Cold Air Intake

First and easiest modification beginners do to their cars is ditching stock snorkel box and installing cone filter. Actually stock filter is not so bad for stock configuration, it's main advantage is that it sucks colder air from under the headlight. When you start making serious power then you need higher flowing cone filter. Stay away from sponge element filters (HKS and Greddy). Buy reputable name cotton gauze filter - Apex, K&N and Blitz are best filters on the market. They are close in filtration quality to OEM paper filters. Their biggest drawback is that they suck hot air from engine bay. I remember reading somewhere that generally as the rule of the thumb 6deg.C (11 deg.F) air charge temperature drop equal to 1% of engine power gain. Maybe these figures are too optimistic but they make sense - more oxygen engine digests for combustion - more power it makes. Colder air is denser, hence it contains more oxygen. I placed my filter under stock pop-up headlight. My stock headlights are gone and there is more than enough room there. Filter separated from engine bay by plastic partition sheet (not fixed yet on this picture) and sucks nice cold air. Pipe is made of PVC tubing wrapped in insulation sheet to cover it ugly grey appearance. This setup is very light, pipe together with filter is lighter than pretty looking powdercoated short intake pipe which came with my SP turbo kit.

Intercooler and Pipes

Stock intercooler of mk3 Supra is quite pathetic (like most of stock IC of turbo cars) and IC pipes of 7MGTE Supra are very long and restrictive in some places. It causes slow throttle response and another bad aspect is cold air gets hotter passing through the pipes routed over the hot turbo and cylinder head to the throttle body. You can roughly calculate how increased IC pipes and IC itself volume contribute to turbo lag using formula.
Time = (V / flow rate)*2
Where V = total internal volume of IC and pipes
Flow rate = engine flow rate at given rpm and VE
The factor of 2 results from the approximate doubling of airflow when going from cruise to boost
Finally one more bad thing is stock AFM (air flow meter) located very far from TB (throttle body). Opening the throttle causes negative pressure pulse travel all the way to AFM through long IC pipes and it takes time. MAP based ECU definitely wins here, less laggy.
Previously I had Spearco stock replacement intercooler. No complains about it, it flows 1080cfm, should be good for 600-700bhp. But since I became power greedy I decided to order mk4 Greddy 3-row IC kit. It is supposed to support 900hp, I will never outgrow it. Kit comes with aluminium hard pipes - 2 elbows size 70mm between compressor and IC, all other elbows size 80mm. Also included numerous bolts and brackets (not needed on mk3), blue silicone couplers and 80 to 70mm silicone reducer, small plastic coolant overflow reservoir with embedded Greddy logo on it and aluminium coil - from enclosed Japanese instruction I understood it was supposed to replace stock mk4 power steering cooler. And, of course, kit included huge 600 x 300 x 115mm core intercooler. Greddy couplers look thin and weak, I am going to replace them by more reliable 4-ply couplers. Hard pipes are not plug'n'play affair for mk3, first I wanted to route them from compressor outlet down along the engine block to IC inlet but after discussing it with my friend Victor (nickname SOARA, because he drives 1JZ Soarer with T78 turbo. Now he is in the process of swapping 1JZ for  2JZ and T88) we decided to route pipes around strut tower. SOARA did all the custom job of cutting and slicing pipes from the kit to suit my configuration. After welding pieces together they fitted flawlessly. There were 4 short pieces between compressor and IC and one pipe from IC to TB. Biggest advantage of this setup is that I can always keep my eye on hose clamps tightness - compressor outlet coupler is notorious for blowing at high boost. Don't ask me how I know... It's too bad to be stranded because of the  blown coupler without proper tools to fix it. Also proximity to exhaust manifold if pipe routed down will not make coupler's life more comfortable.
I use Tial 50mm blow-off valve to release boost pressure and make turbo's life easier. It fllows a lot of air and has very competitive price of $200-220.

Throttle Body and Intake Manifold

The idea of getting rid of stock intake manifold was very appealing to me since I first saw it implemented on other people 7MGTE. There are 3 main reasons to go this way:
 - shorter intercooler pipes, less turbo lag, more responsive engine
 - less heat absorbed by pipes between IC and TB, stock configuration pipes have to cross engine bay
 - possibility to increase flow when going custom manifold.

By stroke of luck one local mk4 guy (R.I.P. Joker) purchased at the same time brand new JUN manifold and SOARA was able to borrow and take all dimensions from it. So, our manifolds are shameless copies of JUN. Lower OEM runners and flange were used for this work, other parts are custom made. The work in process pictures you can see here. Write up is in Russian, but you get the idea how it was done. My manifold has no provision for ISC valve since I was going AEM EMS and decided to keep things simple and set ignition and fuel tables to control idle speed. It works, now my idle is relatively steady around 950rpm. Cold start without ISC valve was quite problematic, in the beginning after starting motor in the morning I had to hold gas pedal for 1-2 minutes not to let the motor die. Later I wired adjustable VSV controlled by AEM LS driver to bypass throttle plate and give motor some more air for cold starts.
When I was swapping manifolds I decided to do old hot rodders trick and fabricate phenolic thermal insulating gasket. Thickness of the gasket is 4 mm, it does not require elongated studs or bolts when fitting manifold. I also used stainless steel bolts, they transfer less heat to compare with mild steel bolts. Phenolic is very fragile material, I spent 4 or 5 hours to fabricate it using old OEM gasket as template. Then phenolic gasket has to be glued with RTV to manifold flange. Next day it can be port matched to manifold. Attention!!! If you are going to use this method, do not apply too thick film of RTV or it will protrude into the ports when you bolt manifold to the head and compress gasket. Now apply another thin film of RTV onto the another side of gasket and install it when RTV dries up. In my opinion phenolic gasket helped a lot in reducing manifold temperature, it gets hot from radiated heat only in stop-and-go traffic.
Now I needed bigger throttle body to match intercooler pipe size. I could not find big TB for reasonable price here in Ukraine, tried to search various dismantlers web sites in States to no avail. Again my friend SOARA came to rescue and fabricated custom 76mm TB. It utilized stock TPS and best thing - cost me one third of the price of Q45 TB.
I installed manifold and AEM EMS at the same time and after sorting out issues with AEM (I was EMS tuning newbie, car had horrible driveability and gas mileage, but this is another story) I started paying closer attention to air distribution among cylinders. From spark plugs reading found that cylinders 1, 5, 6 run slightly on the rich side, blamed it on intake manifold and decided to reduce fuel for these injectors. When I opened AEM individual cylinders fuel trim tables I was pleasantly surprised - injectors of cylinders 1 and 6 already had +3.1% fuel trim and injector 6 had +1.6% trim. Seems like AEM engineers when developed 7MGTE startup calibration for stock motor implemented this fuel trim to compensate for unequal air distribution of stock intake manifold. I zeroed out this fuel trim and spark plugs read good air charge distribution since then.
 

Methanol / Water Injection

power from boost control solenoid or use it for any other action with programmable EMS (like dumping additional fuel for cooling or reducing spark advance angle).
I installed my pump under the passenger seat and methanol/water tank behind the same seat. I want to keep as less weight on the front of the car as possible. It helps to increase rear tire traction. Another aspect is keeping more of the car mass as close as possible to the centre of gravity helps to reduce polar moment of inertia and makes car easier to steer and control in the corners.
Tank looks ugly, that was the only transparent plastic tank I was able to find. I can always see the liquid level inside without opening filling cup, it was important for me. One more thing I like about my install - I always hear if pump is running or not, Aquamist pump is quite noisy.
Recommended water to fuel injection ratio is 10-12% for intercooled motors, non-intercooled guys tend to inject as much as 25-30% of water, some of it pre-compressor. Much more preferable is progressive water injection, but 2s system from Aquamist is quite expensive and will not support my power goals. It uses cycling high speed valve and optional accumulator. This approach is far superior from my point of view to compare with all other progressive kits on the market - they cycle the pump or change pump supply voltage. Many progressive WI kits use Shurflo pump, it flows much more then Aquamist and compatible with 100% methanol. Now I am considering to buy kit from Derek Devices, it was tested and proved on many high power Supras and works great with AEM EMS.
Usually nobody injects pure water, experiments proved that 50% mix with alcohol is the most effective, water helps to reduce detonation and alcohol reduces intake air temperature and increases slightly octane number. Reduced intake air temperature in turn reduce octane requirements of the motor, MW injection sometimes called poor man's race gas. If your WI system flows too much for the power your motor makes then higher alcohol ratio is better for you. The most popular kinds of alcohol used in WI are methanol (RON 106) and ethanol (RON 105). I would not mess with injecting windshield fluid mix unless you are absolutely confident about chemical components of the fluid.
The most common misconception about alcohol/water injection sounds like:
"We used WI on our car and it did not run any faster" or worse "We used WI on our car and lost power on dyno"  This happens more often when pure water injected. WI users have to understand few basic things:
1. If your motor did not detonate in any given condition and you inject water at the same condition motor will not benefit, most likely you will loose power. You have to use piggyback devices to lean out fuel and increase spark advance angle to take full advantage from WI. Don't forget that water in combustion chamber tends to retard flame development angle and therefore engine needs more spark advance to have PPP (peak pressure position) at about 14-15 deg. ATDC . Also you can step by step increase your boost pressure  to the point where motor would knock without WI. In other words you have to go more aggressive with your fuel, ignition tables  and boost pressure. Programmable EMS is the best here. Most of the modern ECU adopt their fuel and especially ignition tables to octane of the gasoline you use. If they see no detonation from their knock sensor(s) ECU assumes you are on high octane gas and jumps to the more aggressive ignition map. It can happen sooner if you reset your ECU.
2. Don't inject too early. In the beginning I adjusted pressure switch to activate pump at 0.2 bar boost pressure. Car stumbled at pump activation and finally I set it on 0.9 bar. This means with my non-progressive system I was injecting too much water at lower part of my torque graph, more then required 10-15%. In fact detonation is most severe at the peak of motor's torque curve and you have to start WI slightly earlier. Injecting water at too low boost pressure also makes your engine feel laggy, for better spool up you need lean mixture down in your rpm band and hot EGT.
3. You will lose power if inject into the engine running overly rich air/fuel mixture. To be on the safe side most stock ECU dump a lot of fuel under boost for additional cooling , and motors run in the range of 10.0-11.0 AFR. Now, if you are using WI you don't need that much fuel and can lean it out safely to 11.7-12.0 AFR or leaner. I run 12.2-12.4 AFR. Also remember about the fact that on most engines running speed/density  air temperature sensor located in the intake manifold or before the TB. Thus air temperature reading will be affected by WI. and ECU will see it as colder air ingested and add additional fuel to keep required AFR. Considering this fact if your system flexible enough it makes more sense to relocate air temperature sensor upstream of the MW nozzle.
Recently I found interesting charts showing cooling capacity potential of different mixes.

As you can see from the leftmost graph injecting 3% water to fuel ratio of pure water is equal in cooling capacity to enrichening fuel mixture from 12.5 to 10.0 AFR. This is dumping a lot of fuel. Latent heat capacity of water is 2256 kj/kg, of methanol 1109 kj/kg and of gasoline 350 kj/kg
For those living in colder climate I would recommend to change your nozzle for smaller size or disable system at all for winter.
Tons of interesting reading on subject of alcohol/water injection you can find on Water Injection Forum.

Update - December 2006

One year ago after reading very positive reports about Derek Devices MW injection kit I decided to ditch my non-progressive Aquamist and buy Derek's kit. Man behind it is Derek O'Banion, he did good job compiling parts to create progressive kit which works with AEM EMS. Main components of the kit are Shurflo pump, 4 meter long hose, M15 nozzle (0.25gpm@100psi) and solid state relay. Solid state relay uses one of the fuel injectors signal to cycle the pump and progressively increase flow. With AEM EMS it is possible to set up secondary fuel injector driver to control the pump - very neat.  Idea to have MW controlled by AEM was very appealing to me but after reading reports about failed solid state relays on Derek Devices mail list I decided to buy Coolingmist progressive controller. Now my setup is Derek's "on-off" kit + Coolingmist progressive controller. Controller uses boost pressure signal to vary pump's voltage supply - not as neat as high speed valve, but still better then "on-off". Controller has two adjustable potentiometers - one to set beginning of injection, second one to set boost pressure at which you want pump to run full blast. I start injecting at 6psi and pump is at full blast at 22psi. To test how progressive controller works I

fed it with compressed air through the pressure regulating valve and gradually increased pressure from zero to 30psi. I was impressed, even at low simulated boost pressure 10psi spray was good, no dripping, pissing, etc. Regret I didn't make video of this test.
I installed the pump and 5 litres tank in the trunk behind the driver side panel and routed the hose under the driver seat, over the transmission into the engine compartment. Pump and tank have to be lower than spray nozzle location to prevent siphoning of MW mixture.
Derek's kit flows a lot more then Aquamist, if Aquamist 1s with 0.9mm nozzle provides enough MW mixture for around 500hp, Derek's system supports 700-800rwhp Supras. This system is too big for me to inject 50/50 mix, I use 70/30 mix of methanol/water, pump is compatible with 100% methanol. Kit allows to run high boost on pump gas, some guys run as much as 25-28psi, I tested it on our 98 RON gas with boost controller set to 1.7bar. Consumption of MW mix is much higher now, Aquamist consumed around 0.5-0.8 ltr/day, with Derek's kit I have to refill around 2ltrs daily. Higher
than 50% content of methanol requires additional lubrication for pump to save internal viton seals. I use 2 tablespoons of Redline alcohol fuel lubricant per 1 gallon of pure methanol.
One more smart feature of the kit is built-in adjustable NC (Normally Closed) pressure switch on the discharge side of the pump. AEM EMS can be configured to safeguard motor if for some reason there will be no pressure in the discharge line (pump failure, empty MW tank, etc.) Though this is not flow sensor like on more advanced Aquamist DDS-2, pressure switch can't detect blocked nozzle. I adjusted pressure switch to activate (go open) at boost pressure 0.7-0.8bar. If AEM detects no pressure in the line at boost higher than 0.9 bar it will dump 18% more fuel, pull ignition 4 degrees and switch over to lower boost target of 0.9 bar.

Update  - April 2009

I had good results with Derek’s pre-throttle progressive MW injection, but good results maybe not enough J, so I decided to upgrade it to direct port injection. After some research I found that Coolingmist has everything I need for conversion job. So, I needed:

1. 6 straight nozzles, capacity depends on your hp goal. I purchased 2 sets – 6 pieces 246cc/min for my small 61mm turbo and 6 pieces 506cc/min for my future bigger turbo.

2. 6 quick-connect fittings, 1/8 NPT (nozzles come without this fittings)

3. 6 nickel plated T-fittings, they come with quick-connectors from all 3 sides

4. 2 check valves with quick-connect fittings

5. 10ft of ¼” hose

6. 1/8 NPT 27 tpi tap, if you don’t have one

I did some very rough calculations assuming that I will be running 80/20 MW mix and pressure drop across the check valves ~10psi, boost pressure 30psi. I estimated effective nozzle pressure for 246cc nozzles ~90psi, for 506cc nozzles ~60psi. It depends a lot on how restrictive the system, beginning from the suction strainer, suction hose size and length, discharge hose length, etc. Additionally you can expect high methanol percentage mix to flow slightly better then water due to it’s lower surface tension.
According to my above rough calculations if I run six 246cc nozzles at 30psi boost with my 61mm turbo (close to 600 flywheel hp) I will get 8% of water by weight and around 30% of methanol to 70% of gasoline by volume.
For my future 1000bhp turbo (I hopeJ) with 506cc nozzles I would get 9% of water by weight and 36% of meth as fuel. This is almost dual fuel setup. Another thing to consider is switching over to bigger Shurflo pump, my standard pump definitely will not support flow of six 506cc nozzles.
I decided to put together something resembling fuel rail using ¼” hose from Coolingmist. Since this “fuel rail” ID was small it has to be fed from two sides. Feeding MW mixture from one side would cause other side nozzles flow less. I connected two ¼” hoses through the Y-fitting to my old Derek’s kit hose and used these two hoses to feed “MW rail” through two check valves. Check valves have crack pressure high enough not to allow meth to be sucked into intake manifold under vacuum condition. If methanol injected pre-throttle this check valve not required if injection point higher then MW tank.
You can see my MW rail on these 2 pictures, not super clean looking but functional. Coolingmist carry 6-port manifolds, but I think my setup flows better and looks less complicated.

Now about results. OK, there were few reasons for going direct port injection. Though I frequently checked my spark plugs for signs of detonation and unequal fuel distribution and never found anything wrong I was still concerned about equal MW distribution between cylinders. My datalogs was showing too high from my point of view knock sensors activity. I hoped to calm this activity down. Another reason for switch was to feed more meth into the cylinders. So, after I tuned fuel map with direct port injection I found no reduction in knock sensors voltage, which means all this sensor activity was background mechanical noise, or what is called knock signature specific to my engine. In fact I found first signs of detonation on spark plugs after switching to direct port injection. Later on I blew my headgasket when was running car hard in scorching summer heat. I suspect if I had my old pre-throttle injection I would get away undamaged. When meth injected upstream TB - mixture is completely vaporized when reaching cylinders. It causes massive air charge cooling which reduces motor octane requirements a lot. This effect is nearly absent in case of port injection, vaporization takes place mostly during compression stroke. This has it’s own benefits, but for street car I would prefer to run simple and cost effective pre-throttle injection. After blowing my HG I purchased AEM progressive kit, installed pump and tank behind passenger seat and will be running it as pre-throttle injection on the street with 75% methanol mix. For 1.3bar boost on pump gas it is more then enough. I keep my port injection system for running high boost on the track and plan to inject 100% meth through it. It will be nearly dual fuel configuration, injecting methanol close to 50/50 ration to gasoline. It is not substitute for race gas, but paying $12-18 per liter of exported race gas is not very attractive option for me.
Another very important aspect is proper atomization when each of 6 runners equipped with it’s own nozzle. If nozzles size is too big methanol pressure will drop (time to step up for bigger pump) Poor atomizing means larger droplets of methanol in air stream and this will cause slower burn rate in combustion chamber. Slow burn rate is bad thing, it can allow time for additional flame front to start, in other words – detonation. Ideally we would like burn rate to be very fast and somewhere few degrees ATDC (my uneducated guess) where the leverage for producing torque is best.
I am thinking of getting rid from progressive controller for direct port injection, I start injecting at 0.7 bar of boost and ramp it to 100% at 1.7 bar and lower pump rate at lower boost pressure is recipe for poorly homogenized air/fuel mix and wide variation of droplets size followed by irregular combustion process because of inconsistent and uncontrolled burn rates in combustion chamber. If fuel droplets in the air/fuel mixture are not of the same size combustion process is very unpredictable - smaller droplets burn quicker, larger droplets burn slower. If ignition spark advance is set for slower burn speed of larger droplets then smaller droplets which burn faster will give you trouble (detonation). I think progressive controller better used for pre TB injection when MW mix has enough time to vaporize before entering combustion chamber.
Click the picture below to watch the video of six 246cc nozzles fed by 150psi pump in action.