Anybody ever put a scope on their knock sensor?

[Jasper]

Be careful not to put a mono jack into a stereo plug, or you can burn out one of the channels.

It is going to be alot easier to make sense of your recordings if you have some logged data to refer to. Are you sure you can't free up an analog input somehow to log the 0-5V output of the knock module?

[chris davey]

well I just got a stereo plug then. How do I know if the jack is mono or stereo?

AFAIK the MT doesn't take any additional inputs other than what it needs to run the engine but I will check.

Don't know if this helps with the previous questions but

In terms of filtering the knock signal, I use one high pass scientific filter at 4800 Hz, one low pass scientific filter at 12.4kHz, and a scientific band block filter from 6.2kHz to 9.6kHz, all with 66 (whatever the unit in Audition is) roll off (max). Although the primary knock frequency for EJ205 engine's 92mm bore is centered around 5.68 kHz, my experience says it ranges from about 5-6 kHz in its entirety. What I believe to be a first order harmonic (correct me if I'm wrong) is very noticeable from about 10-12 kHz.

[chris davey]

another good one

Engine knock noise is transient, which means it spreads its frequency content across the spectrum. If knock is severe enough, you'll see a thin vertical line in Gram. The freqency that corresponds to the resonant frequency within the combustion chamber (knock's fundamental frequency) is amplified by the resonance and is usually the loudest portion of the knock noise. This resonant frequency depends on cylinder bore and speed of sound within the combustion chamber. The resonance causes the sound to persist for a few milliseconds and its frequency falls during that period because temperature within the combustion chamber is falling, reducing the speed of sound. Harmonics of the fundamental frequency can also be observed if knock is severe enough and those harmonics will exhibit the same persistance and fall in frequency as the fundamental knock frequency.

[Jasper]

IIRC, a stereo plug in a mono socket will be OK, but not vice versa. I'm sure google will tell you more. The socket on both your knock module and laptop *should* be mono, but I would double check just in case.

[chris davey]

cool, going to test it out tonight I think I will just suck it and see. Worst case is a dead $100 laptop

[chris davey]

I used this at http://audacity.sourceforge.net/

Only 2.1mb.

The resonant check one is me tapping on it with a hammer. Appears to be lower than 6.8k like you guys. I am using a piezo transistor clamped on the block not the stock knock sensor. Would it be better to use just a knock sensor? You can actually hear the engine rev through this, is that the same as a knock sensor with a mic?

So would I be looking for resonant noises than slope down like other pics I have seen around the 4.5-5k mark then?

tapping


tapping close


engine running


Also, this program has pitch. Would this be of any use? Looks "similar" to the spectrum but seems to only show the strong signals.



(I didn't download adobe as this lappy only has 2gb left on it plus it would take ages. If I need to then I will though.

car running was from idle up to 3500rpm and then back down.

[Esteban]

Did you have any luck with this chris?

[amichie]

Does anybody know if Toyota use automatic gain control (AGC) in their knock sensor circuits?

Is simply attenuating the knock signal going to trick the ECU into thinking there is a smaller knock signal.

I have traced out the board in my ECU and found the knock signal is being amplified/buffered by two opamps (configured as inverting amplifiers) with fixed gain. The output goes to a big NipponDenso IC that I assume has an A/D converter in it amongst other things.

Question is I wonder if the processing is clever enough to figure out if the signal has been attenuated due to a bad connection or loose knock sensor.

[amichie]

I have put some variable gain on the knock circuit. I can now adjust it from the original gain of say 100% down about 15% of the original gain. At about 20% the ECU starts to complain and turns on the check engine light. Turn the gain back up abit and switch the engine on and off and the check engine light goes off again.

My actual knock signal is still seriously nosiy. There is a clear resonant tone at about 7.4kHz with lots of noise on top of it.

I think I may try another knock sensor and see if its any better.

Also.

With the engine switched off and ignition on, just a light tap on the intake plenum with a shifter spanner gives me about 2 Volts peak to peak from the knock sensor.

The actual signal is about 2 to 4 times the amplitude that is indicated in the picture below.

When you put your foot down the signal is much bigger again.

BTW I have absolutely no audible knocking/detonation.