Knock sensors and knock detection

[Ludikraut]

I'd like to start a discussion on knock sensors and knock detection. I'm basically trying to tune a heavily modified 2.0l 4g63 and would like to be able to make some educated decisions as far as attenuating the knock filter maps.

What I've learned so far:
- Looks like the Evo knock sensor is a piezoelectric microphone
- Looks like the stock ECU knock filter maps use RPM and Load axis and are populated with Volts (presumably some kind of threshold for the 0-5 V signal from the knock sensor)

I've attached an Excel spreadsheet that contains the 12 knock filter maps from a stock 2003 ECU ROM, including a graph for each one.

What I'd like to determine:
- Why 12 knock filter maps? What condition(s) would determine the use of one map over another?
- What frequency is the stock knock sensor tuned to?

l8r)

[SaabTuner]

Originally Posted by Ludikraut

- What frequency is the stock knock sensor tuned to?

l8r)

It varies with crankshaft angle, combustion chamber design, and cyllinder size. Here are the typical knocking frequencies for a pent-roof cyllinder, which I believe is of around 500cc/cyllinder, relative to crank angle:



There are also various resonant modes the knocking can take on. This image shows most of the modes for a typical pent-roof combustion chamber:



Lots more information here: http://www.vehicular.isy.liu.se/Publ...PhD_580_LE.pdf

And here: http://www.vehicular.isy.liu.se/Publ...LIC_962_IA.pdf
(carefull look at the Yoshiyama-Tomita model, if Ion-current modelling is your bag)

Also, feel free to search the old NACA (aka NASA) archives for "engine knock" or "detonation" here: http://naca.larc.nasa.gov/

Lots of good documents there. TR912 is of particular interest. In either case, whole buttloads of good information on knock both in general and in specific. Lots of high-speed Schliren photography of knock, interesting conclusions and unusual methods. It's old, but interesting notwithstanding.

-Adrian

[DynoFlash]

Good data