KENICHI YAMAMOTO AND THE CHATTER MARKS FROM HELL: THE MAZDA ROTARY ENGINE
As elegant and straightforward as it seemed on paper, Toyo Kogyo engineers quickly discovered that the rotary engine had many serious problems. At the time the license agreement was approved, even NSU had yet to build a truly production-ready engine, and the early single-rotor prototypes suffered a very rough idle and prodigious oil consumption. Cooling was also problematic and the gap in the water jacket around each spark plug housing produced thermal stresses that would eventually crack the rotor housing.
The biggest problems, however, were with the complex apex and corner seals at each rotor tip. Those seals were responsible for maintaining compression and segregating exhaust gases from the intake charge, but they experienced considerable stress from the combustion process, limiting their useful life. When the seals wore out, the engine suffered dramatic power loss. Few of the early tip seals had anything approaching an acceptable lifespan, usually failing after less than 200 hours of operation. Harder materials lasted longer, but exacerbated another problem: the tendency of the apex seals to leave vicious “chatter marks” on the inner surface of the rotor housing. During the first two years of development, Toyo Kogyo reportedly scrapped hundreds if not thousands of ruined test engines.
Although the Japanese engineers were progressing more quickly than their NSU counterparts were in some areas, the development process was undoubtedly expensive, and Toyo Kogyo might well have given up had it not been for the determination of Tsuneji Matsuda. The story among company employees was that the normally implacable Matsuda had actually kowtowed before Toyo Kogyo’s principal financiers while pleading for the resources to continue the rotary project. Whether or not that was true, Matsuda made it clear that he considered the rotary the key to the company’s survival — a commitment that eventually won over even one of the engine’s harshest internal critics, engineer Kenichi Yamamoto.
Yamamoto, born in Kumamoto in September 1922, had received a degree in mechanical engineering in 1944 from the prestigious Imperial University in Tokyo. Conscripted after graduation, he had been commissioned as a naval lieutenant and sent to work with Kawanishi Aircraft at the firm’s naval works in Tsuchiura, where he was eventually assigned to work on kamikaze aircraft. When the war ended, jobs for skilled engineers were scarce, so Yamamoto ended up as an assembly line worker at the Mazda plant in Hiroshima, which had only recently reopened after being damaged in the American atomic attack. Yamamoto’s articulateness and technical drawing skills (which he had continued to practice) did not go unnoticed, however, so a few years later he was transferred him to engine development, working on the design of Toyo Kogyo’s first OHV engine.
By his own account, Yamamoto was not pleased when the company licensed the rotary engine in 1961. He considered the rotary conceptually sound, but he was all too aware of the many pitfalls facing any new engine design and saw the whole project as a boondoggle and a waste of resources. Given those doubts, one can imagine his reaction when he learned in April 1963 that he had been reassigned to lead the new Rotary Engine Research Department.
Yamamoto’s first six months in his new job did little to assuage his doubts. Despite the dedication of his hand-picked engineering team, known internally as the “47 Ronin” (in reference to the famous group of masterless samurai whose quest to avenge their fallen lord in the early 1700s is popularly regarded in Japan as the epitome of honor, loyalty, and duty), the rotary engine’s major problems seemed intractable — particular the chatter marks, whose cause was initially elusive. Nonetheless, Yamamoto resolved to give it his best effort, particularly after he heard Matsuda deliver a speech to Toyo Kogyo suppliers that June, outlining the existential threat posed by MITI’s consolidation plans.
By the time Toyo Kogyo exhibited prototypes of the new rotary at the Tokyo Auto Show in October, Yamamoto had become so frustrated and discouraged that he told Matsuda he wanted to resign. Matsuda persuaded him to stay by appealing not only to Yamamoto’s company loyalty, but also to the memory of the siblings they had both lost in the bombing of Hiroshima.
Matsuda rewarded Yamamoto’s perseverance with an infusion of new resources. In 1964, Toyo Kogyo set up a state-of-the-art rotary engine lab with 30 test cells and computers to process the test data, still a novelty in the mid-sixties. Over the next three years, the company would quadruple the size of its rotary engineering staff.
Gradually, Yamamoto and his team came to grips with the rotary’s major flaws. The chatter marks were eventually traced to the apex seals hitting their resonant frequency within the engine’s operating range, which was addressed with changes to the seal design and materials. The apex seals of Mazda’s early production rotaries were self-lubricating pyrographite, impregnated with aluminum for greater strength, which eliminated the chatter marks and provided a useful life of at least 60,000 miles (100,000 km). Meanwhile, better oil seals, developed in partnership with Nippon Oil Seal Co. and Nippon Piston Ring Co., finally reduced oil consumption to a manageable level. By 1967, Toyo Kogyo was finally ready to launch its first rotary-engine production car.
MAZDA COSMO SPORT
To showcase its new engine, Toyo Kogyo decided to develop an entirely new car not based on any existing model. Known internally as Project L402A and later christened Mazda Cosmo Sport, it was the first Mazda sports car: a low-slung monocoque coupe with a very low hood line that took full advantage of the rotary’s compact dimensions.
Although Toyo Kogyo had gone to Italy for some past design work, the Cosmo Sport was styled in-house, looking rather like the bonsai offspring of a 1961 Ford Thunderbird and the 1963 Chrysler turbine car. Unlike Mazda’s early kei cars, the Cosmo Sport had a front engine and rear-wheel drive. Front disc brakes were standard and the sole transmission was a four-speed manual gearbox.
The earliest Cosmo Sport prototypes had a two-rotor engine known as the L8A, with a total swept volume of 798 cc (49 cu. in.). (Unlike NSU, Toyo Kogyo engineers had largely abandoned single-rotor engines, concluding that multiple rotors provided better low-end torque and idle quality. Mazda would briefly explore a return to the single-rotor concept in the 1970s in search of greater fuel economy, although they never offered a single-rotor engine in any production car.) To improve low-speed performance, the L8A had two spark plugs for each rotor, one firing 5 degrees after the other. To manage the separate advance curves for the twin plugs, there were two complete ignition systems, including twin distributors.
The first two running prototypes of the new car were finished by October 1963, and Matsuda actually drove one to the Tokyo Auto Show later that month. However, Toyo Kogyo displayed only the engines at that show, delaying the Cosmo Sport’s public debut until September 1964. According to some accounts, Matsuda opted to wait as a show of respect to NSU, whose first Wankel Spider had debuted in Frankfurt only a few weeks before the 1963 Tokyo show. According to others, NSU pressured Toyo Kogyo to delay the launch and discouraged plans to show the Cosmo Sport overseas, lest it steal the Spider’s thunder.
Initially, the L8A had peripheral exhaust ports and a combination of side and peripheral intake ports, which linked to the primary and secondary barrels of the standard four-barrel carburetor. While the additional peripheral intake ports improved high-end power, Yamamoto’s team found them detrimental to low-speed response and idle quality and finally decided to delete them, leaving only the side intakes. That change left the L8A somewhat underpowered, so the engineers increased the rotor diameter, raising total swept volume to 982 cc (60 cu. in.). With a single Zenith-Hitachi four-barrel carburetor, the revised L10A engine was rated at 110 PS (108 hp, 81 kW) at 7,000 rpm, with a maximum of 96 lb-ft (130 N-m) of torque at 3,500 rpm.
Toyo Kogyo built about 60 preproduction cars for evaluation in 1965 and 1966, but the Cosmo Sport didn’t actually go on sale until May 30, 1967. It was not only the first production Mazda with a rotary engine; it was the world’s first two-rotor production car, debuting more than four months before NSU’s Ro80 sedan. (Curtiss-Wright had previously tested its two-rotor RC2-60 U5 engine in a modified Ford Mustang, but that engine was never offered for public sale.) Since Toyo Kogyo had only recently revised its license agreement to allow sales of the rotary engine outside Japan, the Cosmo Sport was initially offered only in the home market, with a starting price of ¥1,480,000 (around $4,100). Only a few cars ended up overseas, many of them purchased by other automakers or rotary licensees like Curtiss-Wright, who were eager to figure out what made the Cosmo tick.
The curious foreign journalists who had the opportunity to test the Cosmo Sport were mostly impressed. It handled well, with quick steering and basically neutral balance, but the real star was the engine. The L10A was not especially quiet (although some reviewers found its exhaust note quite charming), but it was exceptionally smooth and would rev to 8,000 rpm with an alacrity and enthusiasm alien to most contemporary reciprocating engines. Low-end torque was not abundant, but as engine speeds increased, performance brightened considerably. Reaching 60 mph (97 km/h) took less than 9 seconds and advertised top speed was 115 mph (185 km/h), impressive for a small sports car of the era, and faster than many V8-powered American sedans.
The Cosmo Sport was not sold in large numbers — only 343 were built between May 1967 and September 1968 — nor was it intended to be. Its construction involved a great deal of hand labor, and it’s hard to imagine Toyo Kogyo made any money on it. If the little Mazda coupe was not a profitable exercise, however, it was an effective proof of concept that drew attention from around the world, including many markets the company had yet to enter.
Even more attention came in August 1968, when Toyo Kogyo entered two Cosmo Sports in the Marathon de la Route endurance race at the Nürburgring. The Nürburgring cars had various engine modifications, including the restoration of the L8A’s auxiliary peripheral intake ports, but were otherwise close to stock. One car was felled by a broken axle during the race, but the other took fourth place, the first flush of a long and often illustrious competition career for Mazda rotaries.
27 CommentsAdd a Comment
I’ve always loved Mazda’s rotary cars. Fantastic article, and I can’t wait for part 2…
Great story looking forward to part 2. A friend in Tasmania had several of those bertone Luces nice cars the later models had the 1800 Capella engine.
Thanks for the Mazda rotary article. I’m looking forward to reading Part 2. Despite growing up around Mazda rotary-powered cars, I learned quite a bit!
It’s a real shame that no one can seem to lick the engine’s fuel and oil consumption problems. I have heard some discussion of Mazda using rotaries in hybrids, which makes some sense to me. Rotaries are so small and, on paper at least, elegantly designed.
Man, that Luce coupe is a looker.
I don’t know about hybrids, but Mazda has done quite a bit of development on a hydrogen-fueled rotary, which has been offered on a limited basis for fleet sales in some markets.
If the next-generation 16X engine materializes, Mazda is hoping to reduce fuel consumption substantially, in part by adopting direct injection. Still, since piston engines keep improving in that regard, as well, I don’t know that the rotary will ever match the reciprocating engine in specific fuel consumption. Some things can be mitigated (like wall quench), but other factors, like the combustion chamber surface area to volume ratio, are sort of the nature of the beast.
The Luce R130 is indeed a very nice-looking car. I’d never seen one before I started researching this story.
Very interesting article, well, as usual, Aaron!
The topic was somewhat forgotten in France after Citroën heavily invested in the technology, eventually failed to make it work and had to drop the project in the early 70’s. They had been so serious about it that the models developed in the late 60’s, the GS and the XM, were designed for a rotary. They had to hastily develop a reciprocating engine for the GS and make it fit in the engine bay that was not large enough.
The XM eventually was painfully fitted with a Peugeot engine.
Anyway Citroën was never able to design a good engine. This huge investment and its failure played an important role in the demise of the company.
“They had been so serious about it that the models developed in the late 60’s, the GS and the XM, were designed for a rotary.”
You mean the SM, don’t you?
I believe Nicolas was probably referring to the CX, which replaced the Citroën DS in 1974. I’ve never heard anything about the SM being intended for rotary power — of course the production cars had the Maserati V6 — but I think the CX was. The XM was the CX’s eventual successor, introduced in the late eighties.
Right Aaron, my pen slipped, it was the CX.
The XM was its successor.
The SM, stangely enough, was fitted with the (in)famous Maserati V6 even though Citroën had such a faith in the future of the rotary as the ultimate replacement of the reciprocating.
Timing may have had something to do with it. Citroën didn’t build the first M35 single-rotor cars until the fall of 1969, and as I understand it, they were essentially evaluation models, not yet intended for large-scale production. The BiRotor wasn’t introduced until 1974, about four years after the SM debuted. Even if Citroën were keen to give the SM rotary power, it probably wouldn’t have been ready until a few years after launch, even in a best-case scenario.
If things had worked out differently, I imagine Citroën might have added a rotary engine to the SM later, perhaps in a second-generation version for the mid-seventies. Of course, even if the Comotor engines had been more successful, the SM was not, and might have been dropped without ever getting a rotary engine.
For them the rotary was the future type of engine for all applications, just as well as they were persuaded they had a market for the SM.
With NSU, Mazda and others working on it it’s understandable.
Your article is very interesting by showing how Mazda made a success of it, or at least could partly make a living with it, well… that’s a success, isn’t it?
Strangely enough it didn’t catch on as an aviation engine either.
[quote=Administrator] Citroën didn’t build the first M35 single-rotor cars until the fall of 1969, and as I understand it, they were essentially evaluation models, not yet intended for large-scale production. The BiRotor wasn’t introduced until 1974, about four years after the SM debuted.[/quote]
Starting in 69 a limited number M35, and in 73 GS Birotor, were sold to selected, faithful (and masochist) clients but the engine proved such a burden to maintain that Citroën offered to buy them back and scraped them. A few people only turned down the offer. The maintenance contracts were canceled for them. The few models still in existence are now very expensive collectors’ items, the day dream of all the GS enthusiasts.
So there was actually a future for the rotary! ;-) As usual the car that nobody wanted became the car that nobody can afford.
The source I was looking at (John Hege’s The Wankel Rotary Engine: A History) suggests that Citroën had basically intended to buy back the early evaluation engines from the outset, which would make a lot of sense.
I don’t know about France, but in the U.S., automakers are legally obligated to provide parts support for production models for a specific period of time, typically 15 years — obviously not an appealing prospect for cars or engines that don’t end up in mass production! For that and other reasons, some automakers have tended to offer such evaluation vehicles only as a closed-end lease or other type of loan-out, with no option to actually purchase and keep the vehicle at the end; I assume that not actually selling it avoids triggering certain legal requirements.
The Europeans have basically the same obligations as the Americans. As far as I understood, the deal was under specific conditions and since Citroën offered to buy them back it could cancel any support for those who rejecter the offer. It’s stupid it didn’t keep one example for history.
Mazda is the only one who succeeded with a rotary over the years while all the others flopped.
This is an interesting article as usual, I’m waiting for the second part. While you’re at it, how about an article covering GM’s attempt to build a rotary engine?
I thought about it, but in researching this article, I’m finding that detailed information about its development seems to be surprisingly scarce. While the development of the NSU, Mazda, and Curtiss-Wright engines is pretty well-documented, GM played it very close to the vest. To really do it justice would probably require talking to some of the engineers who worked on it, assuming that the people involved are still living, and willing (and able) to talk about the program.
No need to mourn it’s passing. A technological dead end. I don’t miss the
ffffttttt exhaust “note” of them at all.
Used to be a few about Brisbane, Delighted to see and hear that rust and enlightenment of the owners has made them almost extinct.
Good riddence. So it could rev to 5 digits.
Wow, FANTASTIC article! Thanks for the great piece on Mazda, the detail and depths you go to are above and beyond. One of the best history-of-automaker stories that I’ve read. Thanks again!
Another great article Aaron. Really appreciating your narrative drive and level of scholarship. I’m starting to believe the R100/1200 body was designed by Bertone as well, but can’t verify. Do you know of any text that addresses the connections between the Italian design houses and the Japanese manufacturers in depth?
I so far haven’t found anything to suggest one way or another whether the first-generation Familia was done by Bertone, although it’s certainly plausible given that Bertone did the first Luce and the Luce Rotary Coupé in that period. Even if Stilo Bertone didn’t do the Familia or the first Capella, those designs have a definite Italian flavor, much more so than subsequent products of Toyo Kogyo’s in-house design studio, which feel more typically mid-seventies Japanese.
I really like that little sidebar referring how to calculate the Wankel’s full displacement. I know Japan has different regulations than the U.S. and that Mazda had no choice to only count one chamber for each rotor (Geometric Displacement) due to extra taxes being placed on “bigger” cars. Either way, I really hope Mazda brings their Wankel rotaries back to the streets, because that awesome RX-Vision concept needs to be on the roads
In retrospect could the prospects of the Rotary have been slightly improved to a certain extent had Mazda and not Citroen established the Comobil later Comotor joint-ventures with NSU, where the development of the Rotary follows a more developed Mazda like trajectory instead of the engine being prematurely released as was the case with NSU and Citroen?
Would that have been enough had it been feasible to largely butterfly away the poor reputation and resolve the issues of the early Rotary engines or would more changes have been required? Taking into consideration of course the current disadvantages of the Rotary would still remain.
This is one of those “could / would /should” questions. The “could” part is straightforward, at least technologically speaking; Mazda obviously managed to keep the rotary at least reasonably viable for many years, so there’s no technical reason the fruits of their efforts couldn’t have been shared across a broader consortium.
The “would” and “should” parts are more complicated. NSU-Wankel patent licensing agreements generally included technology-sharing provisions because NSU did not have deep pockets and figured that pooling research data would be the best way to advance the art. (Part of the reason GM’s patent license was so expensive is that GM didn’t want to participate.) Toyo Kogyo participated in that and probably contributed quite a bit to it. However, the way that knowledge was applied ended up being dictated by other factors, including a maze of different licensing agreements (like the one with Curtiss-Wright that affected distribution in North America), the concessions NSU minority stockholders wrested during the Audi merger, and the fact that the Comobil/Comotor project contributed to Citroën’s financial over-extension and eventual bankruptcy.
Would NSU stockholders have agreed to set up such a consortium or joint venture with Toyo Kogyo? Possibly, and in that area, the progress Mazda had made with the technology might have been attractive. Would Volkswagen have been amenable to it after the NSU-Audi merger? Harder to say; I don’t imagine that Volkswagen or Audi were terribly keen on Comotor given the NSU shareholder concessions, and Toyo Kogyo being involved in that venture rather than Citroën (or succeeding Citroën) would not have changed that part of the equation. The minority shareholder deal ended up signing away most of the opportunity to profit from the rotary, and Volkswagen still wouldn’t have been in a position to use the rotary in its own products. (I don’t think Toyo Kogyo would have been in a position (financial or political) to acquire NSU instead of Volkswagen.) The main point of change in that scenario is that Toyo Kogyo by then had sunk enough money into development that they were more reluctant to simply shelve it.
Should they have? It depends on a lot of things. If Toyo Kogyo had stepped in AFTER Citroën had been forced to bow out, the venture would have had to rebrand and might still have had a shaky image. There were customers who would have been interesting (AMC, to name one), but Mazda rotary combustion engines of the early seventies were still not without flaw, and the issues might have caused their licensees to jump ship early. (Toyo Kogyo commitment to the rotary had a lot to do with pride and the sunk-cost fallacy, to be honest.) Also, the OPEC embargo would still have really pressed the rotary’s limitations in terms of fuel consumption, which would have still hurt the project’s commercial prospects.
So, an interesting idea, but I tend to see it coming to a similar end.
Do not envision a joint-venture between NSU and Mazda in place of NSU and Citroen significantly changing things once Volkswagen acquires NSU and merges it with DKW/Auto Union to form Audi, what with the fuel crisis hurting the rotary engine’s projects in Europe. That said the NSU Ro80’s issues would have been largely butterflied away for one thing, though not sure if it would be enough for Volkswagen to actually consider a rotary model for say the 914 (as was experimented in real-life) if not a small mid-engined sportscar or some other vehicle (.
For Citroen there is one less contributor to its financial over-extension and eventual bankruptcy, leaving only one or few more elements (e.g. Project F, possibly acquiring Maserati for V6, SM instead of DS Sport, etc) to be remedied on its end without being involved in the Comobil/Comotor project. What Citroen goes from hereon is another matter, in the absence of the French government forcing Peugeot to acquire Citroen perhaps Citroen ends up increasingly entangled with Fiat or collaborates with one and more carmakers on a few joint-ventures outside of Fiat (e.g. Alfa Romeo? Subaru? both? etc).
Essentially the rotary engine in this scenario would be better regarded compared to real-life thanks to Mazda’s early tie up with NSU and a thorough development programme (instead of being prematurely released), yet ultimately undermined by the fuel crisis and only really considered worthwhile for sportscars.
The issue pertaining to the NSU-Auto Union merger (which is explained in more detail in the Ro80 article) is that Volkswagen ended up signing away the lion’s share of Wankel-related profits and agreeing that if it used the rotary in its own products, it would have to pay license fees like any outside customer. In the short term, that was a big win for NSU minority shareholders, but it probably contributed a lot to the eventual stagnation of rotary development outside Toyo Kogyo/Mazda. On the other hand, NSU’s survival WITHOUT a merger would have been very tenuous because they had also overextended their resources with the Ro80 and what became the Volkswagen K70. Whether Toyo Kogyo would or could have afforded a merger with NSU that would have obviated the need for the Volkswagen deal I don’t know; it would have been a politically complex situation, to say the least.
NSU was undoubtedly aware of everything Toyo Kogyo had been doing regarding rotary development, which to my understanding was a condition of the original license agreement. (I assume NSU would still have had to formally license technology subject to Toyo Kogyo patents, although I don’t think that would have been an insurmountable obstacle.) However, the problem they both faced is that each was exploring different solutions to the challenges involved (like the apex seal issue), and it wasn’t yet apparent what would work best. Toyo Kogyo didn’t attempt anything quite as daring as the NSU floating seal design (which was a brilliant idea undermined by inadequate development testing), but the differences between the earliest 12A twin-dizzy engines and the better-developed 13B found in the first-generation RX-7 are pretty substantial. It’s not that Toyo Kogyo was smarter or luckier than NSU in this regard; it’s that they kept working on it and refining their approach to factors like sealing and porting.
However, the upshot I assume you’re getting at is that an alliance between NSU and Toyo Kogyo might have resulted in a more lasting commitment to developing the rotary engine as a commercial prospect, with a Comobil/Comotor-style entity offering engines to other companies that were interested in the technology, but either couldn’t afford or didn’t want to spend the money on developing their own. The actual reason that didn’t happen was probably mostly that Volkswagen had very limited financial incentive to bother and Citroën, as mentioned, ran out of money. If Volkswagen were not in the picture (or the settlement with the minority stockholders had turned out differently), NSU survived on its own, and Toyo Kogyo took Citroën’s place as development partner, it might be plausible.
One other fly in the ointment with that scenario, though, is European hostility toward Japanese automakers. As much as the emergence of Japan as a major player on the automotive scene aroused horrendous racism and nationalistic furor in the U.S., that hasn’t significantly dissuaded Americans from buying Japanese cars, to the point that domestic automakers have more or less abandoned many segments of the market to the Japanese and Koreans. European markets have not been nearly so amenable, and even products Japanese automakers have designed specifically for Continental or British tastes have often been commercial duds. Applying that chauvinism to the early seventies, it’s also not hard to envision a scenario where Toyo Kogyo partnership in a Comotor-type JV ends up leading European punters and pundits deciding that rotary engines are too Japanese, which combined with the pressures of the OPEC embargo might also have been a death knell.
Volkswagen could have attempted to use the rotary in more niche segments like Mazda did with front-engined sportscars though mid-engined in Volkswagen’s case, otherwise Volkswagen could be a passive beneficiary at best upon buying NSU.
As it says in the Ro80 article, there was a lot of enthusiasm within Audi-NSU for the planned Ro80 successor, including some fairly serious talk of installing its 1.5-liter KKM 871/EA871 engine in the Audi C2. However, the minority shareholder settlement would have made using that or other Wankel engines in Volkswagen models fairly costly, which I think was a big part of why the whole plan ultimately didn’t go anywhere.