I've had this before with the bigger cam, after putting in 103 octane fuel I determined it to be false....so I zeroed out all of the spark retard tables at the rpms I was seeing it.

Now I'm seeing it again with a much smaller cam.

I can drive for miles without any KR...then once the Knock Counts register some counts and it gets KR, it then gets 5+ degrees of retard with any amount of throttle trying to accellerate....cruising is fine.
I'm seeing the 5+ degrees of KR without the knock counts going up any.
Then there's also times where I see the knock counts go up and no resulting KR/spark retard.

I've switched PCMs and switch Knock modules with LT4 modules and LT1 modules without any success. I have 4 spare PCMs and a box full of new knock modules, probably 4 or 5.

So, what are the reasons for KR / Spark retard when there are no knock counts being sensed??

And reasons for knock counts to increase with no resulting spark retard?

I really don't want to zero all the tables out again.

 

Its common to get knock retard without incrementing knock counts, and its possible for the knock counts to increment without causing knock retard. Knock counts generally increment a few thou on cold start. Knock counts appear to be generated internally by a PCM calculation using various sensor inputs.... perhaps looking for combinations of things which it thinks MIGHT cause detonation.... just a guess though

 

Any insight on how this happens during regular driving conditions? Maybe its going into the "low octane" operation some how? I could see it going into the low octane once knock is detected, but I'd think it would reset after cutting the engine off but it deos not. Once it starts getting the retard at all acceleration levels, it continues even the next time the engine is started up and you take off driving it has 5+ degree retard any time the allerator is depressed more. Only when the PCM power is disconnected or reflashed does it go back to driving for miles with no KR until it is again detected at a hard shift or real knock then it goes nuts again.

In all instances I'm talking about knock at regular driving, I know the starter will cause flase knock at startup etc.
It does seem as though it almost remembers where it detected knock and tries to compensate by automatically pulling timing when those conditions are reached again....hence why disconnecting the battery or reflashing causes it to act normal again until it detects real knock again.

 

If you don't see the knock counts increasing, its safe to assume the PCM is responding to a signal from the knock sensors. That's the point I'm trying to make..... "knock counts" do not appear related to the knock sensor. They APPEAR to be the result of an internal calculation made by the PCM. Can't say that for sure, but after looking at data logs for 100's of people, I'd say the empirical data supports that assumption.

 

Gotcha, thanks Fred for your insight.
I'll try pulling some timing across the board to see if it has any effect and go from there, its already lower by 3-5 degrees from the stock tables but I'm running a little more SCR than stock, even more than my truck with same cam since I have stock LT4 heads on the truck.

 

Fred, after looking at what feels like endless scans, I think I see whats going on...and BTW, I think I've cured it now of removing any timing and getting knock counts.
I observed that the retard never occurs until it gets its first knock count / retard reading...then I observed that it would start removing timing anytime I got in the same area of rpm / MAP as where the real knock with counts occurred even though the kncok counts would never increase again when getting in the "range". The more you drove, the more areas had "real" knock where the counts increased...and it seems to be a range of about 10 KPA and about 200 rpms from where the real knock occurred that it would start pulling timing out when you approached that "real" knock point. This is why eventually it was pulling timing almost anytime you were accellerating.

So what I did was I then started removing timing at the exact points where the "knock counts" increased and no where else no matter how much retard there was.
After several iterations of this, about 8 times I now get no kncok counts and no retard with or without counts now! Granted, I did end up taking out 5-8 degrees over stock at the cruise ranges say 1400-2000rpms and 50-75 KPA, but everywhere else I have about 3-5 degrees more than stock currently!
So now its on to WOT tuning with the wbo2 and seeing what more spark can be added there.

 

That is really interesting. If, as Fred surmises, the knock counts are calculated, then I would think the calculations would be accurate for stockish compression ratios and cams, but less effective the more you deviated from stock. Predicting knock from the available sensor info and a lookup table or algorithm that maps knock conditions seems simplest and most likely. However, the fact that you observed knock retard BEFORE the knock counts seemed to affect timing, and only after real knock do they correlate well with retard, seems to imply that the car LEARNS when it knocks, remembers the conditions, and builds it's own map of when to increment knock counts. I'd like to think that is possible, but I doubt the PCM is that smart. It would be pretty good evidence that is the case, if you tried again with a different octane, and saw different conditions triggering knock counts without retard.

My car frequently has KR without KC, and the detonation tendencies are pretty far off stock, so I doubt it learns.

 

Actually NO, once the PCM is reset, I would never see retard until the first time the knock counts increased and timing was removed...then only after that would I see retard without the counts increasing at around the range where the knock counts increased the first time.

In all the recordings I've looked at with various motors, TPIs, LT1s, LC2s, and TBIs, only when knock counts increase is when "real" knock is occurring, otherwise if you are seeing retard without counts, then the PCM must be calculating and antisipating knock based on past conditions. "Real" knock meaning something being detected by the knock sensor like detonation, loose motor mount, banging exhaust etc

 

How does the PCM determine you are using low octane fuel? Supposedly it "learns" and runs a low-octane timing routine. You may have found out how it does that. I've always heard about the low-octane "learn", but i never saw anyplace that is was indicated the PCM was in low-octane mode.

Unfortunately, my "learning" on the stock PCM's ended 7 years ago, when I switched to the MoTeC ECU.

 

I suppose it is "learning" when the knock was and is antisipating it assuming it was caused by low octane fuel in the lower cruise rpms like that, but it seems to never reset itself unless the PCM is reset..even after shutdown, the next startup continued to remove tons of timing with no knocks.
Seeing the "Low Octane" multiplier tables I did lower the multipliers somewhat and even set them to 0 but it still did the same thing although it didn't seem as bad.....but maybe setting the multiplier to 0 is equivalent to saying "leave the computed amount alone " like multiplying by "1" since none of the multipliers were higher than 1 to begin with I don't think..... Didn't try to make them negative if it would even let me to see if it would "add timing" instead of removing

What is odd is I've never seen this phenomenon in the truck PCM and its a OBD1 '95 PCM too.

 

Ok, thanks for clarifying. So, knock counts would correlate to the sensor being triggered, and retard would be based on calculations. The opposite of the way it is widely assumed to work. Nice observation based on what you have witnessed, and it could explain why my retard completely goes away sometimes for a while after a reflash. Tuning our advance for counts instead of retard would be less agressive, I think. I get retard alot more frequently than counts.

If you are correct, mapping counts to the timing curves would thus correlate better to engine needs than retard. Using retard would yeild an over-conservative spark map. I think.

 

Yes, for 2 reaons.
1) If the sensor never detected knock, then timing should not be removed in the first place. This is why there is no retard after resetting until once knock is detected and knock counts increased. It appears that once it sees knock counts it then calculates a range to begin removing timing when getting in that "range" in order to prevent knock at that point again...thus where you see retard without counts.
2) If you get say 6 degrees of knock, you should never remove a full 6 degrees of timing in your table. The timing may only be 2 degrees too advanced but can cause X times the amount of retard...therefore I only remove 1-2 degrees at a time in at the direct point or frame just before the knock counts, and maybe 0.5 degrees in the area just before that.