We've yearned for this day for as long as any of us can remember. For nearly a decade, Mitsubishi and Subaru have duked it out with ever-faster versions of the Lancer Evolution and Subaru WRX STi. Until this year, we've been forced to watch this battle from the sidelines.
While the rest of the country thrilled to the geezer wrestling that was the Mustang vs. Camaro battle, we were staring dumbly at Japanese magazines we couldn't even read, wishing desperately that the real battle could come here. Now, with both an EVO and STi in our project car fleet, it finally has.
Stepping into the ring, the STi has the power advantage, while the EVO leads handling with unprecedented nimbleness and precision. The battle begins, then, with each car trying to match the other's inherent strengths.
STi: Float like a butterfly
Judged by itself, the STi handles brilliantly, but judged against the bantamweight nimbleness and relentless stick of the EVO VIII, it feels vague, rough and short on precision. But here's the problem: The EVO manages its handling magic while still maintaining the civility needed to transport your wife, your dog and an infant or two over railroad tracks without anyone risking kidney failure. We don't know many aftermarket Subaru suspensions that claim the same thing.
Save one.
Bridgestone's Praxis air suspension, which we sampled back in the November 2003 issue, doesn't claim it can be a track hero and a marital preservation tool at the same time, but it does claim to switch from one to the other in less than a minute. We sampled this suspension on a much less powerful stock WRX on the track and the relatively smooth roads around Palmdale, Calif., and were suitably impressed, but can it handle the solid-fuel thrust of the STi over the bumps and grinds of the big city?
One thing we did learn from our brief experience with the kit is that, good as it is, it's not a complete solution. Any Subaru suspension needs the additional rear roll stiffness of a larger rear anti-roll bar, and the alignment changes that happen when raising and lowering the air springs mean some adjustable toe links would be nice. The plan, then, is to augment the Praxis system with some extra goodies from Hotchkis.
Two days after we took delivery of the STi, three giant boxes showed up from the Tire Rack, Praxis' exclusive distributor. The kit is very complex, requiring not only the four struts, but also an air pump and tank, a valve block to distribute the air, an ECU to decide where it goes, wheel position sensors to determine ride height, an accelerometer to tell the ECU not to make height adjustments during hard driving, and a large wiring harness to tie it all together. Installation is a big, time-consuming task, but it's technically simple, requiring no special tools and the drilling of only four holes.
Everything is made specifically for the WRX, so the mounting brackets fit perfectly and every branch of the wiring harness is clearly labeled and exactly the right length to reach its destination. You do have to cut the air lines to length and route them through the car, but if you can be trusted with a razor blade, you can handle that. We managed the installation in three or four sessions, driving the car in between. If your WRX is your only transportation, you should still be able to install it yourself.
After installation, you have to calibrate the system by precisely setting the ride height in each setting with three floor jacks. This is fairly tedious, but it's the only way the ECU can know what the wheel position sensor input really means. While you're driving, the system constantly monitors ride height for any changes caused by load, slow leaks, or changes in temperature.
While it ignores bumps, acceleration squat and brake dive, if the computer sees the average ride height over a 40-second period is out of spec, it will adjust it on the fly. These adjustments involve one end of the car moving up several inches, then lowering back to the preferred height. This can be a bit disconcerting at first. Even worse, if you happen to be parked with one wheel on a bump for a while, the system may try to compensate, leaving you slanted once you hit flat ground. Then if you drive hard for the next 10 minutes the car will stay crooked, since the built-in accelerometer knows not to adjust during squat and dive. In other words, the system has some inherent quirks that you must learn to live with.
More adjustability
Before doing any testing, we headed to Hotchkis Performance for more parts. Every Subaru we've ever driven has benefited hugely from a larger rear anti-roll bar. Shifting cornering load to the rear wheels, as the bar does, reduces understeer and makes the car far more lively to drive. Hotchkis suggested bigger bars at both ends, however. The extra roll stiffness reduces overall body roll significantly, which helps keep the tire's contact patch flat on the ground. Its competition-series anti-roll bar set still has enough rear roll stiffness bias, Hotchkis promised, to make all our tail-out powerslide dreams come true.
The Hotchkis bars are beautifully made, with impressive attention to detail. The bars themselves are tubular, which allows them to be both lighter and stiffer than stock. The front bar, for example, weighs 5.3 pounds, 1.4 less than stock, in spite of being larger and stiffer.
The concept of hollow bars is simple. The stiffness of an anti-roll bar goes up with the fourth power of the diameter. In other words, if you double the diameter of a bar, its stiffness will go up by 24, or 16 times. This means the metal on the outside of the bar's cross section is far more important than the metal in the core, so why not just take the core out? If a 1-inch bar is 16 times stiffer than a 1/2-inch bar, then a 1-inch bar with a 1/2-inch hole in the middle will still be 15 times stiffer than the original, but far lighter. Making hollow bars takes more expensive machinery, but Hotchkis' CNC-bending machines do the job just fine.
The front bar replaces the already precise factory ball-joint-style end links with heim joint links. While the stock links are precise, they're not very strong, and tend to break under the increased load of the bigger bar. In the rear, it's the mounting brackets suspending the bar from the bottom of the car that are weak, so Hotchkis includes stronger brackets.
The rear suspension naturally toes-in under compression. This means that during cornering, the outside rear tire turns inward slightly, stabilizing the car, or causing understeer, depending on your perspective. This also means that lowering the car with the Praxis system makes the rear toe-in, which leads to relentless understeer. Setting rear toe near zero in the lowered settings means toe-out in the Touring setting, which will cause frightening snap oversteer.
Our solution is to install Hotchkis' adjustable rear toe links. While rear toe is adjustable with the stock links, it takes an alignment rack to really tell where toe is set. With the large, turnbuckle-style Hotchkis links, we should be able to mark separate settings for touring and track modes on the alignment rack, and simply turn the adjuster to those preset marks when we're at the track.
Finally, up front, we're taking a similar approach with front camber. Factory eccentric bolts in the strut mounting bracket allow camber adjustment, but not with much precision. Camber plates move the adjustment to the top of the strut, where each degree of camber takes quite a bit more movement.
In place of the metal spherical bearing most camber plates use to locate the top of the strut, Hotchkis uses a top bearing of its own design using a mix of Teflon and polyurethane to provide noise isolation and the flexibility to follow the strut's movement through its travel.
Alignment hell
Putting all this adjustable stuff to use is impossible without a good alignment shop. We called on the services of a super-jammy Hunter laser alignment rack at Victoria Tire and Wheel in Huntington Park, Calif. With three ride height settings and countless extra points of adjustability, you can't drop a car like this off at the alignment shop and walk away. We had to dive in and get dirty.
Praxis has recommended alignment settings in Touring mode that are supposed to provide reasonable tire wear over the suspension's full range of adjustment. We followed the suggestions for front toe (1/16-inch toe-in), but set front camber more aggressively at -2.3 degrees in Touring mode. This translated to -2.5 degrees in Sport, and -2.6 in Track mode, since camber increases slightly under compression. In the rear, we found camber at -1.5 degrees in Touring and didn't bother changing it. We set rear toe to 0 in Touring mode, made a mark on the Hotchkis adjusters, then lowered it to Sport, reset and re-marked the adjusters. When we tried the same in Track mode, the front bumper blocked the toe-measuring lasers and the readout went blank. We stopped there.
Setting up the alignment rack, there are several times when the car must be jacked up to reach bolts or center the floating plates under the front tires. When it's in the air, of course, the Praxis suspension will try to lower itself, so there's a switch in the trunk that will tell the ECU to knock off all that silliness. Still, we reset the ride height switch each time we made an alignment measurement, just to be sure.
Even with the more precise Hotchkis adjusters, rear toe adjustment is a little touchy. The tiny amount of slack in the adjuster threads is enough to affect rear toe readings, so you have to turn toward the mark from the same direction each time and tighten the jam nuts in the same sequence to get the same setting.
OK, so is our STi ready to take on the EVO? No fair, if we haven't tweaked the EVO.
Here goes.
EVO: Sting like a bee
Just as the EVO's handling shines an unflattering light on the STi's suspension, the EVO's otherwise staggering powertrain looks positively squirrelly next to the brutish output of the 2.5-liter STi. Time for the Mitsubishi to hit the gym.
In the September '03 issue, we pitched the Stock EVO VIII against the best German iron. Hedging our bets, we also invited another EVO mildly tweaked by Vishnu Performance. The lessons we learned from that car are going directly into our EVO.
The first step, then, is to uncork the stock exhaust. There is little to complain about on the stock exhaust, but, by removing stock packaging and noise constraints, it can be made better. The critically important downpipe casting, which gathers exhaust directly after the turbo, is actually quite good. In most factory turbo cars, this piece forces flow from the wastegate to merge with flow from the turbine too abruptly, causing an exhaust flow traffic jam that chokes performance. The EVO VIII's casting, though, provides huge wastegate flow area, and merges the two flows in a wide area a few inches downstream of the turbo. We could hardly ask for better, so we only started looking for improvements downstream.
From this casting to the catalytic converter, the downpipe is slightly mangled, as it leaves room for heat shielding and underbody braces. It was in this complex region that we were most impressed with Rhys Millen Racing's exhaust. Calling on experience juggling performance and durability under punishing rally conditions, RMR squeezed a 3-inch downpipe into the space of the stock 2.5-inch piece without ignoring the ground clearance and heat management concerns that made the stock piece what it was.
The stock heat shielding is most important near the oil pan, where it prevents unnecessary heating of the oil. RMR addresses this by including a self-adhesive aluminized silica heat-shielding pad for the pan. Next, clearance for the front subframe braces is provided by replacing the front brace with a lower-profile four-point brace that's powdercoated red, because red makes things faster.
The rest of the 3-inch exhaust is straightforward and simple, but packed with well thought-out details for underbody clearance and smooth flow. On the road, the sound level is perfect, giving enough growl to make us feel good, but not enough to wake the neighbors or attract any more attention than the massive wing already does. The flatulent exhaust note of past 4G63-powered cars is gone, too, with a smooth, angry, turbine-chopped growl in its place.
Crap gas
Dyno results, however, were strange. Our car baselined at 225 hp on HKS's all-wheel-drive Dynojet. With the exhaust, output went up, but by a different amount and in a different part of the powerband with every pass. Our shitty, knock-prone Southern California gas was screwing with us, forcing the ECU to constantly dodge the dreaded ping. As you'll see in a moment, this should not be typical.
The next step, as usual, was a trip to Vishnu Performance. We relied on Vishnu for tuning on our WRX project car, and the following ordeal should explain why we've come back with the EVO.
The Vishnu EVO we tested back in the September '03 issue showed huge gains from retarding both cams 5 degrees, so we sent our car to Vishnu with a set of AEM cam gears in the trunk. Its mission: tune the car as is, with the stock cam timing and the RMR exhaust, then adjust the cam timing and tune it again. We wanted to see just how much power was available with each step.
That request should've been simple, as Shiv Pathak, Vishnu president, hydrocarbon-sniffing tuner, and occasional SCC contributor, had already tuned dozens of EVOs and, at that point, already had the maps pretty much dialed. Problem is, his EVO tuning is done with the ChipTorque Exede, a very sophisticated piggyback ECU, and a piggyback is limited to making adjustments relative to the stock ECU. Our ECU, for some reason, was different, so the offsets that worked on every other EVO were an utter failure on ours.
Pathak's only choice now was to custom-tune our car, an exhaustive iterative process he tackled with obsessive-compulsive thoroughness. Dialing in our two settings (stock cam timing and 5-degrees retarded) took him 189 dyno pulls, most of a day, and a tank and a half of gas.
Then the real work began. We returned to HKS, wanting all our testing to be done on the same dyno, and performance was back to its old, erratic self. Those 189 tuning runs were on crappy, 91-octane Northern California gas, but now we were on crappy, 91-octane Southern California gas. Despite having the same octane rating, the two fuels are clearly not the same. Luckily, we had anticipated just such a problem, and Pathak was already there, computer in hand. It took four hours and another 100 dyno pulls to get things dialed on our uniquely shitty petrol, but in the end, with pupils throbbing in a hybrdocarbon-poisoned haze, Pathak declared victory. The RMR exhaust, AEM cam gears, and Vishnu's exhaustive tuning took output from 233 to 273 hp with driveability and gas mileage improvements to boot.
Your results should vary
All this gasoline drama and our screwy ECU mean our results are a worst-case scenario. Our ECU was tuned more aggressively, with slightly leaner mixtures and more advanced timing than most. Strangely, our baseline dyno results are a close match for other stock EVOs, but the more aggressive stock tuning may explain why our car took the highway to knock sensor hell as soon as we put a free-flowing exhaust on it. This shouldn't happen on anyone else's EVO, and it certainly shouldn't happen in an area with 93-octane gas.
The exhaustive retuning drama should also be unique to our situation. Pathak has been making dozens of trips to various parts of the country to tune cars. With at least 10 cars gathered at each location, he's seen a comprehensive picture of regional fuel differences and is able to ship his Exede piggyback ECUs with different maps depending on where you live.
Just how big is the difference in fuel? The week before our second tuning session, Pathak was in Texas, where he was tuning on the same kind of Dynojet in similar weather conditions (hot and humid). Several Texan cars with the same modifications that made 273 hp on our car, made an even 300 hp at the wheels on 93-octane Texas gas. Color us jealous.
Testing?
Yeah, well, sort of. With an extra 50 hp, you probably want to know what the EVO will do in the quarter mile. So do we. Unfortunately, after serving us in three test sessions for various comparison tests, our EVO's original clutch is getting tired. While it still works fine in normal aggressive driving, it's starting to chatter when cold, and slips when asked to withstand a dragstrip launch. An Exedy twin-disc will solve that problem shortly.
As for the STi, well, we tried. With the alignment exhaustively dialed in and the rear bar set on its medium setting, it managed 71.6 mph in the slalom in both the Touring (tall and soft) and Sport (lower and stiffer, but still streetable) modes. That's a healthy step up from the stock 70.7, especially on the stock tires, but still shy of the 72.3 mph run the EVO made against the German cars. In Track mode (very low and stiff), we saw 73 mph on one run, but were unable to repeat it. Our slalom times are an average of the two fastest runs, so no repeat means no score. Even on our smooth test track, the track setting had insufficient travel and bounced erratically.
On the skidpad, things got stranger. In Touring mode, the STi managed an impressive 0.94g, besting the EVO's 0.93, but it did so strangely. All cars make slightly more grip with the driver's weight on the inside, but the difference here was dramatic. Turning left, the STi managed 0.97g. Turning right, it was only 0.915g. In Sport mode, the average jumped to 0.95, with a peak of 0.98g turning left.
Finally, we noticed the problem. The right rear wheel had lost its camber. One of the strut mounting bolts apparently slipped (time for a new torque wrench) and the 1.5 degrees we dialed in on the rack ended up looking more like 0 degrees. Lucky for us, this slip helped the rear of the car rotate on the skidpad. We'll be back next time with both rear tires set that way and see if perhaps we've stumbled on the magic setup.