RomRaider

The bit about setting IAM to the 'initial' value puzzles me. How then could the ECU ever wind up with an IAM that is anything other than Initial, Initial+4, or Initial-4 ? (I assume that's Initial, Initial+.25, or Initial-.25 for 32-bitters)

I had thought - until a conversation with XPT over at IWSTI recently prompted me to re-read this stuff - that the Initial value was only used after an ECU reset. I am still inclined to think so, because I've seen LV snapshots with values other than 4, 8, or 12. (.25, .5, or .75). What am I missing?

Thanks!

the IAM step value is halved with every change in direction. That is how the ECU determines when the IAM has "settled" on the appropriate value (or if the IAM hit min or max - ex. 0 or 1.0 32-bit ecu - for a period of time). So, the step value of 4 (or 0.25) is only used in the first change. When you go from fine to rough correction mode, the IAM doesn't necessarily change at that moment. Other conditions have to be met. When those occur, it very rapidly resets IAM to initial, clears the FLKC table, and then applies the initial step value. It would be difficult to catch the change to initial when logging.

When you reset the ECU, the IAM is set to initial, FLKC cleared, but there's no immediate change to the IAM as opposed to going from fine to rough correction mode and THEN hitting the conditions necessary and then, bam!, the IAM is set to initial, FLKC is cleared, and the initial step value is applied to the IAM.

But how does IAM ever reach 16 if every re-evaluation means resetting it to 8 and adding a delta that is at most 4? That would get IAM to 12, but never higher.

Does it re-set IAM to the initial value only once per transition into rough correction mode? I mean, if it transitions to rough mode, then re-evaluates IAM three times, does it start from a value to 8 all three times, or does it reset to 8 only the first time, and then apply changes to the current IAM on each subsequent re-evaluation?

No, the reset to the initial value occurs once after going from fine to rough correction mode once the conditions are met (kca max min. met, fbkc disabled, in rough corr ranges, etc.). Then, the IAM change continues whenever the conditions are met again (or if they are still met). Advancing the IAM also has a slight delay. So, for example, if the initial was 8:
1. conditions dictate switch from fine to rough mode.
2. another set of conditions dictate (kca max min. met, etc.) when to reset the IAM to initial/clear the FLKC table.
3. IAM is reset to initial = 8. Check knock signal. If set, IAM - 4, if clear, IAM + 4 after delay period. Let's say - KS clear (IAM = 12) after delay
4. check same conditions again. If met, check knock signal. Let's say knock signal is set. IAM step value reduced by half = 2 (IAM is changing directions). IAM = 12 - 2 = 10.
5. check same conditions again. If met, check knock signal. Let's say knock signal is clear. IAM step value reduced by half = 1 (IAM changed direction again). IAM = 10 + 1.
6. IAM step value = 1. IAM has "settled" at 11. Switch to fine mode.

If there's no knock, the initial step value won't change -> initial value 8, next=12, next=16 (another way to exit rough correction mode is if IAM is at min or max for a period of time).

Got it, thanks for clarifying.

The upshot of all of this is that I'm going to set my initial IAM back to 0.5 when I'm satisfied with my tune. I set it to 1.0 for the sake of logging right after a reflash, but it sounds like if I ever do get a tank of crap gas I'll be better off if it's 0.5, so the rest of the IAM adjustment stuff works as designed.

Wow! Nice write up!!!!

Would you be able to answer the following question?

I would like to know what would be the maximum/minimum limits of advance modification when say IAM = 1 on 32 bit.

Like what would fine adjustment (max) + feedback knock correction (max) = ??? value?


Reason I ask is because with my current live log device, I can see 'Learned Ignition advance', with a max value of 8 degrees advancement on WOT for a near or on max load...

What would that 8 degrees of advance roughly equate to if you were to convert it to IAM?


Cheers!

Merchgod,

I have questions regarding the 32bit 04 sti rom, A2ZJ710J. The KCA is now calculated differently as shown below. I have read over the descriptions (thats where I got the formula below).

Total timing = base timing + knock correction advance + other timing compensations. Knock correction advance = (((KCA max primary low * primary advance multiplier) + (KCA max primary high * (1.0 - primary advance multiplier))) * IAM) + KCA max additive

For total timing, are the "other timing compensations" = feedback knock correction + fine learning knock correction + compensations due to IAC, ECT, ect. ?

Can you go into more detail about how the primary advance multipler is calculated? Can this value be logged? How does enabling the KCA alternate mode affect this equation? The description says the KCA max additive is basically switched to zero. But then it says it affects several other tables when KCA alternate mode is enabled including KCA max primary high and low tables.

This is more complicated than the old Knock correction advance = (KC advance max map value * IAM) + feedback knock correction + fine learning knock correction. Would it be a bad idea to just zero the KCA max additive table and use the KCA max primary high and low tables to simplify tunning? I would be leaving the KCA alternate mode disabled.

It is a bit complex to go into here, but it is based on knock and conditions that may lead to knock and dynamically adjusts this multiplier from 0 to 1 based on this and therefore impacting the switching between the two maps (0 or 1) or a combination of both (between 0 and 1). I do not have it in front of me, but there should be two ext. logger parameters that will cover this multiplier and the final primary advance. I don't remember what they are called, probably "knock correction advance..." something (probably map ratio for the multiplier).

Which map (KCA max primary high or KCA max primary low) should have the high timing values? Is the engine happy with a multiplier of 1 or 0? Sounds like I could simplify this by making both maps the same, zero the KCAA max additive map, and depend on my FLKC, feedback knock correction, and IAM for ignition retard. Does this sound like a bad idea?

"knock conditions high" is basically the "failsafe" table (multiplier = 0). IIRC, it also the default when conditions aren't met to evaluate the "knock conditions".

Many tuners use the strategy that you are talking about -> it will make it easier for you to predict what the timing will be at any given time.

How is it possible that I have logged flkc but did not see it learned in the FLT as per LV? I have extened ranges to cover for EL and rpm. Usually knock especially in high rpm are learned immediately. Iam goes below 1 but no learned knock to show for.

Learned knock is cleared when IAM changes.

So if Iam is 1 then you get flkc, then it is learned. If flkc occurs again enough to trigger iam to go below 1, FLT table will be cleared?

Correct, any time there is a change in IAM it will clear the learned FLKC.

Btw, what does mean?
What is limp-home mode?