Staged Upgrades

1. Datalogger. The absolute first step is to enable engine monitoring. The MirageCorp Hybrid Datalogger consists of a special cable and PalmOS software that reads most ECU parameters including knock count. For OBD-I models, Dmitry Yurtaev's open-source (GPL license) MMCd Datalogger for PalmOS is a flexible option.
2. Tuneup and pressure test. Modifying an engine that is in less than optimal condition will be at best disappointing and at worst disasterous. Use good wires and plugs gapped to .034 in anticipation of higher boost. Several pressure testers are available that help identify induction leaks.
3. Air induction system. A popular CARB-approved option is the K&N FIPK, which reduces intake restrictions associated with the stock air box. Drawbacks include more engine noise (you can clearly hear the turbos spool up and the bypass valve actuate) and an annoying intake resonance (commonly described as a "hooting owl") that can be eliminated only by replacing the stock bypass valve. Also, the stock air box snout sources air from a relatively cool section of the engine bay that is inaccessible to open-element filters, potentially negating the restriction advantages.
4. Boost controller. An electronic boost controller (EBC) offers features such as multiple boost settings, RPM-based adjustment, and faster solenoids that enhance wastegate control. A cheaper but less-capable alternative is a manual boost controller, which is essentially a bleeder valve that reduces the amount of boost measured by the stock pressure sensor (so that it will leave the wastegates closed longer to build more boost). Drawbacks include the temptation to crank up the boost too far; keep boost under 1 bar (14.7 psi) to prevent detonation and fuel cut (where the stock fuel system is maxed out), and datalog religiously!

At this point the car will be generating a reliable 350+ SAE horsepower with 350+ lb-ft of torque (see my dyno results at this stage). Futher modifications can have a significant impact on drivability and reliability.

5. Larger fuel injectors, higher capacity fuel pump, and the appropriate electronics to compensate for these changes.  More power requires more fuel, so a fuel system upgrade is critical before further increasing boost or installing turbos that can hold boost at higher RPMs. Unfortunately, stock drivability is typically sacrificed; tuning is regarded as a black art, and even a well-tuned setup that circumvents the stock ECU may require frequent adjustments to compensate for environmental changes (temperature, humidity, barometric pressure, etc.). Choose an injector size that results in approximately 80% duty cycle (IDC) at maximum horsepower. An excessive increase in injector size can lead to insidious timing problems; when the fuel controller reduces the airflow signal, the ECU instructs the injectors to deliver a proportionately lower volume as desired, but because it computes less load, it also advances timing to the point where detonation can occur at even moderate boost pressures.
6. Larger turbos, more efficient intercoolers, less restrictive exhaust.  The turbos and intercoolers will reduce EGTs and allow higher boost pressures to be maintained without detonation.  An exhaust system is finally justified to handle the significant increase in air flow over stock. There is much debate over which turbos are "best." Brand new 13G (for pump octane fuel) and 15T turbos (for racing octane fuel) are popular street applications, while larger TD04 upgrades are better for the drag strip, producing more horsepower at higher RPMs at the expense of low-end performance (turbo lag means the turbos spool up more slowly). TD05 options are increasingly available and may eventually offer the best of both worlds (even with adapter plates or headers, they are currently far from plug-and-play).
7. Intake injection.  Some highly-modified engines employ systems that use water, propane or other substances to cool the intake charge to avoid detonation at even higher boost pressures. Drawbacks to this approach include 1) the substance is consumed and must be monitored / replaced regularly; 2) if the injection system fails, the results can be catastrophic.

Completed Modifications

• K&N Filter Induction Performance Kit (FIPK) for less restrictive air intake. Side effects included more engine noise (you can clearly hear the turbos spool up and the bypass valve actuate) and an annoying intake resonance (commonly described as a "hooting owl") that I fixed by replacing the stock bypass valve (see below). This modification was probably good for a few horsepower once the boost controller was installed and the engine began inhaling more air, but I did not verify this on the dyno.
• Optima Red Top battery.  This 1050 CA / 830 CCA sealed spiral cell battery can be mounted in any position and relocated to the trunk if desired. It is also small enough to accomodate most strut tower braces (note that modification of the stock retainer bracket was required for installation). I bought this battery when, after six years of faithful service, the original could no longer supply enough juice to prevent spark blowout at high RPMs. What I didn't know at the time was that my alternator was not only failing to charge at full capacity but draining the battery when parked. The Optima was so strong that it put up with this charging system problem for over two years before throwing in the towel on the way to Le Mans in 2003. I had to replace it with a traditional battery in France to get to the race and back home for proper repairs, but I will definitely buy another Optima when the time comes - most impressive!
• A'PEXi Super AVC-R electronic boost controller to safely allow higher boost and produce more power. I like this unit because it is truly "fire and forget:" simply set the boost pressure limit and the boost actuator duty cycle (BADC) and the unit learns the boost curve automatically. Two different limits can be set and toggled between on-the-fly. The display shows current boost pressure and injector duty cycle (IDC) as digits or a bar chart. The boost pressure reading is accurate and eliminates the need for a separate gauge.

  The BADC setting provides the unit with an estimate of solonoid activity necessary to achieve the desired boost limit. If it is set too low, the unit may never achieve the desired boost pressure, regardless of how long it tries to learn. If it is set too high, the desired boost pressure will be exceeded temporarily, and again the unit will have difficulty learning. This is puzzling to me, as the algorithm should be straightforward:
  

  check pressure
    if below threshold then close wastegates
    if above threshold then open wastegates
  repeat
  

  Another annoyance is that this unit does not work at high altitudes. When driving in the Alps, the unit shuts itself off above 5-6k feet, limiting boost to about .5 kg/cm^2. A'PEXi customer support has verified this behavior but offers no solution. Now that I live in the UK, this flaw is not a concern, but replacement will be necessary if I move back to higher ground.

  If you're in the market for a boost controller, the latest offerings from A'PEXi and Blitz are much more flexible, but with capability comes complexity and some might find them a bit harder to set up properly. At any rate, a boost controller represents the absolute best bang for the buck investment, but if you're not careful, BANG is exactly what your engine will do!

• A'PEXi turbo timer to keep the engine running until the turbochargers cool down.  I would have preferred a turbo cool-down device that was temperature based, but such devices are too expensive.  Blitz offers a model that computes cool-off time by monitoring boost activity before the ignition is switched off, a nifty idea. My unit is basic and can only be set to keep the engine running for a specified time period, which works fine.  I normally leave it set for 30 seconds but bump it up to several minutes after an exceptionally hard drive.

  Note that it is tricky to wire a turbo-timer without disabling convenience functions such as passive alarm activation, keyless entry, automatic headlight shutoff, power antenna retraction, etc. Eric Gross offers some good tips on his Turbo Timer Installation web page for those interested in this modification.

• Redline Shockproof fluids for smoother shifting and less drivetrain loss.  I have MT-90 in the transmission and transfer case and ShockProof Heavy in the rear differential (some prefer a 50/50 blend in the tranny and xfer case). The difference from stock was subtle at best. Some have expressed concern that these GL-4 fluids may be too aggressive for the brass synchronizer blocking rings, but after discussions with Redline, the community is confident that this is not an issue. After 50k miles with these fluids, I have observed no abnormal or premature wear.
• Magnecore wires for better spark.  I had these installed with the first major service at 50k miles. I noticed no difference over the stock wires.
• Spark plug cover plate made by EK2MFG and finished by Dave Best. This is just something neat that some fellow Team3S members designed and sold.
• When it comes to upgrading the 2G brakes, the best bang for the buck is Porterfield cryo-treated rotors (solid, not cross-drilled!) with Porterfield R4-S pads all around for the street and R4 pads up front for the track (see previous mods below), which served me well for years of driving on the Autobahn in Germany and on the track in the UK. After abnormal brake failure at Donington Park and a sheared stainless steel brake line in the left-rear caliper, I decided to order a four-wheel StopTech BBK from Supercar Engineering. The kit includes massive calipers with custom-sized pistons (to balance front-rear bias) squeezing cryoed 332x32 and 322x28 rotors on floating aluminum hats, the rear hats having integrated parking brake drums. Titanium backing plates protect the caliper piston seals, and stainless steel lines maintain pedal feel when the fluid gets hot. The pads are a standard size used by Brembo, Porsche and others, so options are plentiful and reasonably priced. My kit served as a prototype for StopTech's official off-the-shelf front kits for the Stealth / 3000GT platform and for a retail four-wheel kit from Supercar Engineering.
  

  Here are a few pictures comparing the rotors and pads with stock 2nd gen. Here are a few installation notes. Based on my experience with this kit, here are a few issues to consider when deciding about brake upgrades.

• While the stock 2nd gen 17x8.5 wheels accommodate the brake upgrade, a second set of rims eliminates the need to remount tires for track events. Though it is easier to find 18" rims with the necessary caliper/spoke clearance, fewer track tire options, higher cost and compromised handling and durability make 17" a better choice. I decided on "hyperblack" Enkei RPM2 17X9 5-114.3 40mm 18.7 lb rims, which fit perfectly at all four corners without spacers. The Miata.net Tire Size Calculator is very useful when investigating the impact of changing tire size.
  

• With two sets of rims, it is no longer necessary to compromise with all-season radials. An excellent combination is Yokohama ES100 245/45WR17 Ultra High-Performance Radials during the summer and Hankook Icebear W300 235/45R17 Winter Radials during the winter, which handily outperform Ultra High-Performance All Season Radials for about the same price. At the track I run Yokohama A-032R 275x40R-17 H-compound competition tires.
• At higher boost levels, the stock plastic Y-pipe has a tendency to blow off off the throttle body, particluarly as the rubber gasket ages. The gasket is not sold separately, so most owners replace the entire Y-pipe with something less failure-prone after tiring of "the tennis ball mod." Several aluminum options are available on Ebay for very reasonable prices, but reported problems include lack of rolled flanges, improperly sized IAC nipple, flimsy materials that crush when the clamps are tightened down and weld burrs that could be ingested by the engine, potentially causing seious damage. I opted to go with an IPS Stainless Steel Y-Pipe powdercoated black with stock BPV flange, T-bolt clamps and reinforced silicone reducer. It is a beautiful piece!
• Many folks feel that the stock exhaust system is too restrictive and should be replaced with an aftermarket alterative, but dyno testing indicates that the stock exhaust actually flows quite well and that such an investment is not necessary unless boost is increased drastically.  The biggest restrictions are the precats, the main cat and the downpipe, in that order, and replacing each is necessary for the others to make much of an impact. When my stock downpipe developed a leak in the flex section, I replaced it with an EK2MFG downpipe that offers two 2.5" pipes mandrel-bent into a 3" pipe for improved flow over the stock tee connection. It bolts up to all three cats for full emissions control or to precat eliminators and a test pipe for the track. Other options are reviewed in John Monnin's downpipe comparison; while all improve upon the stock tee connection, most eliminate the front precat (emissions risk) and lack one or both flex sections (complicates installation and could crack headers as engine mounts age). For more information, take a look at Erik Gross' page.
  

  

  The stock cat-back system flows nearly as well as aftermarket systems, and the selectable exhaust valving becomes increasingly important for volume control as restrictions upstream (i.e.- cats) are eliminated. I retained the stock system until it rusted where the pipe joins the muffler and spent considerable time researching replacement options. Aftermarket systems are available in two configurations, one similar to stock where exhaust enters a common muffler on the left side before being split and routed to the right side via a crossover pipe, and one where exhaust is split at the differential and routed to both sides equally. There is room for a proper muffler only on the left side, so systems of the latter type tend to use canisters or undersized mufflers that are unable to suppress an annoying drone/resonance in the cockpit between 2700 and 3300 RPMs where the car is driven most often (it's possible that side branch resonators could address this). None of the aftermarket systems replicate the active exhaust function, though the A'PEXi Exhaust Control Valve (ECV) has potential.

  • The HKS system looks virtually identical to stock (quad tips) and has only a slight resonance but reportedly suffers from fitment and corrosion problems.
  • The IPS Quad-Tip system also looks very similar to stock and flows very well but is extremely loud.

  • The ATR system has the stock configuration, good fitment and an aggressive sound but is loud and has large, round tips.
  • The Borla system is relatively cheap and sounds good outside the car but has a severe drone. Borla customer service provides inserts to address the irritating drone (at the expense of flow), but owners report that they are mostly ineffective.
  • The Greddy system flows well, sounds good and has no resonance but exits on only one side of the car (aesthetically not as pleasing).
  • The Tanabe system flows well, sounds good and has only a slight resonance but its large, round tips protrude too far beyond the rear bumper.
  I went with a stainless steel system custom fabricated by Zorstec. It employs 3" mandrel-bent pipe with a long, 4" diameter, ceramic-baffled resonator between the main cat and a beautifully crafted splitter feeding identical mufflers on both sides and exiting unique dual-pipe oval tips.
  
  
  

  Here are some sound clips:

  • stock exhaust (29k)
  • Zorstec exhaust (coming soon!)
  • unrestricted exhaust (27k)

• Engine monitoring is critical for tuning and general diagnostics, and while various data loggers for OBD-I models (1991-1993) have been available over the years, no solution was offered for the hybrid ECU models (1994-1995) until 2004 when Mirage Computers, Inc. released their offering for Palm M-series PDAs. It works great with my m515 and provides an indispensable look at engine behavior (including boost by connecting the output from the SAVC-R's 3-bar MAP sensor to the EGR-T lead). When I ditched Windows in favor of Linux, Chris Dooley provided the info I needed to develop a Perl translator for PDB -> CSV files. If you have a 1st gen car with OBD-I ECU, check out Dmitry Yurtaev's open-source (GPL license) Palm data logger. What I'd really like is a configurable dash that would allow me to decide what to monitor and how.
  
• The American-spec cars have injectors that max out at around 6000 rpm when running 1.00 kg/cm^2 of boost. The stock fuel pump also runs at a reduced voltage at low RPMs, which may contribute to detonation issues in that range when running higher boost. Though I observed acceptable air/fuel ratios during my dyno session (i.e.- the mixture is not leaning out and causing a dangerous situation), I installed a Denso Supra fuel pump with Hotwire kit that provides full voltage at all times. Since the increase in volume would overrun the stock fuel pressure regulator, I also installed an adjustable FPR. This sets the stage for upgrading from the stock 360cc injectors to DSM 450cc black top injectors to support 13G turbos.

  NOTE: When hotwiring the fuel pump, be sure to replace the stock voltage relay with a jumper (between pins 2 and 3) so that noise is not backfed to the ECU, causing garbage data in logs (potential fix here) and potentially other anomalies!

  
• Proper mixture is achieved by cycling the injectors by an amount appropriate for the measured airflow. Raising boost increases airflow which the ECU maps to higher duty cycles; excessive boost can lead to "fuel cut," where the ECU determines that injector flow at 100% duty cycle is insufficient and abruptly disables the injectors to prevent a lean condition that could damage the engine. The ECU also calculates ignition timing from airflow, advancing it to improve responsiveness and retarding it to prevent detonation ("knock").

  With stock 9B turbos limited to stock boost pressure, maximum airflow as the engine generates 320 flywheel horsepower results in IDCs of about 80% and timing of around 28 degrees. With stock 9B turbos limited to 1 kg/cm^2 (14.223 psi) pressure (which they cannot sustain at redline), maximum airflow as the engine generates 367 flywheel horsepower results in IDCs of approximately 97% and timing of around 24 degrees. Note that IDC values reported by loggers should be viewed with suspicion; accuracy can be verified by computing duty cycle from pulse width as IDC = RPM / 1200 * IPW.  Also, the ECU reports timing 10 degrees higher than actual (explanation here).

  Higher-capacity injectors deliver more fuel per cycle, rendering the rates computed by the ECU inappropriate and resulting in an extremely rich condition. An air - fuel piggyback controller (e.g.- A'PEXi AFC) effectively decreases the airflow signal by an amount proportional to the increase in injector size so that a lower duty cycle is mapped by the ECU. Unfortunately, the low apparent load results in timing too advanced for the high actual load, leading to detonation at otherwise acceptable boost pressures. Piggyback controllers with timing control are available (ITC, eManage, AEM) but tuning becomes more complex. Injector capacity should be increased only enough to achieve the maximum fuel delivery needed for the maximum anticipated airflow (more here and here).

  Unlike 9B turbos, 13G turbos can sustain well over 1 kg/cm^2 (14.223 psi) to redline, resulting in a significant increase in airflow that necessitates 450cc injectors. Even larger turbos coupled with high-octane racing fuel and other modifications that suppress detonation can sustain higher pressures yet, increasing airflow to the point where 550cc or larger injectors are required. For this platform, a general rule of thumb is injector size * maximum observed duty cycle = potential flywheel horsepower, but more precise calculations can be carried out on Jeff's site.

  After nearly 10 years of ownership, I finally overcame my fear of tuning and installed an air/fuel controller and black top injectors (cleaned and flow tested). The consensus is that 450cc injectors require no timing control, and I'm not out to corral every last horse, so I decided to go "old school" and get an A'PEXi AFC. I looked at the AFC NEO ($299 new - compatibility warning here) but decided to start with the old 5-knob version ($70 used). It was easy to wire in and introduced no undesirable behavior when zeroed out with my stock 360cc injectors. Jeff's injector swap procedure was indispensable, and the four-hour estimate is about right. I fumbled a rear plenum bolt and the EGR gasket (which I just couldn't get lined up), both now irretrievably lost somewhere in the depths of my engine bay, but otherwise things went smoothly.

  As for airflow correction, 360/450 - 1 = -20% is just right when the AFPR is set to stock pressure and with no intake leaks revealed by pressure testing. All three long-term fuel trims hover right around 100, timing is as expected according to Matt's map, idle is perfect, open loop NBO2s are about where they were before (.90-.94), there is no knock under load, and max IDCs (IPW*RPM/1200) are in the high 70s and low 80s (vs. very high 90s with stock injectors).

• Armed with ample fuel and a good exhaust, the next step was a turbo upgrade. The stock 9B turbos are very responsive but cannot sustain more than about 11psi to the 7200 RPM redline. Larger turbos can sustain higher boost pressures at high RPMs but at the expense of lag. In my case, the objective is good low-end response and at least 1 kg/cm^2 to redline, suggesting 13 class turbos as most appropriate. Many 13 class turbos are modified 9Bs with clipped TD04 turbines and housings bored out to accept 13G or 13C compressor wheels, and DSM TD04L-13Ts can be modified to fit the 3S platform, but I opted to go with two factory-fresh (unmodified) TD04L-13G from MHI with the larger turbines which were installed by Sumiyaka in February 2008.

  Below are thumbnail links to two of the turbo compressor flow map comparisons I posted to the 3SI wiki to demonstrate the difference between 9B and 13G compressors (and 360cc and 450cc injectors). The Y axis represents boost and the X axis represents flow. A horizontal line intersecting the two green dots in the first chart would represent 17 psi, and the amount of air most 3S engines demand at 7000 RPM falls somewhere between them. Anything above (higher boost) or to the right (higher volumetric efficiency) of an engine's "green dot" is unused capacity. A larger compressor will offer higher efficiency at maximum boost and RPM (drag racing) but typically lower efficiency at lower boost and RPM (daily driving).

  

  

• High Intensity Discharge (HID) lights are significantly whiter and brighter than the stock halogens, particularly important when driving through the pitch-black Yorkshire winters. I went with a Mcculloch kit, which included ballasts, ignitors, wiring compatible with the stock plugs and 5300K 9006 bulbs. Plug-n-play! I discourage use of HID temperatures greater than 6000K, which are too dim.

Planned Modifications

• Folks with substantial engine modifications complain that the stock clutch slips under hard acceleration.  Some aftermarket clutches offer more holding power, but usually at the expense of drivability, transmission load and clutch longevity.  With my more standard mods, the stock clutch lasted just over 100k miles and worked well under all but the most aggressive launch conditions, so I replaced it with stock parts.  Here's a page showing clutch, synchro and seal parts for the Getrag W6MG1 (which Dodge and Mitsu claim has no user-serviceable parts and must be replaced in its entirety!).

• One of the biggest problems to overcome with these cars is getting cool outside air to the intake.  The engine compartment is so tightly sealed that the engine struggles to draw in even the hot air available to it.  It is amazing how much power increases when the car is driven with the hood popped.  I have some ideas on how one might design a custom cool air intake system and I hope someone markets an effective solution.

Previous Modifications

• Michelin Pilot XGT-Z4 245/45ZR17 Ultra High-Performance All Season tires. In general, these are an improvement over the Pirelli P7000 they replaced; the Pirellis were a little better on dry tarmac, but the Michelins wear longer and stick better in adverse conditions, important while living in Europe. I went through two sets, the first lasting 50k street miles and the second 25k street and track miles (two track sessions, one dry and abrasive and one wet and slick). This model has been replaced by the more expensive ($50 more per tire) Pilot Sport A/S, though the XGT-Z4 was still available several months after the new ones were on the market.

• Bosch winged wipers to keep the blades from bowing away from the windshield at higher speeds on the Autobahn.  I actually am not too happy with these blades.  The force of the wind on the blade wing makes it very difficult for the wiper motors to return the blades to rest position, and they don't seem to wipe the water from the windshield that much better than regular blades.

• FIRST: Blitz Super-Sound blowoff valve. In my case, the primary reason for replacing the stock BPV was to eliminate the annoying intake resonance that surfaced after installation of the FIPK (see above). Unfortunately, venting metered air to the atmosphere causes an overly rich mixture during activation, felt as a slight stumble shorly after letting off the throttle, which IMO is unacceptable.

• SECOND: Greddy Type-S Bypass Valve. The stock bypass valve is sufficient for normal levels of boost (up to 1.1 bar or so) but is engineered with a slight leak to accommodate limitations of early wastegate solenoids. The Type-S theoretically enables the system to pressurize more quickly and better maintain pressure at higher RPMs while reducing compressor surge between shifts, but in practice the impact is not obvious. I reverted to the stock BPV when the Typs-S developed a huge leak from the uncapped lower activation assist nipple, which is the most common problem with this model. I also tired of the sound it made.

  Here is helpful guidance for adjusting the Greddy Type-S BPV for Mitsubishi engines.
  Here is a sound clip of my Greddy Type-S BPV under various conditions.

• Custom stainless steel braided brake lines for safety and to eliminate brake sponginess.  These were custom made for Mikael in Sweden, Roger in Switzerland, and I in Germany.  I couldn't feel much of a difference on the street after installation but am confident they help maintain the firm pedal feel I enjoy at the track despite elevated brake fluid temperatures.  Also, there is very little chance of failure with lines like this. Note: I subsequently used one of these lines as a replacement for a failed clutch hose near the slave cylinder - same threading, same length, works perfectly!  Thanks to Philip at SCE for the suggestion.
  

• FIRST: Stillen Metal Matrix brake pads for better braking and less brake fade.  These pads wear faster than stock and create more dust on the wheels but improve braking from high speeds on the Autobahn.  The stock rotors continued to warp with these pads.

• SECOND: Porterfield R4-S pads with cryo-treated discs (Brembo blanks) for better braking and less brake fade with no warping!! These pads are slightly less effective than stock until warmed up, but then they grab much harder. Moderate dust. These pads were not effective at track temperatures; I had much better luck with R4 pads up front and R4-S pads out back. I had the calipers stripped and painted when the rotors were installed. This was not cheap, but it really cleaned them up and I think they look terrific.