[PSIG]

Every application would be different, but for controlled fuel applications (race) I use 3% and 5% as my warning and limits, while for street it's typically 5% and 7% due to poor fuel quality control. My angle on this is simply that a running target of 14:1 at a specific point of the tune, with a limit of 10%, is capable of adjusting from 12.6:1 to 15.4.

Those limits are well outside of reasonable for the expected conditions at that table cell. 3.6% would be .5 AFR, so 13.5-14.5 as a reasonable correction range — again depending on application and fuel.

Currently, we can use long(er)-term averaging with very large #cycles between corrections. I have also used rolling average to nail values before, with incremental effect to the running average. Many approaches, but that circles back to why you want this? Be specific.

It could be very different use for me than you, e.g. I want it to compensate for seasonal changes in fuel blends with global table adjustment, while you want it to correct for specific small errors in the tune. Very different uses, and how it would function.

Be specific, and for greatest flexibility and usefulness, consider other uses and how the feature could address those also.

[purplelightning]

I was under the impression that OEM LTFT is typically a gamma value, no? Not saying I think it's ideal.. Seems to be hard to find real concrete data on these things..

[Ilotalo]

Mitsubishi evo OEM ecu has first O2 trim +-12,5%. This zreroes when engine is stopped.

Then there is long time low, mid and high trims +-12.5%. These are stored in battery memory.

When O2 trim goes over/under some value it correct long time trim that way. These low, mid and high is rpm based. Idle, cruise and high rpm. When engine goes boost it ignores these corrections.

I have been thinking this type of trims but i dont know do processor have enough time write those long trims after engine shutdown.

One thing is to make that continous power to that processor.

[JHolland]

Mitsubishi evo OEM ecu has first O2 trim +-12,5%. This zreroes when engine is stopped.
.
.
.
When engine goes boost it ignores these corrections.

What year EVO?
The Mitsu ECUs that I have dealt with are NA but always apply the LTFT, I could see the logic if the ECU was leaning off the mixture, a false reading could cause engine damage but if the ECU is enrichening the mixture then not applying LTFT will lean off the mixture which doesn't seem sensible on a boosted application.

[Ilotalo]

I got galant vr4 an i used one evo8 ECU. There might be those where is only O2 trim, but basic ECU you have non boost trims.

In Finland you don't know the mixture of fuel 98E5 or 95E10 can have 0-5% or 0-10% ethanol. That's make your fuel mixture differend.

[JHolland]

There might be those where is only O2 trim, but basic ECU you have non boost trims.

Older ECUs usually have only one set of trims, from around 1998 the Densos that I have worked on have three different sets of trims. You need to use a proprietary scan tool to read them, OBD2 only supports one set of trims.

[digmorepaka]

I got galant vr4 an i used one evo8 ECU. There might be those where is only O2 trim, but basic ECU you have non boost trims.

In Finland you don't know the mixture of fuel 98E5 or 95E10 can have 0-5% or 0-10% ethanol. That's make your fuel mixture differend.

The more ethanol content you have, the more fuel per the amount of air you need, however for the sake of the example let's assume lambda 1 E0 tune, your ecu will always hit 14.7:1 AFR, lambda 1 on E10 is 14.1. On an E0 lambda 1 tune you will always be running slightly lean if you feed it with E10 regardless of closedloop. Closedloop on that E0 tune assumes that lambda 1 = 14.7 AFR and target it. Without an ethanol content sensor you wont be able to differentiate between E0 E5 and E10. E10 itself also means "up to" 10%, in my country E10 is most often 6% ethanol, so it's all really a guessing game of close enough.

[iltheo84]

The more ethanol content you have, the more fuel per the amount of air you need, however for the sake of the example let's assume lambda 1 E0 tune, your ecu will always hit 14.7:1 AFR, lambda 1 on E10 is 14.1. On an E0 lambda 1 tune you will always be running slightly lean if you feed it with E10 regardless of closedloop. Closedloop on that E0 tune assumes that lambda 1 = 14.7 AFR and target it. Without an ethanol content sensor you wont be able to differentiate between E0 E5 and E10. E10 itself also means "up to" 10%, in my country E10 is most often 6% ethanol, so it's all really a guessing game of close enough.

Closed Loop always targets Lambda 1, not AFR 14.7, because O2 sensors only read Lambda, regardless of the fuel you are using.
The slight differences among different fuels and ethanol content (up to a reasonable percentage) are easily compensated by even the less advanced ECUs.
In open loop, usually OEM tunes run overly rich to avoid any lean condition under load.