Boost/Ignition Controller


This page describes my Boost/Ignition Controller and some of the background design principles.


First of all, for the technically minded, it's worth explaining how the standard boost control system works on the Impreza. Boost pressure is controlled by software in the ECU - the ECU measures manifold pressure, using the MAP sensor, and then regulates boost pressure using the wastegate solenoid. (As well as manifold pressure, RPM is also used as part of the input map for boost pressure to reduce boost at high RPMs.) The actual configuration of the turbo tubing and components consists of a pressurised hose from the turbo to the wastegate actuator with a T-piece off to the wastegate solenoid that bleeds air to the intake side of the turbo under the control of the ECU. (Out of interest this is a slightly different arrangement to other cars such as the Nissan Skyline.) The wastegate solenoid is a non-linear device - it's either fully open or fully closed - so to allow more control (and essentially make it linear), it is pulsed open, then closed, 14 times a second (you can hear it clicking from just above tickover in the off-side of the engine compartment). As boost pressure rises, the time it is closed for increases in proportion to the open time, which provides more pressure on the wastegate actuator, which in turn allows more air to bypass the turbo, thus reducing turbo pressure. The resulting system is a classic closed loop control system. The graph below shows some typical responses to a step change (i.e. when you floor it!). The standard Impreza's response is underdamped, which is fine on a car running standard boost, but the overshoot is potentially dangerous on a car running high boost. The ECU also has a engine protection system referred to as "overboost cutout", whereby it will cut the engine if it detects too higher boost levels (typically greater than 15 or 16 psi on recent UK cars).

Also, a more intelligent controller would bleed off maximum air through the wastegate solenoid until the boost pressure was close to the desired pressure. This would allow boost pressure to rise more quickly and improve throttle response. The standard ECU is already closing the solenoid about 10% of the time from just above tickover. This will provide smooth power and easier boost control at the expense of a fast throttle response. although the downside effect is unlikely to be that significant. Back to my controller - in order to provide a fast response, yet with no overshoot requires careful adjustment of the software parameters. One key element for example is "differential control" - that is, it must be able to predict to a certain extent when the target pressure will be reached, so it can anticipate and drive the wastegate solenoid in plenty of time in order to prevent overshoot.

"Restrictors": As a slight aside, it is possible to increase the overshoot by increasing the restriction (to restrict air flow) in the turbo hosing between the turbo and the first T-piece (upstream of the wastegate and wastegate solenoid) - this delays the rise of pressure in the tubing to the wastegate, thus the manifold pressure will overshoot as in the red line above, providing a small surge of extra power when the accelerator is floored. Some overshoot is OK when running standard boost, but when running slightly higher boost, the overshoot then climbs briefly to dangerously high levels (which incidentally the ECU usually misses with its overboost cutout detection). The Impreza has a small brass restrictor fitted as standard, but this may be closed down slightly by plugging it with solder then re-drilling using precision drills (0.1mm tolerances can be critical!) Alternatively, a restrictor can be made using a suitable length of rod drilled out with a say 1.0mm hole.

However additional restriction can also be of benefit under steady-state full throttle. This is because the pressure drop down the turbo hosing from the wastegate to the solenoid is significant enough to provide enough pressure on the actuator to open it very slightly. The ideal solution is to mount the solenoid nearer the actuator - but of course that may overheat the solenoid. Alternatively, a slightly stronger actuator would alleviate the problem, although this can disturb the control loop (i.e. make turbo pressure less controllable). A restrictor also solves the problem, although again potentially at the expense of loop stability. (Note this is a very easy way of slighty increasing engine power, as long as there is sufficient headroom before overboost cutout.)

Ignition Timing: As boost pressure rises, the ECU retards the timing - in fact by 15 degrees between -10 and +15 psi of boost. Originally I had assumed that boost pressure (i.e. output from the MAP sensor) was used by the ECU as a major input to the ignition map. On further investigation, it appears to have a minimal effect. I would guess that the MAF sensor plays a more significant part (along with RPM etc).
MAP Sensor: As well as closed loop boost control, the MAP sensor signal is also used by the ECU to cut the engine if it detects overboost. This is an important safety feature - something sadly lacking from low cost boost kits. The trick here is to still provide overboost cutout, but at a higher boost pressure. Also the standard MAP sensor is not very linear above about 15 psi, so my controller must take this into account (i.e. non-linear mapping). Finally the MAP sensor is part of the closed loop boost control system as described above - however my controller takes over this control function, which frees the ECU's MAP sensor input to provide ignition advance (although this I now believe is minimal) and raised overboost cutout functionality.
Fuel Mixture: Fuel is metered by the ECU from information derived from the mass air flow (MAF) sensor in the air intake and lambda control from the exhaust oxygen sensor. This results in the ECU quite happily providing sufficient fuel (in fact very rich - as per usual) at higher boost levels (up to 18 psi) - suggesting that there is probably no need to consider uprating the fuel pump or injectors (verified on the dyno runs). There is a slight concern that although the average fuel mixture is OK, individual cylinders may get leaned out. However this alleged danger only appears to be of concern at very high boost levels - and not at the relatively small increases my controller does. Note also that a rich fuel mixture is probably there by design - it moves the knock point further away by providing cylinder cooling. However some argue this is debatable as European engines, for example, run much leaner. My own opinion is that as long as sufficient fuel is being supplied, then leave well alone. (Unless perhaps you use water injection to move the knock point further away.)


My basic design aim was to provide an electronic controller, fitted in minutes by plugging into the main loom to the ECU (i.e. no wires to cut) that provides switchable performance whilst driving - with no other mechanical modifications - such as loud exhausts etc. Also of course it should have no effect on other ECU features, such as overboost cutout, immobiliser, cold start, anti-knock ignition retardation etc. (As an aside, I'm a bit wary of complete ECU replacements, as they usually do not provide all the safety/engine protection features that are in the standard ECU.) Anyway this has all been achieved apart from adding a free-flow air filter (i.e. the only other modification to the car).

The final controller pictured below has three modes - "Economy", "Normal" and "Performance", which may be switched "on the fly" whilst driving. And this is the fun part - gone are the days of getting used to high performance - just run in Economy Mode for a while, then switch back!

Boost/Ignition Controller showing Cable Connector and In-line ECU Connector


Or in other words, how does it work? First of all, running Economy Mode and Normal Mode are easy to explain - In Economy Mode, the wastegate solenoid is kept closed and there is no re-mapping of the MAP sensor. This results in a maximum of about 7psi boost and consequently restricts the acceleration and fuel usage. In Normal Mode, the MAP sensor and wastegate solenoid control signals are simply passed directly through the controller (via integral relays) to the ECU. (Also, should the 12V power supply fail to the controller, the relays will revert to Normal Mode as a fail safe feature.)

Performance Mode is the more complex one. There are two distinct operations - firstly the MAP sensor is re-mapped so that the ECU provides overboost cutout at a slightly higher level; and secondly, the closed loop control of boost pressure is completely taken over by the controller. In fact a "dummy" wastegate solenoid inside the controller fools the ECU into thinking it still has control. The actual boost pressure is settable by a 16 position dial at the rear of the controller. This allows boost pressure to be adjusted - for example, adjusting for summer/winter weather conditions, or turning down for track days etc.

A nice security feature is that without the controller plugged in, performance is reduced to Economy Mode with the engine cutting out whenever the accelerator is pressed.

The results, as dyno'ed for each mode, are on theEngine Torque and Power: Dyno Graphs 2 page and the respective acceleration times are on the G-Force and Acceleration page.

Engine Torque and Power: Dyno Graphs 3 shows the effect of running 17 psi in Performance Mode rather than 16 psi and Engine Torque and Power: Dyno Graphs 4 shows the effect of adding a free-flow exhaust (de-catted centre-section plus free-flow backbox).


Again for the technically minded - the controller algorithms are implemented software on an 8 bit micro-controller. The other chips are a digital to analogue converter and a couple of op-amps, along with some relays and various other support components. The box and cables are all screened to eliminate any interference effects.

Fitted (and removed) in minutes - just plug in-line with the main loom and ECU

Effects on Engine Lifetime: What is the long term effect on the engine? Well that depends very much on how the car is driven (and in what mode). If Economy Mode is always used, then the wear on the engine will be less than normal, all other things being equal. In Performance Mode, the additional power will put the engine under increased strain and is therefore likely to reduce the engine's lifetime - how much is a debatable point. I believe with regular servicing and not thrashing about all the time (!), the effect is likely to be minimal or even insignificant. Using Performance Mode (at say 17 psi) on track days may well be more harmful, and for this reason I usually only run 15 or 16 psi. Note also the additional power is only available under full throttle - so for example, using Normal Mode and full throttle often, will wear the engine faster than using Performance Mode and full throttle rarely.

ECU Reset: Finally I thought I'd add to the ECU reset debate - I have seen no evidence during the development of my controller that the "ECU reset" exists. I have even simulated knock (using a hammer on the block!) to watch the ignition timing change - and it appears to reset itself within a few seconds. So my current opinion is that it's a bit of a myth - although perhaps it applied to earlier models? Certainly there will be some learning ability but probably within very tight confines of the ECUs standard parameters. Also it may well be that going "for blast" (part of the ECU reset myth/"method") improves the car's performance for other reasons - nothing to do with the ECU. However I'm willing to be proved wrong. I'm also dubious about 95/98 RON making any difference on an unmodified car - but that's another story.

Comments, suggestions or questions welcome -

Return To Home Page.

Last Updated: 28-Nov-99 (Various minor changes in the light of further investigations).
Previous Changes:
15-Jun-99 (Minor changes)
20-Feb-99 (Created)

Copyright © 1998-99, All Rights Reserved.