Subaru Impreza 2.5RS

Part X: Riding the Ragged Edge

By Shiv S. Pathak

Photography by Shiv S. Pathak

Just in case it wasn't clear in the SCC Rallycross Smackdown last month, SCC's Project Impreza road car kicked some serious tail at the CRS Ridgecrest Rallycross. Not only did it have enough horsepower to send the competition into a power-jealous tizzy, it also proved to be impressively reliable. Throughout the course of the entire weekend, we didn't experience any engine-related problems. No spurious pinging, no overheating, no oil consumption. As one can imagine, it's only during competition that a performance car can earn the title of durability. After all, during a race, the revs are high, the boost is all over the place and your eyes are on the road (and away from the gauges).

That brings us to Tuner Rule Number One: If anything is going to break, it will break during competition and in front of a lot of people. Of course, a 50-second rallycross run isn't exactly a 20-minute endurance race. That said, it would be terribly naive of us to believe that our turbocharged Impreza is impervious to potential engine grenading. With this in mind, we're going to devote most of this project installment to addressing potential sore spots in our Impreza's aftermarket turbocharged powerplant.

Goofs, Glitches and Grunts
Mass Airflow Meter (MAF)
We're on our fourth one. Despite our attempts at isolation from vibration and protection from contaminants, we have yet to determine the concrete cause (or causes) of chronic MAF failure. We, like everyone else, have our suspicions. But without any successful long-term solutions, we're left feeling rather inadequate. With replacement 1999 MAFs from our local Subaru dealership costing a tad over $300, we're left a little light in the wallet region as well.

One thing is for sure, driving a turbo Impreza would be a lot more enjoyable if we didn't have to worry about a vital engine sensor failing at any given minute with little or no warning. A speed density (MAP-sensor based) programmable EFI system is looking more and more attractive with each successive MAF failure. If you have missed the previous nine installments of this series, we should point out, once again, that the '98 and '00 2.5 RS do not have this problem.

Fuel Pick-up
Months ago, in a search of a higher-flowing fuel pump, we installed a stock pump pilfered from our Project RX-7 Twin Turbo's spare parts bin. Capable of supporting close to 340 hp at reasonably high fuel pressures, the Mazda's pump seemed like an ideal choice for our high horsepower, high fuel pressure application. One problem: The pump's bottom fuel pick-up doesn't extend far enough towards the bottom of the tank. This means that we effectively run out of fuel well before our fuel gauge reads empty. Yes, we've learned this by ourselves and yes, we feel awfully silly.

One solution is to fabricate a new, longer-extending fuel pick-up. Of course, this takes work and is best left for people who have nothing better to do than fabricating jury-rigged fuel pump pick-ups. Instead, we looked for other replacement in-tank fuel pumps. Preferably one that looks and works just like the stock Impreza fuel pump. After a little research we found what we were looking for: Holley Part Number 12-909. Although it is advertised as a (Yes, you guessed it...) replacement RX-7 Twin Turbo pump, it (for some strange reason) looks just like a stock Impreza pump. But one that is supposedly capable of flowing 255 liters per hour!

Installation couldn't be easier. In fact, in order to install the Holley pump, we had to un-modify the very same fuel pump assembly that we had to modify (i.e. bend with pliers) earlier. As an added bonus, we no longer had to use our makeshift intermediate electrical connector as the Holley pump's male electrical connector mated right up with the stock Impreza's female electrical connector. The result? Good flow as well as good fit. Not to mention the fact that we can travel an extra 30 to 40 miles on a tank of gas before it runs dry.

Fuel Slosh
For some strange reason, we have yet to drive a car that doesn't suffer from fuel starvation (also known as "fuel slosh") in one form or another. This terribly annoying situation is almost always experienced during evasive cornering maneuvers while driving with a relatively low fuel tank. This happens when the fuel in the tank gets plastered against the walls, leaving the fuel pick-up momentarily dry. Without fuel, the engine sputters, shudders and falls flat on its face. Of course, this usually happens at the most inopportune time (e.g. while powering out of a turn).

For example, SCC's monster-powered Project RX-7 suffers from fuel slosh while making hard left turns with anything less than one-third tank of fuel. As bad as this is, it's nothing compared with our Subaru slosh-meister which runs dry during hard right-hand turns with anything less than one-half tank. One-half tank? Yes, one-half tank. Simply unacceptable. Besides being terribly frustrating and joy-killing, fuel slosh can also be quite dangerous as it tends to "bubble" the fuel right before it gets sucked up by the pump's pick-up. With tiny air bubbles in the fuel line, it's possible for the engine to run dangerously lean. As one can imagine, fuel slosh should best be avoided with high-output turbocharged powerplants. Fortunately, due to the ubiquitous nature of this problem, there are readily available solutions. One such fix would be to fill our fuel tank with fuel cell foam. This low-density foam is like a giant in-tank sponge designed to keep the fuel in one place. The downside is obvious; you give up a little fuel capacity due to the space taken up by the foam material. The upside is that this modification is easy and really quite inexpensive.

Another slightly more upscale solution would be to install a smaller-capacity supplementary fuel tank (also known as a "surge tank") with an auxiliary fuel pump between the main tank and the engine's fuel rail. Acting as an emergency fuel reservoir, this always-full surge tank ensures that the engine's fuel needs will always be met--even when the fuel in the main tank is getting sloshed around like free-range chickens in the middle of a hurricane. By our next installment, we will have implemented at least one of these two solutions.

Fuel Flow Limitations
As recently measured on the injector flow bench at RC Engineering, our stock fuel injectors flowed approximately 280 cubic centimeters per minute. This is 40 cc per minute less than we were led to believe by listening to other third-party sources. Serves us right for not verifying our information. This means that our fuel injectors aren't as oversized as we once thought. To get a good idea of what horsepower levels these injectors are designed to support, let's do some simple injector math.

The maximum horsepower that a fuel injector can support can be calculated by multiplying injector size (pounds of fuel/hour) by maximum duty cycle (%) divided by Brake Specific Fuel Consumption. (To convert 280 cc/minute into pounds/hour, we divide by a factor of 10.5 and come up with 26.67 lbs/hour.)

As an industry standard, maximum desired duty cycle should be 80 percent. "Why?" you ask? As one would expect from any electromagnetic solenoid, there are mechanical limitations to how quickly an injector can open and close. Most fuel injectors, when operated above 80 percent duty cycle, tend to offer erratic performance, staying open when they should be closed or visa versa. This isn't a good thing. First of all, this behavior can overheat the mechanics of the injector, leading to compromised injector life. Second, an "over-performing" injector squirts fuel even during the exhaust stroke, as the intake valve is completely closed. Unable to shoot directly into the cylinder, the fuel accumulates on the back of the closed intake valve. Although this "puddled" fuel will eventually find its way into the cylinder during the next intake stroke, it will not be optimally atomized or evenly dispersed. That's why.

Brake Specific Fuel Consumption (or B.S.F.C. for short) is defined as the amount of fuel (in pounds per hour) that a given engine will consume to generate one horsepower. Of course, this number will vary from engine to engine. Typically, large displacement, low output, naturally aspirated engines, like the stock EJ25, will often make maximum power when operating with a B.S.F.C of 0.50. High output turbo engines, on the other hand, will often need to operate at or beyond 0.60.

Plugging these Numbers into the Equation
Maximum Horsepower (per injector) equals 26.67 lbs/hour multiplied by 0.80 divided by 0.50, or 36 hp per injector. Multiply by four, and we come up with 170 hp. We can only assume that the engineers at Subaru did the same arithmetic when deciding upon injector size.

So how were we able to extract a dyno-verified 250-260 hp while running stock injectors that were only designed to support 170 hp? Simple. We took advantage of the injectors' mechanical inefficiency. Forced to operate well beyond 80 percent duty cycle, the injectors remained static. That is, they were constantly dumped fuel--even during the exhaust stroke. Puddle, baby, puddle! Not pretty, but it worked. And at an estimated 280-300 hp, it is still working. All we needed was an additional 30 psi of fuel rail pressure and a little luck (which we think we are pushing).

Fuel Stink
That's right. During sustained periods of high boost, Project Impreza's cabin gets saturated with the noxious aroma of raw fuel. For weeks, we tried to find the location of the source of this frightening stink. Like troopers, we spent countless man-hours under, over and inside the car, searching for loose or ruptured fuel lines. Unable to find even the slightest hint of fuel leakage, we diverted our attention elsewhere--to the charcoal canister purge system, in particular.

Could positive manifold pressure (something a stock 2.5RS was never designed for) be pressurizing the canister, causing it to leak gasoline fumes into the cabin? To test our theory, we installed a one-way check valve into the manifold vacuum purge line just behind the intake manifold. Simply put, this little trinket keeps the boost from blowing into the charcoal canister while still allowing the purge system to function as designed. Eureka! With the check valve installed, we're happy to announce that the fuel stink is gone. Our remaining brain cells couldn't be happier.

Ignition Limitations
Misfire is a funny thing. Often, one doesn't know it exists until it's gone, or at least reduced in severity. Although the stock ignition system appeared to be perfectly adequate for our boosted needs, we decided to replace the stock NGK PFR5B-11 platinum tip plugs with HKS Super Fire Racing plugs. Two steps colder than the stock plugs, the HKS replacements certainly couldn't hurt performance.

Almost immediately, we noticed a difference. Not just in high-rpm smoothness (and perhaps a little bit of power), but in air/fuel ratio (as displayed on our J&S Electronics dual monitor). With the HKS plugs, the engine was running slightly leaner--a sure sign of improved combustion! While lean is not good, it does lend support to HKS's claim that their iridium core spark plugs outperform typical platinum plugs in high-output applications.

According to HKS, the melting point of iridium is a whopping 2454 degrees C, nearly 700 degrees C higher than that of platinum. This ability to withstand heat greatly reduces the chance of spark plug damage during sustained racing conditions. Further performance improvements are afforded by the size and shape of the center electrode. Made of this ultra-hard and durable iridium, the center electrode can be machined down to a tiny pin-head, just 0.4mm in diameter. How does this translate to more spark? Simply put, the skinnier the electrode, the less voltage required to generate a spark. And the less voltage required to make a spark, the less chance of misfire under high cylinder pressures.

Boost Control
Or lack thereof. Right about now is the best time to point out that the Minnam turbo kit was designed to run between 5 and 8 psi of boost and was fitted with an appropriately sprung Turbonetics Deltagate external wastegate. However, as boost went up and time went by, it became clear the wastegate needed a little help maintaining consistent boost pressures. We noticed maximum boost pressures would fluctuate by as much as 2 psi depending on the gear (higher boost in higher gears). While such boost creep is not atypical for a mechanically actuated wastegate, it can be risky when running at the edge of our fuel system's capacity. With no fuel to spare, we cannot afford to deal with even the slightest over-boost. The solution? An electronic boost controller, of course.

Enter HKS's Electronic Valve Controller (EVC) IV. As one of the more sophisticated electronic boost control devices on the market, the EVC IV employs something known as "fuzzy logic" to learn and compensate for a given wastegate's mechanical inefficiencies and non-linearities. Not only does this nifty feature eliminate boost creep, it also contributes to improved spool-up since the smart electronic controller knows exactly when to start pulling the reins on the wastegate. As with most electronic boost control systems, the EVC IV consists of a cockpit-mounted controller and an underhood-mounted plastic solenoid box. To install, we mounted the solenoid box on the passenger-side fender well, just next to our fuel pressure regulator. With the supplied heavy-duty vacuum line, we re-plumbed the wastegate signal line, which was originally routed between the compressor outlet and Deltagate wastegate. A manifold pressure signal was also fed into the solenoid box. With the compressor outlet line intercepted and manifold pressure carefully monitored, the EVC IV actively controls the wastegate, keeping it completely closed until desired boost pressure is reached.

Once target boost pressures are achieved, the EVC IV maintains boost pressure by constantly manipulating the wastegate signal. Of course, this talent does not come naturally. It must be learned. And that is where the HKS instruction manual comes into play. If you're the type of person who prides yourself on figuring things out through trial and error, think again. You will need to read the manual. Without it, the EVC IV is no more than a pretty little device that beeps merrily as you power-on your car. But used to its full abilities, the EVC IV proves to be a truly impressive boost controller packed with useful and innovative features. Among such features are two boost presets (high and low), a manual over-ride, a user-definable, short-term over-boost, an over-boost warning and a digital boost gauge.

After an hour of head-scratching and button-fumbling (Hey, we're not used to reading manuals!), we successfully dialed-in the EVC IV. With low and high boost settings at 8 and 10 psi, respectively; we were astounded with the improvements in partial throttle boost response. Crack open the throttle gently and WHOOOSH!!!... full boost before you can say, "Ouch, my aching neck." Fully spooled just above 3000 rpm, Project Impreza assumed an almost sinister disposition. Gone are the days when we could tool indiscriminately with an insensitive throttle foot over broken pavement. Now, each bump-induced micro-movement in the throttle plate sends the boost gauge needle screaming towards the big numbers. Hardly subtle...but as effective as a sledgehammer. Most impressive was the fact that despite the spastic boost rise, we could not get the EVC to loosen its firm grip on wastegate control. Never once did we notice even the slightest hint of boost spike or boost creep. It's almost as if there was a invisible mechanical stopper in the face of our boost gauge, which kept the needle from rocketing to the far ends of the dial. Simply amazing.

Unfortunately, such wonderful, gut-twisting boost response is not without its potentially deleterious side effects. First of all, to say that it takes more effort and restraint to drive the car smoothly would be a gross understatement. Second, it's clear that such immediate boost response demands better fuel and ignition management. Before we installed our fuzzy logic boost controller, boost pressures would rise slowly enough (relatively speaking, of course) that our mechanical rising-rate fuel pressure could stay on top of things, providing just enough fuel in just the necessary amount of time. With near-instantaneous boost response (from 3500 rpm and up), we began to notice a subtle, but unnerving, fuel pressure "lag." This transient lean spot would send our J&S Safeguard knock sensor into action, retarding several degrees of timing for a fraction of a second until maximum fuel pressure was restored. While detonation is safely averted, we can't help but be a little concerned about the long-term viability of our current fuel management system. And that brings us to the next topic of discussion.

Engine Management:
in Retrospect
As hokey as it sounds, our Project Impreza represents the typical bells and whistles "tuner" car. Sure, it works. But a quick glance under the hood or under the ECU cover would reveal several points of possible failure. Between loosening splices, fraying wires, potential shorts and a fuel and ignition management system that operates at the ragged edge of safety, we're not entirely confident that our car will exhibit the exemplary reliability and trouble-free nature which we would expect from a real WRX. What separates our project car from a WRX is proper engine management. And simply put, we don't have it. In order to get it, we could have to invest in a sophisticated stand-alone programmable EFI system. Not only would this approach render all our electronic piggy-back computers and controllers obsolete, it would also afford us the ability to eliminate every single glitch or trouble spot associated with our current state of engine tuning. Of course, a complete programmable engine management isn't exactly chump change. But then again, neither is the "add-on as you go" approach which we've been following up until now. Don't believe us? We tallied up the approximate prices of the components which would probably be replaced if we were to install a full-featured programmable EFI system. (See the table.)

Twenty-three hundred dollars? Surely, for that kind of coin, we can put together one helluva butt-kicking, high-performance, custom EFI system. And with the help of Electromotive Inc. and RC Engineering, that's exactly what we are going to do in the next installment. We'll have it all: Laptop programmability, straight to hard-drive data logging, upgraded ignition coils, four monstrous 550 cc/min low resistance fuel injectors, full rpm-dependent boost control, programmable knock sensing and much more. Stay tuned...