EVO vs. STI

When we last saw our EVO and STi, they were playing catch-up with each other. The EVO needed to catch up with the STi's power output, so we gave it a Rhys Millen Racing exhaust, AEM cam gears, and a custom tune (with a piggyback Exede ECU) from Vishnu Performance. The STi wanted to catch up with the EVO's handling, so we threw the works at it; a Praxis suspension, Hotchkis adjustable anti-roll bars, adjustable rear toe links and front camber plates. Then we aligned the living hell out of it and headed to the track.That's where it all fell apart. The EVO's stock clutch should've been able to handle the power, but we'd already done so many drag launches, it was too worn to perform. On the STi side, things were looking good until the alignment slipped on the right rear wheel and all our camber and toe settings went to hell. Right before it all went goofy, though, we saw .98g in one direction. So, fixing things.

EVO VIII A fresh stock clutch would have solved the problem for a while. We needed something that would last, though. Something we could take to the track over and over and never have to walk away trailing excuses. Well, excuses about the clutch, at least. But this is a street car, and we also wanted something smooth enough we could pull out of a parking space without stalling three times.

A cerametallic, twin-disc clutch sounded like exactly the wrong solution for a daily-driven street car, but we kept hearing tales of Exedy's cerametallic twin-disc being perfectly streetable. We ordered one up and immediately thought we'd made a mistake. It was a gorgeous piece of race hardware, but there was no way, we thought, this thing could ever be streetable. A good street clutch will usually have an organic friction material that engages smoothly. It could have a lot of metal in the material to handle the heat of some hard launches, but cerametallic friction materials are usually grabby.

Most street clutches will also have a marcel spring--basically just a wavy piece of sheet metal--between the friction material and the clutch disc. This spring gets smashed flat when the pressure plate clamps down on it, but compressing the spring spreads out the clutch's engagement, making it easier to modulate. The Exedy twin disc, like any race clutch, has the friction material glued directly to the disc with no cushioning. This should make the discs much stronger, as it's virtually impossible to rip the material off the disc, but what about getting out of the driveway?

The sole concession to civility that we could spot was the fact that both clutch discs were sprung, meaning there are springs between the disc and the hub that are splined to the transmission's input shaft. These do relatively little to cushion clutch engagement compared to the marcel springs we were hoping to see under the friction material. Think about it: With a marcel spring, you could hold the clutch halfway through its engagement zone as long as you want (or until the clutch burns up from all the slipping), but the springs on the hub only absorb a few degrees of crank travel before they're fully compressed. These springs are really there to absorb torsional vibrations from the crankshaft before they get to the transmission and cause annoying rattling noises.

So, by the looks of it, the Exedy clutch was going to be an unbreakable brute, but driving the car was going to suck. When we dropped the car off at Road/Race Engineering to have it installed, the guys there saw exactly what we did. "It's going to cost the same to install the next clutch," they warned us.

In the end, though, we were both wrong. Well, half wrong. As expected, it is an unbreakable brute, with vastly more holding capacity than we need, and virtually no sensitivity to heat and abuse. You can launch the EVO as many times as you want and the clutch performs every time. But it's in traffic that the Exedy clutch impresses the most.

The clutch pedal is heavier than stock, but not that heavy. Your leg won't get tired, even in traffic. Mike Welch at Road/Race put a stick and a spring scale to good use and found some numbers for us. It takes 28 pounds from your left leg to stomp on the stock clutch, and 38 pounds for the Exedy. With two discs, there's twice as much torque capacity available for a given clamp load, so you don't need the pressure plate to grab all that hard.

However, the first time anyone lets out the EVO's new clutch, they stall the car because it does grab lower on the clutch pedal and more suddenly than the stock clutch. But after that first lesson, it's very easy to drive. Despite the aggressive friction material and solid discs, it's surprisingly easy to modulate. Combined with the torqueless off-boost performance we have with both cams retarded 5 degrees, it's a bit of a dud in traffic, but it's a smooth and driveable dud. Back at the dragstrip, the times weren't exactly what we'd hoped. Run after run, the EVO ripped off 13.4-second runs at 103.1 mph. Well, if you were there, you would've called them 13.3s, but around here, a 13.355 is a 13.4. With the worn stock clutch, our car ran 14.0, so it's clearly better now, but we know of many similar cars running much faster. Blame the weather, blame our car, blame the gas, blame the driver, just don't blame the clutch. It grabbed hard run after run, no matter how abusive we got.

WRX STi Back in the land of the STi, solutions were harder to come by. In addition to the alignment problem, the Praxis suspension initially wasn't as good as we remembered it on the stock WRXs we sampled several months earlier. The front suspension, in particular, seemed under-damped, and also bottomed out far too easily.

As you'll recall, the Praxis system uses air springs and adjustable shocks to switch between Track, Sport, and Touring modes. In anything but the Touring mode (the tallest), the front suspension would occasionally bottom out hard enough to pound right through the bump stop to something very solid sounding. The left front strut would also leak air whenever we left it parked for a few hours, leaving us with a lilting, wounded-looking car to return to. The system self-levels when you start it, so the slow leak didn't hurt anything but our egos.These problems are fixed now and the car rides and handles great. We'll spare you the tedious details of how we worked our way around to the solution and skip right to what we learned.

Thing we learned #1: The Praxis suspension doesn't like camber plates. When the engineers at Praxis first suggested that the camber plates might be to blame for the underdamped front suspension and the frequent bottoming, we were incredulous, but after returning to the stock, rubber top mounts (and switching to a new set of front struts after some ignorant nob stole our originals), the suspension was as good as we remembered. Here's the best explanation we can come up with: The rubber top mounts are relatively rigid radially. In other words, they don't let the top of the strut move around much side to side or front to back. They are very soft axially, though, meaning they can deflect straight up and down a shocking amount. On a hard bump, the rubber top mount can let the strut shaft move up as much as 3/4 of an inch. This effectively adds 3/4 of an inch of compression travel. There's part of our bottoming-out problem.

The other part is shock valving. All dampers are filled with mysterious little valves that open and close automatically to adapt damping characteristics to the conditions at hand. How many valves and what they do is all part of the black art of shock design, but let's take a very simplified view of what's happening: Say you want a lot of damping at low shock speeds. Low shock speeds are the ones that control body motion, so firm, low-speed damping will mean better transitional control, less body roll in quick maneuvers, and a car that faithfully follows the road's surface. Damping always increases with speed, though, so if you hit a bump and the shock suddenly has to move, say, five times as fast as it does controlling body roll in a slalom, the shock will resist the motion with five times the force. The resulting impact will knock your fillings out. So in practice, a valve blows open under these conditions, reducing the damping force to, say, 2.5 times what it was in the slalom. With the rubber top mount, small bumps are absorbed by the rubber, so it takes a real bump to soften the shock. Camber plates don't deflect, though, so even small bumps get transmitted into the shock. With the camber plate, then, it takes less of a bump to soften the shock. Road texture, we're guessing, was often enough to do it, and that's why the front of the car felt floaty and underdamped with the camber plates in place.

Thing we learned #2: The Praxis suspension is very sensitive to setup. You need to set the ride heights exactly as the instructions say or the spring rates and damping will be mismatched and the car will feel like crap. Setting the heights, using the stock Subaru eccentric bolts to adjust camber (instead of the Hotchkis camber plates we so dearly miss), and using the Hotchkis adjustable toe links in the rear means a four-hour alignment from hell. Don't get any ideas about changing the ride height calibration to improve the stance, either. Lower the front half an inch from the recommended setting and the spring rates will change too, leaving you with a WRX that both rides and handles like absolute crap. Don't ask how we know.

Thing we learned #3: We're lazy. The Praxis suspension lets you switch between three different ride height and spring rate combinations at the touch of a button, but you have to crawl under the car to adjust the damping to match. That means laying on the ground and fumbling around blindly for hot, dirty parts and then turning them and counting clicks. "Don't be a sissy," you say, but in practice, we're almost always driving around with the height in one setting and the damping in another, which doesn't work very well at all.

Thing we learned #4: Even with all our first-attempt mistakes fixed, the Track mode still has so little travel, it's useless. We've relabeled it "Car Show mode."

Thing we learned #5: Self-leveling isn't always good. Since the air trapped in the springs would raise and lower the car itself as it heats and cools, the Praxis suspension has to be self-leveling. The system periodically checks the ride height sensors and lets air into or out of the bags until all four wheels are sitting where they should be. The system is smart enough not to do this while you're accelerating or braking, since it's way too slow to effectively combat squat or dive, but it's not smart enough to not do this if you're parked on an uneven surface. Since the system keeps leveling for several minutes after you turn the car off (to deal with air in the springs cooling), you can park on an uneven surface, go buy a six pack of Schlitz, and come back to find your car picking a wheel up off the ground in an attempt to put all the wheels where they belong. Drive away gently and the system will relevel itself in less than a minute. Forget to pay for the Schlitz and tear-ass straight from the parking lot in a high-speed chase, though, and you'll find the heights and spring rates all jacked up, since it knows not to level while you're accelerating or braking. Worse case, you could understeer in one direction, oversteer in the other, and quickly find yourself a star of the World's Most Inept Car Chases #17.

Results: Properly set up, with an alignment that gives you zero toe at both ends, 2.5 degrees of camber in the front and about 1.1 degrees in the back, our combination of Praxis suspension with Hotchkis front and rear bars and adjustable rear lateral links works brilliantly. Ride quality in Sport mode, which is the mode we usually drive in, is good. With the rear bar on its medium setting, handling numbers aren't that spectacular. The car understeers at the limit, pulls 0.92g on the skidpad and runs the slalom at 70.4 mph (stock was 0.91g and 70.7 mph), but put the rear bar on stiff and the car wakes up. Skidpad grip jumps to 0.98 g (still on stock tires) and slalom speed creeps up to 71 mph. (The difference in slalom speed is negligible, so let's just say all three were the same.)

It wasn't until after testing that we discovered a tweak that should have a large impact on slalom speeds and is, subjectively, the single biggest improvement in handling precision and driving enjoyment of anything we've done. Best of all, it costs $40. All Subarus have vague, sloppy steering at turn-in. Once you've committed to a corner, the steering is usually very linear and direct, but that first moment when you're picking your line is fraught with uncertainty. At the speeds the STi is capable of, this trait is downright frightening. The problem is that the steering rack is mounted in bushings that appear to be made of a proprietary mix of Teflon and KY Jelly. Turn the steering wheel right and the first thing that happens is the steering rack slides left. Only when the rack settles into place does steering feel return. The stickier your tires, or the older your car's steering rack bushings, the worse this problem gets. A simple set of polyurethane steering rack bushings from Whiteline makes a ridiculous difference in how the car drives. Steering precision is vastly improved and turn-in is so much sharper, we almost ran over the curb on the inside of the first corner we went around. There is no down side. Is anybody at Subaru reading this?

Finally, our new rolling stock. The same jackhole who stole our front struts without bothering to spend the four hours it would have taken to steal the rest of the suspension, also took our wheels. At least he'll be able to sell the wheels. Since we wanted to finish testing with the stock tires, we got a new set of 17s. (We're also convinced 18-inch wheels look dumb on Subarus, but we're all like 30 years old, so maybe you shouldn't trust us on that.) Looking at every wheel Prodrive offers, we couldn't find one we wouldn't kill to have on our car, so we just called and asked them to send whatever 17-inch wheel they wanted. What we got was this GC 010. Not only are they gorgeous, they're lightweight at only 16.5 pounds, and since they're forged in Japan by Rays Engineering, we're pretty sure they're strong, too.

When we installed them, we found a few surprises. We had a new set of Bridgestone RE070s (the stock tire) shipped from the Tire Rack, and had them mounted at Shoreline Motoring in Huntington Beach, Calif. Luckily for us, Shoreline is run by longtime SCC contributor Dan Barnes, one of the most anal tech geeks we've ever met. He immediately told us to throw away the pretty metal valve stems that came with the wheels. The valve stems are secured to the wheel with two nuts (one to tighten, the other to jam the first in place) on the inside of the wheel. Problem is, the nuts stick out right on the part of the wheel the bead has to slide on when it's mounted. This makes the tires difficult to mount and risks damaging the tire bead, the valve stem, or both. Without Locktite, there's also a chance the nuts can back off inside the tire. You'll never know until the valve stem shoots out of the wheel like a dart from a blowgun. Most metal valve stems install from the inside and have the nut on the outside. If the nut backs off, air pressure is still holding the valve stem in place. We simply switched to cheap, rubber stems. When we put the wheels on the car, we were in for our second shock. We were planning to run only about 0.5 degrees of camber in the rear. The stock alignment is about 1.0 degrees, but we have less body roll to worry about, and, frankly, we don't want too much rear grip. With the rear tires standing nearly straight up and all the air out of the rear springs (just one of the thousands of steps in our four-hour alignment), the rear tire ran into the fender. That's when we finally noticed the offset.

Counting on Prodrive's Subaru experience, we assumed the offset would be close to the stock 53mm, but these were 43mm. This is what people in the wheel business call an "aggressive fit." Basically, the wheels are pushed so far to the outside, they barely clear the fenders. This looks good, but die-hard perfectionist tech geeks like us call this kind of fitment "doesn't fit." Changing the offset by 10mm usually screws up steering feel and, in this case, it limits both how wide we can go with our future tire choices, and how creative we can get with our alignment settings. Because of the 43mm offset, the car now has noticeable torque steer and we had to run 1.1 degrees of camber in the rear to clear the fenders. The extra rear camber forced us to go one step stiffer on the rear anti-roll bar to get our 0.98g.

Back to the EVO 0.98g! That just won't do. The EVO's supposed to be the handling king here. We went to great lengths to get that much grip from the Subaru on its stock tires, but by the time we did, our EVO had worn its third set of stock tires to the cords. Time for something new. First up was a bigger rear anti-roll bar. We still consider the EVO's stock suspension to be virtually without fault, but there are times we wish for a hair less understeer, or just a bit more willingness to step the tail out when asked. We've sampled EVOs with Cusco's adjustable rear anti-roll bar, and that's exactly what the car needs.

The Cusco bar is only 1mm larger than stock. That's a tiny change, but it's all the change the EVO needs. With the Cusco bar on its medium setting and everything else stock, the car can be steered more easily with the brake and throttle, and you can apply power much earlier in a corner. Mash the throttle in a long sweeper and the stock EVO will start to understeer, while the Cusco-equipped car will continue to rotate and track through the corner to ludicrous speeds. Since our tires were toast, we decided to step up to something gummy enough to keep the STi at bay. Time for some R-compound rubber. Sticking with the stock 235/40-ZR17 size, we upped the stick with a set of Pirelli PZero Corsas, which we had mounted at Stokes Tire in Santa Monica, California. Usually, making the R-compound leap means a dramatic transformation in performance. Both the EVO and STi come from the factory with near R-compound tires, though, so the jump was subtler. Grip was up, no doubt. In fact, with just tires and the rear bar (set on medium) the EVO now pulls 0.99g (up from 0.95g stock). Remember, this is a tall, heavy, all-wheel-drive sedan that's still comfortable to drive on rough roads.

Steering sharpness and turn-in, surprisingly, weren't really improved with the Pirellis. Fact is, the factory Yokohama A046 are so firm and grippy, there's little improvement to be made in terms of sharpness. The Corsas do maintain their composure better on the track, and, like most R-compound tires, make little to no noise when you're sliding. Great news if you're prone to public powerslides. But even with their improved high-temperature performance, the outside edge of the front tires took a beating when we subjected them to two days of on-track punishment. With the extra grip, there's more body roll than before, and the front could use a bit more camber.

Finally, something that has nothing to do with trying to beat the STi: brake pads. The EVO's brakes are already better than the STi's. They have more consistent pedal feel and don't fade as easily. The pads do feel a tad wooden, though, as if they're always just a little bit glazed. They aren't glazed, but they could be a bit grippier. We swapped just the front pads with Ferodo DS2500s. The Ferodos work well over a surprising temperature range. Stone cold they have better initial bite and a more linear feel than the stock pads, and as they get hotter, nothing seems to change, while the stock pads eventually start losing bite. There doesn't seem to be any downside to the Ferodos, either. No squeal, no obviously excessive wear, not enough extra brake dust to cry about. It's not often you get such improvements with no down side.n