Hide minor edits - Show changes to markup
April 24, 2007, at 03:33 AM
Added lines 10-15:
Two posts later…
The knock control system is active from ~2000rpm to ~6000RPM, give or take a few 100rpm based upon application/mapping. It’s also active above moderate load (under boost). It is very unusual to hear audible ping within those conditions with deltadash consistently showing no drop-outs in knock correction. It may be possible that the knock correction isn’t taking out enough timing do totally surpress knock but I have yet to see the detection ability of the system to be amiss.
BTW, if you try filling up a stock car with 87 octane, it will ping instantly, pull back timing, send the ignition advance multiplier well below 4 which will make the car run on its rich fuel map. For the most part, it will run ping-free. But every now and then, it will try to advance back to normal. And in doing so, it may ping audibly.
April 24, 2007, at 03:31 AM
Added lines 1-9:
Knock Sensors, KC, and IAM
Originally posted by Shiv at NASIOC.
The code at our disposal allows us to change the rate of retard, rate of advance, operational load range, operational rpm range, learning authority range, etc,. However, the basical code that distinguishes against noise and knock is unchanged as its a bit more sophisticated than a volume threshold above which noise is interpreted as knock. It’s been remarkably effective in almost all the applications I’ve seen allowing us to keep knock sensors functional under a lot more conditions than I’ve been able to get away with aftermarket programmable systems which have been historically too sensitive or not sensitive enough at the same time (depending on engine load/rpm conditions).
Also, just because the factory knock control system doesn’t react to a few pings one time, doesn’t mean that its ineffective or deaf to detonation. From what we’ve been able to tell, it simply shifts its dynamic thresholds up and down based up on prior knock activity. It is very likely that the next time the car is run through that RPM/load point, that it will pull back timing and store the knock correction levels in memory. If it does this in enough places in the table, the advance multiplier will drop and stay dropped until its sees consistent maxed-out positive knock correction. This is how it responds to knock in octane limited situations. Frequent resetting of the ECU will erase this knock learning and I have yet to see this reproduced in the aftermarket with any success. It’s not perfect but it’s better than anything I’ve seen.
At the end of the day, if knock sensors were perfect, there would be absolutely no reports of engine failure as knock would never occur.