[KenL]

It has been mentioned on MINI2 that the MINI ECU can "learn" how you drive and it has been implied that this can lead to a quicker car if you drive it quickly in general.

I am sceptical about this and think it may be an urban legend. I also don't really see the point of it and although possible may be a complex thing to implement for little or no gain to the manufacturer of the car.

However, I am happy to learn (and being an electronics engineer am interested in the technical side) so if anyone really knows for sure I'd be interested to learn how it is done.

Cheers,
KenL

[theminiv]

i have heard that it adapts itself to the driver too. I think it does, cause the mini is such a clever little car!!!

[NEMINI]

ECU adaptibility is far from a 'new' thing or limited to MINI, its been the norm for most computer equipped automobiles for years. 10 years ago when you disconnected your battery and everything was reset the computer used to default to baseline settings and it was very noticable and improved as you drove. The reasoning for this wasn't so much for performance as it was for emissions control.

[Richard Crofts]

GMINI03 has it right. All ECU's 'learn'.....this is why when you take a car that is playing up in it can be reset to standard default settings (this would be impossible if there was no way of the ECU getting varied) If you fit air intakes and exhausts etc the ecu adjusts over time to account for increased air flow and reduced backpressure/exhaust retrictions etc etc...this over time can alter the memory/settings of the ECU and thus compensate by reducing/increasing and regulating timing of fueling etc to optimise emmissions not nessesarily power (although this is often a by product, as the car is attempting to create the most efficient burn charicteristics it can and this usually increases power also)

[northandy]

If only it was clever enough to cure the yo-yoing I get on my MCS !!!

[Rob.]

I have read this as well, though not sure how true it is. I don't really see how it would affect performance significantly.
And, ECU learning sounds like Terminator could come true

[ButtonIt]

I have noticed this "Learning" effect. I don't use my MCS much during the week, just the 50 mile round trip to work and back, where I drive very conservatively. When it comes to the weekend and I start to give her a bit of stick I notice that she drives like a dog to begin with. By Sunday she's driving like a beauty I may be imagining it, but I think it's true.

[Andy@Ross-Tech.com]

There are several different learned adaptations that take place, but the most significant and noticeable is fuel trim. Here is an explanation of fuel trim and what it does for us.

The DME controls Air/Fuel mixture in order to maintain power, efficiency, and emissions. A/F is expressed as either a ratio (14.7:1 for example) or as a Lambda value. With iso-octane ("ideal" gasoline), Lambda of 1.0 is equal to 14.7:1 A/F. This is known as "Stoichiometric", a condition where there is a perfect balance between oxygen molecules and the various hydrogen and carbon based molecules in petroleum. With the oxygenated gasoline that most of us use, actual A/F ratio of 15:1 is closer to stoichiometric.

If Lambda is greater than 1.0, then there is a surplus of air and the engine is running lean. If Lambda is less than 1.0, then there is a surplus of fuel and the engine is running rich. It should be noted that the ratios are mass-based, not volume-based.

So, why don't we always run at 1.0 all the time? Well, we do MOST of the time. At cruise and idle, mixture is held tightly to 1.0 to keep the catalytic convertor at optimal efficiency, so the emissions are minimized. However, when we need acceleration, the mixture gets richer. Why? Maximum power is made between 0.85 to 0.95 Lambda (12.5 to 14.0 A/F with iso-octane). So, under acceleration, mixtures get richer. Sometimes you want to get even richer under acceleration to keep detonation (pre-ignition of the mixture from excess cylinder temperatures) away. The MCS engine, which has a relatively high compression ratio for a supercharged engine, which especially under lots of boost, is very succeptible to detonation).

So, now that we know that the ECU wants to be able to control the A/F ratio. It has a prescribed set of values (maps) for a given RPM, Load, etc. So, the DME tells the injectors to pulse for exactly XX.X milliseconds and that SHOULD get us the proper A/F ratio that we want. Well, if you tell an employee to go do something, you want to make sure they actually did it, right? The ECU has some snitches (the front O2 sensor and the MAF, for the most part) that will report back whether or not the desired mixture has been attained. The rear O2 sensor is used mostly to monitor the condition of the catalytic convertor, although in some applications it also contributes to trim information.

Based on feedback from the snitches, the DME learns to apply a correction factor to its commands to the fuel injectors. If you know that your employees take longer than the standard allotted time to do a specified job, you will need to adjust for that in your planning (injectors are in a union, so it is tough to fire them). The learned values go between the maps in the ECU's Flash ROM (the "chip") and the signal to the fuel injectors. These learned compensations are known as "trim". So, when you see "trim", it means "compensation".

"Add" means additive trim, which is addressing an imbalance at idle. When the ECU is using additive trim, it is telling the injectors to stay open a fixed amount longer or shorter. The malfunction (e.g. vacuum leak) becomes less significant as RPM increase. For additive adaptation values, the injection timing is changed by a fixed amount. This value is not dependent on the basic injection timing.

"Mult" mean multiplicative trim, which is addressing an imbalance at all engine speeds. The malfunction (e.g. clogged injector) becomes more severe at increased RPM. For multiplicative adaptation values, there is a percentage change in injection timing. This change is dependent on the basic injection timing.

[Paul]

Andy, how does this work with ECU remaps by tuning co's? I'm a bit naive when it comes to how they can remap fuelling etc, and then the ECU's learning still work?

[Goonery]

Quote: Originally Posted by northandy

If only it was clever enough to cure the yo-yoing I get on my MCS !!!

I think the answer is one or two posts above yours, try disconnecting the battery , everything should reset to an as new state, worth a try

[Andy@Ross-Tech.com]

Quote: Originally Posted by Paul Mullett

Andy, how does this work with ECU remaps by tuning co's? I'm a bit naive when it comes to how they can remap fuelling etc, and then the ECU's learning still work?

As long as the chiptuner didn't do anything really dumb, adaptation still takes place the same way. Let's say that under load, the stock ECU wants to see an A/F of 12:1. It watches the O2 sensors and applies a trim value to keep the actual A/F close to the intended one. Now, Let's say that under load, the chipped ECU wants to see an A/F of 13:1. It watches the O2 sensors and applies a trim value to keep the actual A/F close to the intended one.

There are other adaptations as well, notably for fuel quality. The engine is running just fine, then the knock sensors detect detonation. Immediately, the timing is retarded and the DME makes note of this. It applies a trim value to keep timing a little bit retarded for a while. If it goes enough time without seeing detonation, it will advance the timing back to its normal levels.

[northandy]

Quote: Originally Posted by goonery

I think the answer is one or two posts above yours, try disconnecting the battery , everything should reset to an as new state, worth a try

I'll give it a go, although it is due at the dealer next week anyway...

[obehave]

Quote: Originally Posted by northandy

I'll give it a go, although it is due at the dealer next week anyway...

Disconnecting the battery won't work.