im Merlins ecuflash guide this method is stated as one of the good way for determining false knock. spots where Knockadc(pink) spikes over Knockbase(yellow) are real knock, other is just a mess of engine noise. anyone uses this method succesfully ?


image host

 

any input ?

 

I tried to use it one time with no result.
As I understand, logging speed it too slow, so I've managed to get a clear knock image only once. Knock Sum can be 20-25, but Knock ADC won't cross Knock Base. I think Knock ADC changes very rapidly.

 

nearly same results for me too, but maby that's phantom knock?
interesting if someone has double checked it via det cans while logging knockADC and knockBase

 

Whether it is phantom or not, Knock ADC must cross Knock Base. This is the raw data to calculate Knock Sum.

 

The idea is to take notice of the average noise floor vs. the KnockBase line. If it appears that the noise floor is very close to the base line at all times, the engine may be noisier overall. If you see giant spikes, those can still be false knock since the knock sensor can't tell you if it's from a brace hitting the downpipe or a true knock event. Also, lowering boost, changing to a higher fuel octane and minimal ignition timing can help prove/disprove false knock.

At first glance of the sample graph, it looks as though the knock multiplier LOW and MID rpm ranges are a bit noisy, though the knock multiplier HIGH looks like its leaving plenty of noise floor headroom. Again, this would a situation that needs to be confirmed before making adjustments. It seems that on most standard logs I have seen, knockFILTADC is, on average, half of KnockBase.

-Jamie

 

EvoLive map is the way forward here. You can get hundreds of samples a second with the DMA logging. Live map thread

 

If only 89970010 or 80700010 have DMA...

 

You could ask donner and ziad if they could do it. I know they are both pretty busy, but you ever know.

 

I agree, DMA is the best way to make use of this.

-Jamie