RomRaider

I have a question that I hope can be answered here, I have been looking at my logs with regards to IAM and sometimes in my logs when IAM drops It begins to rise slowly to 1 again, but i noticed in my recent logs that my IAM is at .4375 and does not rise anymore. It just stays there. Is it supposed to do this and if so can someone explain the logic that got the IAM to stay at .4375?

Thanks for the help,

James

You likely have some kind of issue with knock due to any number of reasons - mechanical, your tune, your gas quality/octane, etc. The IAM will not increase until certain thresholds of positive FLKC is reached and the same for decreasing (negative FLKC thresholds), so the amount of advance at an IAM of .4375 is likely optimal for the current situation.

I have a question about the knock control strategy.

When the KC table is "zeroed" out, IAM is effectively disabled (WRT timing response). Does the ECU still attempt to use IAM when the normal criteria for re-evaluating IAM are met? Or does the ECU recognize that the KC table is blanked and revert to FBKC and FLKC only?

It would never enter rough correction mode because the minimum advance map value for rough correction is never met (usually something like 3.8-4.9).

I'm trying to understand FLKC a little better...I read in another thread that FLKC is only positive if the IAM is less than 16 (or 1 depending on 32 bit/16 bit ecu). This makes sense.

From what I can gather, FLKC is only negative when there is knock. Is there a time when FLKC can be negative, but there isn't knock?

Also, would it be correct to say that a car in perfect running order that absolutely never knocks will always have an IAM of 16 (or 1), and FBKC and FLKC are always 0?

FBKC and FLKC are only negative when there is perceived knock. The knock sensor is a microphone set in the engine block that picks up vibrations. An algorithm is applied by the ecu to the sound picked up by the sensor in order to filter out other transient noises from true knock but, in some cases, vibrations such as those caused by a loose turbo shield may be perceived as knock resulting in negative FBKC and FLKC. These events are, however, usually sporadic and are therefore more likely to result only in negative FBKC and not FLKC. So, to answer your first question, there are some scenarios for which you will have negative FLKC without actual knock.

If the car is in perfect running order, which includes everything being tightened down, IAM should remain at 16 (or 1) unless the initial value is set lower and the ecu has just been reset or it enters into rough-correction mode but, if all is well, this would be difficult to capture in a log as it would quickly climb back to 16.

Actually, FLKC is limited so that (FLKC + (IAM * advanceMAX)) will not exceed advanceMAX. So, if the IAM is 1.0, there can be no positive value.


Its not necessarily based on current knock - it is learned correction applied across load and rpm ranges. No, there's nothing else that would make FLKC negative other than some knock in that load/rpm range sometime in the past (knock as perceived by the ECU).


No, because the knock sensor setup is not 100% accurate at distinguishing knock from other engine noise at all times. Likely there are also cases where it does not pick up a knock event. Regardless, even if it were 100% accurate, no matter the tune, you would still occasionally have a knock event. Key is that you pay careful attention to knock events at moderate to high load and those that are consistent or persistant.

That brings up another question I have...I think I can put my logs into airboy's spreadsheet and see if it's actually a new knock event...but how does the spreadsheet do that? If I wanted to do it manually, what values would I compare to see if it's a new knock event, or just a learned correction from an old knock event?

Also, in the FAQ on the knock control strategy, it says this:

"Fine Learning Table Offset - current applied FLKC cell. If this value remains the same while logged FLKC changes, then the ECU made an adjustment to the FLKC in the current cell (barring any sync issues normal to logging)."

Can you elaborate on that a little more? I'm guessing it has something to do with all of this...

Yes, you would log that in addition to FLKC. If you look at the entire FLKC with Learning View, the offset would correspond to numbering each of the cells starting in the upper left hand corner and left to right, top to bottom. So, when the offset changes in your log, you are in a different load/rpm cell (it doesn't indicate that a change was made, simply a different cell). If the offset remains the same and the logged FLKC changes, then a change was made to that cell (remember it can decrease or increase within the limits specified by the system). If it increases, then there was a relative lack of knock, decrease means knock event as perceived by the ecu. Although I haven't looked at it, MickeyD's or Airboy's Spreadsheets probably accomplish this by looking at the 'Fine Correction Rows (RPM)' and 'Fine Correction Columns (Load)' in your rom and then determining what cell you are in if you logged rpm and load.

I just went and pulled this log, and here's how I interpret it, when comparing FLKC and and fine learning table offset:

Starting at row 10 in the excel file (which is time 1134), the ECU was in cell 12 of the fine learning table...initially, this cell was set at 0, but it then changed to -1.05, presumably because of knock;

Then it moved to cell 20 in the fine learning table, which stayed at 0;

Then it went to cell 21 which was initially at -1.05 (which was in response to knock sometime in the past), but progressed up to -0.7;

Then it went to cells 22 and 30 which were both 0;

Then it went to cell 31 which stayed at -0.35;

Then it went to cell 39 which went from -2.11 from a previous knock event and improved up to -1.76;

On and on like that. Is that all right? <crosses fingers>

Yes, but the only issue is that all the parameters on one line of the log are not sampled at exactly the same time. So, at time 1134, it is possible that the FLKC was actually for the previous cell, but then you moved to the next cell and the offset was logged for that cell on the same line. Best thing to do is to use Learning View before and after your run so you'll know exactly what changed during the run.

Got it...that was my next question, is why I would need to use the Learning View tool if I can just see this stuff in the excel file. This all makes much more sense now, thanks for taking the time to answer my questions.

Learning View displays the entire FLKC table all at once. To do that with logging, you would have to hit every load/rpm range while driving.

One last question (for now)....when you reset the ecu, it resets all values in the FLKC table to zero, right?

Yes and when you flash your ECU with ECUflash with most ECUs it also resets the ECU.