In a WOT pull tonight, I ended up with a scary amount of knock. This is in an area where the timing is stock. Is each knock even real and distinct (meaning the stock map has WAAAY too much timing in this region), or can one knock event set off a whole chain of other knocking if the tune is kind of on the ragged edge? Log below snipped to show relevant time span, don't worry about AFR, I only have the stock AFR sensor so it would have been jumping up and down going "Look at me! Look at me! I'm pegged rich!" for the entire pull.


Time RPM g/rev Thrt% FBKC FLKC IAT RelPres
143171 4405 1.16 100 0 0 72 -1.06
143421 4472 1.16 100 0 0 72 -1.05
143671 4536 1.16 100 0 -1.05 72 -1.06
143921 4589 1.17 100 0 -1.05 72 -1.07
144171 4652 1.16 100 0 -1.05 70 -1.07
144421 4692 1.16 100 0 -1.05 70 -1.1
144687 4748 1.16 100 0 -0.7 70 -1.07
144906 4822 1.16 100 -1.05 -0.35 70 -1.08
145156 4885 1.15 100 -1.05 -0.35 70 -1.09
145390 4918 1.15 100 -1.05 -0.35 70 -1.11
145625 4992 1.16 100 -1.05 -0.35 70 -1.11
145859 5049 1.15 100 -1.05 -0.35 70 -1.13
146093 5104 1.16 100 -0.7 -0.35 70 -1.16
146343 5149 1.16 100 -0.7 -0.35 70 -1.14
146578 5192 1.15 100 -0.7 -0.35 70 -1.18
146843 5239 1.16 100 -0.7 -0.35 68 -1.17
147078 5288 1.16 100 -0.35 -0.35 68 -1.17
147312 5353 1.16 100 -0.35 -0.35 68 -1.19
147578 5388 1.16 100 -1.4 -0.35 68 -1.2
147812 5434 1.16 100 -1.4 -0.35 68 -1.21
148062 5503 1.17 100 -1.4 -0.35 68 -1.23
148312 5529 1.16 100 -2.45 -0.35 68 -1.23
148546 5567 1.16 100 -2.45 -0.35 68 -1.27
148781 5612 1.16 100 -2.45 -0.35 68 -1.27
149015 5653 1.16 100 -2.45 -0.35 68 -1.28
149281 5695 1.16 100 -2.45 -0.35 68 -1.29
149531 5752 1.15 100 -2.1 -0.35 68 -1.28
149765 5772 1.16 100 -2.1 -0.35 68 -1.33
150015 5821 1.15 100 -2.1 -0.35 68 -1.31
150265 5875 1.15 100 -2.1 -0.35 68 -1.36
150515 5922 1.14 100 -3.15 -0.35 68 -1.36
150765 5946 1.15 100 -3.15 -0.35 68 -1.32
151000 6000 1.12 100 -3.15 -0.35 68 -1.35
151218 6037 1.13 100 -3.15 -0.35 68 -1.39
151484 6059 1.12 100 -2.8 -0.35 66 -1.39
151734 6092 1.11 100 -2.8 -0.35 66 -1.4
151968 6138 1.09 100 -3.85 -0.35 66 -1.37
152218 6174 1.09 100 -3.85 -0.35 66 -1.41
152468 5978 1.05 100 -3.85 -0.35 66 -1.17
152703 6204 1.02 100 0 -0.35 66 -1.2
152937 6066 0.37 16.86 0 0 66 -8.51

No. From what I see.

It knocked once at 4800 rpm, it pulled 1 degree and started to add it back.
Another knock at 5388, 1 more degree of timing pulled on feedback.
Another knock at 5529, 1 more degree on feedback. starts adding it back
Another knock at 5900, 1 more degree pulled on feedback. Again starts adding it
And then one more knock at 6138.

You need to fix your timing at each one of those knock events.

That's how I read it too, but I wanted to confirm. Like I said, this is on the stock timing on 94oct on an RS. It's amazing to me that the factory tune is knocking this badly.

factory tunes have knock retard all over the place - mine had a lot more than 3.8 degrees.

I would reduce timing through that whole range (assuming fuel is on target) to keep an overall smooth 3d shape. Knocking timing out in whole degrees at individual steps may make it feel jerky when travelling through the range on a WOT run. I would probably start with 0.7 degrees unless it is doing this on every run through.

to answer your question (even if it didn't necessarily happen in THIS particular case), yes.

knock tends to have a "hysteresis." in other words, there is a dependency on past events. once knock starts it tends to keep going.

as an example, say you can run 15* of advance with no knock whatsoever. in the quest for "mo powa" you try to dial in another degree. the car knocks and the EMS by some mechanism pulls one degree, which nets 15. the car still knocks. the EMS must dial it back to 14 degrees to STOP the knock. once things are stable the EMS may or may not dial back in some advance. 15 will likely work just fine again (it did before), but if it goes to 16 then it'll be necessary to drop to 14 once again to quell it.

of course if the EMS is operating properly this all happens in the fraction of a second. nevertheless the fact remains that an operating condition that runs without knock may NOT be stable IF knock just occurred.

fwiw
ken

OK, that's what I was wondering.

What's the physical process that produces this "hysteresis of knock"?

hard to say. :)

i think it boils down to the instability of the conditions in the chamber after the knock event. a lot of times knock will result in an incompletely burned charge which may leave behind much more reactive compounds than a "clean burn" would have. on the next cycle they blow up, and the process runs away.

Knocking timing out in whole degrees at individual steps may make it feel jerky when travelling through the range on a WOT run. I would probably start with 0.7 degrees unless it is doing this on every run through.

Yeah, the knock control system by default pulls 1 degree but that doesn't mean you weren't less than 1 degree off target. Start from the top and work your way down.

Not in this case (log), but knock does cause more knock because of extreme heat during the initial knock event.

not to mention that the sharp increase in heat is more likely to create "short term hot spots" on the piston.

An ADI system that constantly re-mapped itself based off the FLKC table would be sweet.

I have seen behavior where fixing a knock even at lower RPM will fix behavior where I see multiple knocks during a pull through a wider RPM range. Say, pulling 2 degrees in the 4400-5000rpm 2.6-3.0 load region where I saw a knock event will avoid the problem where the engine will see a few knock events throughout a pull from 4000-6500rpm even if timing is not touched from 5000-6500rpm.

This would lead one to believe the extra following knock events were caused by the first one, and not by the timing or fueling during the later knock events. I think this is a reasonable conclusion.

An ADI system that constantly re-mapped itself based off the FLKC table would be sweet.
:evileyes:

/me dons his noob hat

What's ADI?

Anti-Detonant Injection (secondary water/meth system).

/me dons his noob hat

What's ADI?

it's a term that Ken (Ride5000) had to beat into people's skulls for a solid 2 months. :lol: