IDEAS ARE COMMON IN ANY BUSINESS, but new ones are always rare. Good new ones are hen’s teeth. In 1933, a German engineer named Felix Wankel patented a new type of engine. It was small, relatively efficient, and ferociously smooth. It featured a triangular piston, or rotor, spinning eccentrically around a central shaft, like a hula hoop around a waist. Unlike a traditional piston engine, there was no valvetrain or reciprocating motion, which meant little mechanical noise or vibration. There was just glassy power, and from a handful of moving parts, which meant little to go wrong.
This story originally appeared in the September, 2017 issue of Road & Track - Ed.
In 1959, after a bit of development, Wankel’s engine was shown to the public. The world went ape. Manufacturers from General Motors to Daimler-Benz and Toyota sought rights to build the technology. Thirty-four of the companies that applied for those rights were from Japan, but in 1961, the sole Japanese license was awarded to Toyo Kogyo Kaisha Limited—a Hiroshima motorcycle and machine-tool company originally established to manufacture cork. The firm wanted to try building world-class cars, and its management thought a distinctive technology might help carve out a place in the industry.
Toyo Kogyo was later renamed Mazda Motor Corporation. Wankel’s engine came to be known colloquially as a rotary, and every other carmaker eventually threw in the towel on its development. For road use, the design offers inherent engineering challenges, from fuel thirst to huge oil consumption and low torque. But Mazda stuck with it. Fifty years ago, the company built its first rotary road car. It built rotary pickups. It took rotaries racing, where they sang: both literally, because the things wail when uncorked, and figuratively, because racetracks love a small and powerful anything.
Mazda North American Operations runs a living car museum called the Heritage Collection. With its help, we went to Mazda Raceway Laguna Seca and climbed into two bonkers race cars from the rotary’s peak. Each was built in the 1990s but developed in the late 1980s, a time of unique possibility. Together, they’re a testament to Mazda, Wankel himself, and the remarkable opportunity of a singular era.
1990 MAZDA RX-7 IMSA GTO
TO KNOW MAZDA IS TO KNOW THE RX-7. It is the only Wankel-engine car to have made a dent in the consciousness of the general public; more important, it forged Mazda’s reputation for sports cars. The name stood for Rotary Experiment 7—a hint, perhaps, that even Mazda knew the Wankel question was open-ended. Experiments are begun knowing they might not work.
Only this one did. The first RX-7 was introduced in 1978. It was small, rear-drive, and powered by a beer keg of an engine called the 12A. A 12A is two rotors, a displacement just under 1200 cc, a 7000-rpm redline, and three primary moving parts. RX-7s famously had a cockpit warning buzzer near redline, because the engine was so smooth and quiet, engineers worried people would forget to shift.
The RX-7 was redesigned twice, staying in production until 2002 (U.S. sales ended in 1995). Early on, Mazda gave the car an engine called 13B—wider rotors than the 12A, which meant more displacement. That particular beer keg was eventually turbocharged, spitting out lunatic power for something so compact and light. In naturally aspirated form, it powered nearly half of the 800,000- RX-7s built, and it helped the car win races, from SCCA rallies to an overall first at the 1981 24 Hours of Spa. It was also durable as hell. In an era when you could kill a car in an endurance race by simply driving too fast for too long, the Mazda team that won Spa was told to beat the snot out of the engine and pray for everyone else to break.
In 1990, Mazda entered the second-gen RX-7 in IMSA GTO racing. GTO meant Grand Touring, Over 2.5 Liters. The 1.3-liter 13B was made larger to suit, by adding rotors. GTO rules made this legal, so long as the Frankensteinian product had roots in showroom bits.
The monster that resulted was called a 13J. It was a four-rotor, basically a couple of 13Bs bolted end to end. A 13B is somewhat modular—with a long enough eccentric shaft, you can stack the engines like Lego. (There’s no end to this, at least in theory. Find a good machine shop and a few RX-7s, you can build your own rotary freak show—a four-rotor keg, an eight-rotor, whatever. Try that with a conventional engine, adding pistons, without your own foundry.)
Four 13B rotors meant 2.6 liters. Fuel-injected, the engine made 610 hp at 8500 rpm on regular-unleaded pump gas. GTO rules meant the car was a “silhouette”—a steel-tube frame, a production roof panel, and bodywork that looked like an RX-7 if you were maybe a little drunk. The
2400-pound car was designed by Lee Dykstra, who drew the famous Group 44 Jaguars. Defending GTO champ Pete Halsmer was hired as one of the drivers, coming off a season in the V-8 Mercurys of Ford icon Jack Roush.
GTO was a manufacturer-sponsored knife fight, populated by heavy spenders like Ford and Nissan. The Mazdas were hundreds of horsepower weaker than the V-8s but more nimble and hundreds of pounds lighter. And Halsmer was a beast. In 1991, he and our test car won the championship.
Tube-frame race-car construction has not changed significantly since 1990, so in 2017, GTO cars seem perched out of time, neither new nor old. The RX-7 has no doors, because the car’s frame would get in the way. There is a wing but no real downforce, just hefty spring rate and wide slicks that take a lap to warm. In back, production RX-7 taillights are buried between huge composite fenders.
In person, the car looks prickly, like an inflated puffer fish. The rear bodywork is held on by four cheap-looking plated latches. It covers mostly air, because the outside of a tube car has little to do with where its designer has placed mass. So you look under the RX-7’s broad rear windshield and see space either open or packed tight. A differential made by stock-car supplier Mid Valley Engineering, a sea of plumbing, and a tiny blade anti-roll bar. The right side of the car is dominated by a sewer pipe of an exhaust covered in wire mesh. The mesh appears cagelike, as if its removal might prompt the muffler to escape and eat your dog. Up front, the engine is buried so far aft as to be almost hidden under the dash.
There is no way into the car without wrestling through the tiny side window like a doofus. There is also no way to drive the thing without earplugs. The engine is always yelling, even at idle, abusively loud even for a race car. It would be loud for a bomb in front of a megaphone. The pitch changes with rpm, growing more shrill. So you drive out of the paddock, feeling out the short-throw clutch (RAMPA RAMPA RAMPA goes the engine, lumping along spastically, on its idle map). Screwing with the dog-clutch five-speed that doesn’t work unless you’re ferocious with your shifts (RAMPA RAMPA RAMPA). Then warming the tires and piling out of the pits.
And the first time you belt open the throttle, it hurts.
The noise is apocalyptic. An uncorked four-rotor blasts language out of your body—if you have any emotional tie to cars whatsoever, your internal monologue almost immediately mutes. It took me the better part of a lap to regain conscious thought and drive on more than instinct. Bathing, the whole time, in engine wail, this nutty, bright honk, a kind of stretched and reedy alto that reminds you of old Formula 1, a decade or two back, before the sport decided to stop thinking about the spectator experience and instead disappear up its own rumpus. Just the compounding sound of a weird spinny triangle thing that seems to feed on its own motion.
If you have not yet figured this out, rotaries are like nothing else on earth.
The Wankel dominates the car. Peak torque breezes through the rear tires at 7000 rpm, some 390 lb-ft. Everything you do is sussing how to manage that, sorting where the chassis wants your effort. It turns out that the answer is everywhere, because unlike a lot of pro race cars, a GTO RX-7 never wants to help you. A 13J demands to be kept boiling and flitting; if you so much as flirt with leaving its thin powerband, the car feels parked in mud. This is married to surprisingly heavy steering, given the car’s blueprint—you drive with your upper arms, levering lock into the wheel. The brakes are attached to a pedal like a stone wall and nothing special, just pro-grade race-car brakes, modulated by pressure. They feel remarkably normal next to the engine, so you never worry about them.
It is undeniably work. An odd cross between maintained momentum, like every slow race car ever built, and gut-check speed, like every fast one. The RX-7 has enough grip that it makes the majority of its corner speed on entry; once you figure this out, you find yourself inching up to fast laps, bending the car down to an apex on the brakes. You also spend the first couple of those laps wondering why you’re being such a pated weenie and braking too much. Even if the car is sliding on entry. Which it wants to, often, and in a way that makes you either laugh or think seriously about insurance. Sometimes both.
Halsmer once told an interviewer that the RX-7 cornered faster than his old Cougar. The former was more difficult to drive, he said—power way up high, not enough torque to steer with the throttle. He was right, of course, but he could have been describing every good Mazda, from Miatas to the RX-8.
The Halsmer RX-7 is currently vintage-raced by Mazda PR man Jeremy Barnes. In Laguna’s garages, Barnes told me the car was “a nightmare bitch from hell,” but he said so casually, in a tone one might use to order a sandwich. Which is not how you talk about a Miata, but then, a Miata does not make 600 hp. “I love it,” he said, “but it just wants to hurt you. It doesn’t care who’s in charge, you or it, but someone’s going to be.”
In 1994, Mazda took this very car to the 24 Hours of Le Mans. RX-7s had run there before, but the model was never aimed at an overall win. When the company chased that goal, it used a version of the same madness, just faster and more expensive. And an order of magnitude more insane.
1990 Mazda 787 IMSA GTP
LE MANS IS THE MOST DEMANDING road race on earth. Cars have been doing 200 mph there, day and night, rain or shine, since the Sixties. Mazda is the only Japanese carmaker to have won the race overall. It did so only once, with a car like this.
Le Mans first saw a Mazda rotary in 1970, when a Belgian team put one in a Chevron prototype. The car broke after 19 laps. The result played into a popular notion of the day, that Japanese engineering was too fragile for the rest of the world. In 1973, the race met its first Japanese team with a Japanese car and Japanese drivers: a 224-hp prototype called a Sigma MC73, powered by a Mazda rotary. The car’s clutch failed after 79 laps. Stereotyping continued.
Sigma was led by an energetic manager named Shin Kato. When he returned to Japan after the race, he ed Takayoshi Ohashi, head of Mazda Auto Tokyo, which would become Mazdaspeed. The two men met in the city of Okazaki, in a temple, where they talked for hours. Kato and Ohashi wanted to go bigger. They agreed this could be achieved one of two ways: “take time to develop our know-how,” Ohashi said, or start racing immediately, failing and learning on a world stage.
“We had no doubt,” Ohashi said, “that the second solution was the one for us.” Kato suggested Mazda enter Le Mans, because it was the toughest, most prestigious event he could think of.
These men knew little about international motorsport, just that the process would be painful. Years later, Ohashi told an interviewer that he left that temple with a shiver in his spine. He and Kato went to France, where they met Le Mans officials.
Ohashi said the French looked at them funny. Mazda entered Le Mans for the first time in 1974, but the race was a nightmare. Tires exploded. An engine had to be dismantled in the pits. They tried Le Mans again in 1979, with an RX-7 that didn’t qualify. Two works RX-7s ran in 1981; each blew its gearbox before the 12-hour mark. In 1982, a factory RX-7 finally ran all 24 hours.
For 1983, against every ounce of conventional wisdom, Mazda decided to go big: The company would design and build its own prototypes—race cars not based in production, aimed at the sharp end of the grid.
The choice was akin to running a marathon before learning to walk. But somehow, in ’83, a stumpy, little, 13B-powered Mazda called the 717C went to Le Mans and won the Group C Junior class. It finished 12th overall. Seven years and six redesigns later, the Mazda garage at la Sarthe held four rotors and a hyper-modern carbon fiber tub. (The Porsche 962, then the dominant Le Mans Prototype, used aluminum and steel.)
The new car was called 787. A 1991 update, dubbed 787B, won Le Mans overall. Context only burnishes the accomplishment. For 10 years, the French classic had been dominated by giants: Porsche (1981-1987), Jaguar (1988, 1990), and Mercedes-Benz (1989). Mercedes first entered the race in 1930, and Porsche and Jaguar began running it in the 1950s. They knew the event's quirks, to say nothing of what it took to build a modern front-runner–an abuse-proof, 200-mph race car producing thousands of pounds of downforce.
Mazda's effort was drawn by former Formula 1 designer Nigel Stroud. It was a happy piece of wonder. For one thing, the winning 1991 car was sponsored by Japanese clothing company Renown, which agreed to have the car painted orange and green, like an argyle sock. For another, it was powered by a bellowing, obnoxious four-rotor called the R26B.
The R26B was the same displacement as a 13J. Like that engine, its exhaust spat fire on downshifts. But it was a ground-up design: three spark plugs per rotor, to the 13J’s two; an obsessive attention to materials; and fanatical development.
Every Le Mans design is obsessed over, but this was something else. Consider the months before the 1990 season, when a small team of Mazdaspeed employees were told to add 100 hp to the R26B while somehow using less fuel. Chiefed by engineer Yoshinori Honiden, the task represented a nutball goal for a non-turbo endurance engine already operating at the edge of sanity. They worked nights, weekends, and holidays.
“All these young engineers sacrificed” their private lives, Honiden said. They “ate nothing but prepacked noodles. It was very painful for me to have to impose this on them.” But they found the power. By nothing short of a miracle, they also made the engine less thirsty in the process. By 1991, the 787B produced 700 hp at 9000 rpm, with a usable powerband around 3000 rpm wide. These were huge numbers for a racing rotary, but the front-running Porsches made around 50 hp more. Those cars were also a known commodity—their basic design was almost a decade old.
I was lucky enough to test a 962 several years ago. Comparing that car to the 787 helps illustrate Mazda’s achievement, but also the difference between upstart and veteran. Next to the Porsche, the Mazda seems jewel-like, smaller and more dense. You fall casually into the former and fight limbs into the latter, as if Iron-Manning yourself into a suit of car that doesn’t want you there. The 787 is claustrophobic, windshield in your face; the 962’s instruments seem at arm’s length by comparison. The Mazda’s roof is more of a hat over your helmet. The tall shift knob and its woolly linkage are so close as to feel installed in your arm.
The theme carries beyond the cockpit. The Porsche looks easier to work on, more air under its skin, the seeming result of decades in endurance racing, where everything eventually breaks. The Mazda, by contrast, gives you the sense that its creators wanted nothing in the way of shrink-wrapping one of the most reliable engines in the business. (Old joke I learned in club racing: Because rotaries are generally tolerant of abuse, the line goes that you can’t overrev one, just piss it off.)
Finally, the Mazda offers countless neat touches: trombone-like variable intake runners, to maximize torque. A doorsill sticker printed with the engine computer’s diagnostic blink codes, for easy reference. Or the synchromesh, five-speed Porsche gearbox, borrowed from the 962, because it was essentially unburstable.
That Mazda was able to purchase that gearbox hints at the Porsche’s development budget and productionized ubiquity. Nearly 100 962s were built, but just six 787s were made, including the winning B. The 1990 787 we tested finished eighth at Le Mans in 1991. Up close, it looks like the hand-built gem that it is, and wonderfully aged. The car has the patina of use—rock chips, paint rubbed from panel edges.
And no 962 ever sounded like this. Those cars are sedate whooshes and grumbles, muted by turbocharging. A 962’s exhaust is so low-fatigue, you almost forget about it. The first time I blipped the 787’s throttle, parked in the paddock, I jumped in the seat. It whooped. The effect was somehow compounded by knowing the engine’s physical size—it made the thing seem friendly and relatable, as if you could carry it around like a pet. (Cue yelling in the family kitchen, over the sound of a screaming R26B: “MOM THIS AIR-RAID SIREN TOTALLY FOLLOWED ME HOME CAN WE KEEP IT?”)
As with the RX-7, I was allowed 20 minutes at the wheel. Mazda’s people told me to shift around 8500 rpm, where the engine pulled like bejesus; it was revved only slightly higher in period sprint races. (For qualifying, they occasionally went to 11,000.) The 787 was screamier than the RX-7, if slightly broader in power delivery. And completely unlike the 962, which slathers torque across the tach. On a sunny, dry day, for 20 minutes, the Mazda’s whoops and yowls were bliss. At four in the morning on the Mulsanne, they must have been like a knife to the cortex.
The rest of the car only ramped up the abuse. Even with ear-plugs, an extended dose of the noise alone would have turned my brain to liquid. Add in the 787’s fat cornering load and boiling cockpit heat, and that liquid would have run down my spine and pooled in the seat. The steering offered the compounding weight of a car with significant downforce—the wheel gets heavier as you go faster, air flying the tires into the ground. The Mazda seemed to bat into corners more aggressively than the Porsche, requiring more caffeinated attention. A 962’s steering is slower, more work in a quick slide but easier on the arms. The only quality the two cars seem to have in common is a set of brakes that feel like they could turn you inside out.
Human beings obviously coped with this sort of thing for hours on end, but I have no idea how.
Rule changes and economic shifts meant that Mazda never again tried to win Le Mans overall with a rotary. Which both amplifies the accomplishment and makes it bittersweet. As does the death of the marque’s last rotary road car, the RX-8, in 2012.
The RX-8 was far from perfect—the same old problems of torque and fuel and oil—but it was hard to drive one without smiling, buzzed on the sense of difference. When the car was discontinued with no announced replacement, many people assumed that Mazda had simply accepted what much of the world believes: Rotaries and modern roads don’t mix.
Perhaps. It would be nice, really, if the 787B weren’t the peak of an alternate kind of thinking. But in hindsight, as with every RX-7 built, the car’s story remains proof of a universal truth: It’s better to think differently, and reach a glorious dead end, than to never try at all.