although the torque converter transmission had been conceived with Chevrolet in mind, Kelley also showed off the prototypes to engineers at the other automotive divisions to see if any of them were interested in the new design as a potential alternative to Hydra-Matic.
The strongest interest came from Buick chief engineer Charles A. Chayne. Buick had been very resistant to the earlier Automatic Safety Transmission and had declined to adopt Hydra-Matic, which Chayne had caustically dubbed “Hydra-Jerk.” Some of Buick’s antipathy toward Hydra-Matic was probably attributable to divisional pride; some years earlier, the division had spent a lot of time and money on the abortive “Roller,” an infinitely variable friction drive transmission that Buick general manager Harlow Curtice had been obliged to cancel back in 1934. Nonetheless, the concerns about harshness were probably not without merit. Unlike Oldsmobile, Cadillac, and Pontiac, which used open driveshafts with universal joints at both ends, Buick (and Chevrolet) in those days used torque tube drive, which combined the enclosed driveshaft and rear axle into a single rigid assembly connected to the transmission via a single U-joint. Since the primary purpose of the torque tube was to transmit drive torque, the massive axle assembly would have amplified each of Hydra-Matic’s firm shifts into an uncouth thump, hardly in keeping with Buick’s upscale image.
Chayne and Curtice both sampled the prototype torque converter automatic and found it much more to their liking. It was mechanically straightforward and offered seamless if rather stately acceleration. Chayne subsequently assigned Buick’s own engineers to collaborate with Kelley’s team on the development of a production version of the torque converter transmission for Buick.
After much development and extensive testing, the new transmission, which Buick christened Dynaflow, was finally announced in January 1948. It went on sale in March as an option for the top-of-the-line Buick Roadmaster. While Dynaflow was mechanically simpler and somewhat lighter than Hydra-Matic, the Buick transmission was no cheaper — list price was $206 (more than $1,800 in 2010 dollars), some $20–$30 more than the contemporary Hydra-Matic.
Dynaflow retained the five-element, dual-impeller torque converter of the early prototypes, but the planetary transmission adopted what today is commonly known as a Ravigneaux gearset (after French inventor Pol Ravigneaux, who patented several variations of the layout in the thirties and forties). This comprised two sets of planetary gears sharing a common planet gear and a single planet carrier, which drove the output shaft. Hydraulic servos controlled a multi-disc clutch and two brake bands, which could be alternately engaged to provide high (direct drive), low, and reverse; the latter two shared the same 1.82:1 ratio. There was also a parking pawl to lock the driveshaft.
The original Dynaflow is often described as a two-speed automatic, but selecting Drive engaged the planetary transmission’s direct drive clutch and the only automatic shifting the transmission provided was via the torque converter. The driver could manually select Low, which would disengage the direct drive clutch and engage the low band, but shifting back to direct drive required moving the selector back to Drive. This was by design; while it would not have been difficult to engineer the transmission to start in low and shift automatically to and from high, Kelley and his team wanted normal operation to be as ‘stepless’ as possible.
The tradeoff was performance. While non-automotive torque converters often had stall ratios of 4:1 or more, those transmissions were designed for heavy-duty engines that spent much of their operating lives at or near full throttle. For Dynaflow, using a higher stall ratio would have had several undesirable consequences, including a higher stall speed and more fuel-wasting slippage at part-throttle cruising speeds. To avoid that, the original Dynaflow had a stall ratio of only 2.25:1, which was taller (numerically lower) than second gear of a contemporary Hydra-Matic and significantly taller than the 2.67:1 low gear of Buick’s standard three-speed manual transmission.
In partial compensation, Dynaflow-equipped Roadmasters used a special high-compression version of Buick’s 320 cu. in. (5,247 cc) straight eight that added an extra 6 hp (4 kW) and 4 lb-ft (5 N-m) of torque. Even so, Dynaflow’s off-the-line response was lethargic — albeit very smooth — unless you manually selected Low, which could be done at any speed up to about 45 mph (72 km/h). Unfortunately, frequent manual gear changes exacerbated the already heavy fuel consumption and would eventually take their toll on the transmission’s low band and direct drive clutch, which were intended for only occasional use. Buick cautiously described the lower gear as “emergency low.”
Despite those shortcomings, Dynaflow was well-suited to the character of postwar Buicks, which emphasized unhurried plushness over performance or road manners. The average Buick buyer of the time was not terribly concerned with fuel economy and welcomed Dynaflow’s lazy smoothness. It was too bad that Buick no longer offered formal cars; Dynaflow lent itself admirably to a processional pace.
Although Buick had beaten Chevrolet to the punch, GM’s largest automotive division had also evaluated the corporate engineers’ torque converter transmission and had been working since 1946 on Chevrolet’s own production version. Dubbed Powerglide, this finally debuted as a $159 option for 1950 Chevrolet DeLuxe models.
In its original form, Powerglide was much like the early Dynaflow — not surprising considering that both were production derivatives of the same basic design. Both transmissions used a two-speed Ravigneaux gearset providing 1.82:1 low and reverse ratios and both had a five-element torque converter with dual stators and dual impellers, although Powerglide’s stall ratio was slightly lower, at 2.20:1. Powerglide’s principal novelty, developed and patented by Kelley and William S. Wolfram (U.S. Patent No. 2,651,918), was an unusual auxiliary fluid coupling incorporated within the torque converter and sharing the same oil supply. The auxiliary coupling’s “impeller” was actually an additional set of vanes mounted around the inner circumference of the converter’s turbine torus while the “turbine” vanes were mounted around the inner circumference of the primary impeller torus. Each set of auxiliary vanes was curved in more less the opposite direction of the curvature of the primary blades.
The auxiliary coupling was intended to address a minor but disconcerting flaw of most fluid clutches: a lack of engine braking when coasting with the throttle closed, as when descending a hill. Under those conditions (technically known as overrun), the momentum of a Powerglide-equipped car would turn the converter turbine, whose auxiliary vanes would transmit that motion to the auxiliary vanes on the primary impeller and thus provide an engine braking effect. The auxiliary coupling also allowed the car to be push-started at speeds as low as 12 mph (19 km/h). The downside was a bit of additional drag within the converter during normal acceleration.
In other respects, Powerglide operated just like Dynaflow and suffered the same limitations. The performance penalty was even more pronounced with the Chevrolet six than with Buick’s big straight eight, particularly since Powerglide included a taller 3.55:1 axle ratio, compared to 4.11 for Chevrolets with manual shift. Despite the bigger, more powerful engine that was standard with Powerglide, Chevrolets with automatic were more than five seconds slower to 60 mph (97 km/h) than standard-shift cars and returned less-than-frugal fuel economy.
Although these limitations did little to dampen buyer enthusiasm, Chevrolet quickly moved to address Powerglide’s performance with an extensive redesign of the transmission, introduced for the 1953 model year. The revised transmission had the same internal ratios as before, but was redesigned to start in low rather than high. The hydraulic system was revised to provide automatic upshifts and downshifts at speeds up to 42 mph (68 km/h) and the low band and clutch pack were beefed up accordingly. Meanwhile, the five-element torque converter and auxiliary coupling were discarded in favor of a simpler (and undoubtedly cheaper) three-element unit providing the same stall ratio.
This revised arrangement was a compromise of Kelley’s original vision, but it was much better suited to Chevrolet’s needs and worked well enough for most buyers. Chevrolet would continue to offer the two-speed Powerglide (with several subsequent improvements) into the seventies.