openecu.org

I noticed that the ecu learning ability in exports JDM is different to the JDM ecus.

I know that this is for safety on export markets.

What I would like to know where in the ecu affects learning ability.

I drove in a wrx tonight that have a exhaust mod and possibly a up-pipe upgrade that is leaning out in the 3000rpm range. When it pings, the ECU pulls the IAM from 16 or any value to 4 or 6 and boost drops to 6-8psi and increases slowly as IAM goes up, but n ver reaches it full value until the car is happy. The IAM stay at the low values until the car is reset or learning over a very long period; unlike the JDM. The owner states that it take a week for the car to feel better again.

What values can be change to improve learning a bit or even which maps will affect boost that much once IAM reduces?

I am currently working with a similar rom right now. Subaru Impreza WRX 2003MY - AH750 5141. His ECU is AH750 5141

I will have to pull his and see if it is the same.

I think below affects the Boost. When the IAM is 8, I can get full boost, below, less than 10.

#00 - Advance Mult for Boost Disable = 6
#00 - Advance Mult for Boost Re-Enable = 8

If the car gets a hard ping in the stated untun region, the IAM drops but takes like years to 8 or 10. But the ecu can get a ping and reduce to 10 or 8 and happy there for awhile. A long while


Does:-

#00 - Fine Ign. Learning Advance Interval

affects the learning ability? The lower the number, the faster the learning, since the jdm seems to be lower than 100 and export above 100, 183 to be exact?

Don't mess with the boost disable when IAM drops below 6. It's there for a reason. If the IAM is dropping to 4 on said car, then that car has the crapiest tune in the world. Furthermore, if that car is an 04+ car without any tune on the ECU then he should know better. You need to stop with the resets on that ECU because that's just causing the engine to repeatedly detonate. A tune should not cause a car to det and have it's owner hope the ECU learns around it.

When you have a good tune the IAM will raise after a reset from it's starting point (which can be set to whatever number you want... it's 8 on the OEM tune). If the IAM drops after a reset then the engine has serious det issues which must be fixed with either a retune or finding the faulty parts which are causing the new det if the engine was previously running correctly on the same tune. Spark plugs will sometimes cause det issues... either incorrect ones or ones that a perticular engine just doesn't like.

Hey, don't get frightened. Read my objective about.

We needed to reset to see whats going on. The car does not learn like the JDM. So you can sit there for the hold day and it goes no where without stats to what is going on the some jdm models I see with coarse ign retard and so on.

The car can be @ 10 for a long time or even 8. But I wanted to show that if after a reset and it pings, it would go to a lower value and stay there.

It suppose to have never tune, except the mods. I have foud the spot and it is too lean. 12.93+ afr just before full spool. The EGT rise a little sharply in that area.

The areas mentioned above would help others identify those settings that affects the area discuss.

My interest is the learning ability which I think I have found.

No one is getting frustrated, you misunderstood completely. The speed of learning characteristics of the ECU have nothing to do with whether or not the ECU will ping or not. The tune of the ECU has to do with that. While it may take longer for the USDM learning corrections to take place, that's not what I was trying to state... I was trying to state that even if you change that portion of the logic to try and speed it up you'll still be in an unhappy state of tune because regardless of that learning speed if the engine is not happy with the tune your final number will still be the same, you'll just get there quicker, that's all.

Again, no one is getting irritated, but it's hard for me to know what you are aware of or not so that I can tell you the most likely cause for the det issues you're having. That's all. Take care.

No. Not angry or anything.

What I noticed is what I stated above. I am not sure if I was clear about it, but I understand the whole tuning factor and the effect of ecu learning.

When I sort this problem out, I would like the ecu to learn like the jdm. Its painfully slow. Hence my questions.

Take for instanst, the ecu would feel good @ IAM=6 or 8 for hours and different loads. Get a ping it may drop to 4 from a ecu reset or battery disconnect and will take days to get to 6-8.

But I am aware of problem causing it, because it is outside of CL for the ecu to add fuel to correct the problem.

Once the owner allow me to pull the rom, I will adjust that area and see what happens.

He is classifying his car as a normal 2L now when boost is low

That's what I meant... you're seeing the slow transitions because the ECU is seeing knock. For instance, let me give you an example of what my ECU does after I reset it.

Do a reset and IAM drops to 14 (that's my starting point instead of 8 ). If the engine is happy and there's no knock the IAM climbs up to 16 within 1 mile of driving with some part throttle boost or even full throttle boost... doesn't matter. If it's not happy it will drop the IAM to 12 or 10 or whatever it needs to keep KC values positive without causing knock. Now, let's say my IAM drops after a reset to 10 because of knock. The ECU will not readvance the IAM back up without a lot of "knock free" driving. It's simply how that works. The ECU does not retest the limit of the knock threshold at too frequent of a rate because usually knock (on a properly tuned engine) will only occur if you have crappy gas or something environmental causing the knock (for instance). So the time delay for it's retest tend to offset (from my findings) about the same time as it takes to go through 1/2 or a full tank of gas... which is usually a couple of hundred miles. Of course it's a lot more complex then that, but that's the jist of it.

So with that said, and I know I sound like a broken record, don't expect the ECU to detect knock, drop IAM from 8 to 6 and then within 5 miles try and bring the IAM back up... it won't happen. Also, if your IAM is dropping from 8 to lower there's definatly problems. That thing should be pegged at 16 all the time for trouble free opperation. Does that help at all or am I just restating things you already know?

To me IAM values of 14-16 are good, 10-12 ok, 10 or below suck.

Not at all.

I would say I started of with a question and end up learning what it was and so we are now educating new users.

Below 6 only occurs after reset or serious pinging in the stated region. I heard it myself Was amusing, since the customer is eager for change.

What you state is true. It is just that we prefer JDM capabilities and with new tuning online, we need that ability

Boost enabling would not be interferred, since if there were issues, this setting is good.

There is two problems really.

We have bad gas here, althought the guy is going to mix race gas soon. Our octane is 86-88 or even 89. Don't know.

A wrx would not behave this way stock. Change the exhaust, IAM might be @ 8+, but never drop.

Second problem is exhaust system to rear car and maybe an up-pipe.

These are recipies for lean conditions out of CL into OP, since pinging in OP.

This is the problem. Once fix, I would bring learning up to or close JDM. Changes just occurs too slow. Most probably I will pick a number in between 63-183.