Knock Control Problems
 

Well after weeks and weeks of reading and searching on here, I am starting my own thread. I have been working with a local tuner and now Mellon to get my car tuned. The car has anywhere from 0 counts of knock one run, up to 13 the next run. This is all WOT runs. Mellon and I decided it is phantom knock after the car still acts the same with c16 and does not really respond is timing changes. The problem with adjusting the knock threshold is that it never happens in the same rpm. It will be random from 2500 to redline. Car has BBK turbo and full bolt ons including a lightened flywheel. Has anyone ever had a problem like this before?

Thanks

Try logging knockbase, knockadc, knocksum, RPM on a WOT run, do one short log file for each WOT run, then post them here (plot all three items vs RPM on an Excel scatter graph with one axis for the knockbase and knockadc, and another axis for the knocksum). If you only log these four items and disable logging to screen you should be able to get 50 items per second. You still won't catch every ignition event, but you'll catch as many as possible and we will be able to assess the noise profile of your engine. If you want some reading see my recent knock thread. Mellon has played with this already and should be able to help you fix it. Of course you also need to try to remove any mechanical sources of the noise.

Thank you very much for your time, I will do as you said on Monday. Im away from the car until then. Also, I have been through the engine bay and underneith many times to try and find a cause of some noise. No luck with that.

Sometimes people end up finding a source inside their engine for such intrusive knock. A lot of false knock from stiff engine mounts and forged pistons seems to be at low load and about 3000-3500 RPM. Your logs will be interesting.

Ditto when my downpipe was rattling against one of the lower braces on the undercarriage.

I've tuned many EVOs and this is the only one that exhibits this behavior where it's at different RPM ranges. We tried richer AFR, less timing, race gas, less boost etc...nothing helps it become consistent.

I've had a few including my own that had problems in the 2500-4500rpm range but nothing like this at all. I'm out of ideas but subscribing to stay in the loop.

I believe you swapped in a brand new knock sensor too right Chris?

The thing is the car makes great power when the knock isn't cropping up on some runs. We stopped at 430awhp 370tq on 93 octane, so It seems to be healthy otherwise.

Chris and I talked about completely disabling or desensitizing the knock sensor since we believe it's false knock, but it would be very risky to do so.

It does sound different to the norm, even without seeing the logs. It may well be that a small increase in knockbase will allow it to still run knock control satisfactorily - depends on how much it is being breached at present. A non-timing responsive noise even just a little over the threshold will quickly accumulate, and you can make it disappear sometimes with just a 10% increase in knockbase, which will make little difference to the timing subtracted for real knock. Let's hope it is that easy.

If it were my own car I would be okay with the risk and tune the car to perfection based on the few clean passes which weren't affected by sporadic knock. Then I'd simply desensitize the knock sensor to a point where it's just enough to do the job.

Chris and I talked about this but I think he's really interested in finding the solution rather than tuning around it. I don't blame him but I have a bad feeling he may never find the culprit short of disassembling the engine and even then I'm not convinced it would be something obvious.

One customer of mine that had false knock in the 2500-4500rpm area ended up jumping time and wiping out the valves etc... He shipped the head off to Buschur and when it came back we couldn't reproduce the odd false knock. There was no stage 3 treatment done to our knowledge, just a simple freshening up. That suggested to me that his issue must have been in the valvetrain... cam/lifter/rocker/valves etc..

Sorry it took so long, Ive been having computer issues. I also had a brain fart and actually did 3 different logs, logging knocksum, base, adc, and rpm every log. I did at least have log to screen off. It picked alot up anyway I think. My computer wont let me do a print screen at the moment but I can email the 3 logs to whoever. So if anyone would post them up for me that would be great, just send me your email. If I need to go log all of that again the way you wanted I can.

Thanks

sounds like a metal on metal problem - ie down pipe OR your knock sensor is either loose OR not torqued up right.

I have checked everything I can think of, not the downpipe for sure. I replaced knock sensor with new one and torqued it correctly.

Got pics going now.
http://i148.photobucket.com/albums/s...enate/log1.jpg
http://i148.photobucket.com/albums/s...enate/log2.jpg
http://i148.photobucket.com/albums/s...enate/log3.jpg

It looks generally noisy. If you really are happy it is not knocking, I would suggest multiplying the knock multipliers (either the three values for an 8 or the table for a 9) by 1.2 and testing again.

I multiplied the entire table(evo ix) by 1.2. Here are 3 new logs.
http://i148.photobucket.com/albums/s...g1_knock12.jpg
http://i148.photobucket.com/albums/s...g2_knock12.jpg
http://i148.photobucket.com/albums/s...g3_knock12.jpg

Seems better below 4500 RPM. If you are still confident it isn't knocking (I would hear this in the cabin even without det cans if this was real knock and you didn't have a very loud exhaust), then you need to increase further at 4500-6000. Whilst you can't see every spike that is triggering the knock sums, there will be some doing it that aren't logged. You do have a slight issue at high RPM where if you increase it much more, then the sensor isn't miles away from saturating (if knock_base hits 255 you can't exceed it!)