Knock system

V6sleeper

Well-Known Member
I plan to do work on this when the car comes back out. That is the reason for watching the signal from the ESC to the ECM. I have checked over the car pretty well, not to say I did not overlook something. My curiosity is what kind of noise level is being picked up vs what is the ECM recognizing as knock.
Also check to see if the knock sensor is not over tightened , that will cause over sensitive readings.
 

Pronto

You can't knock them up with spit.
I plan to do work on this when the car comes back out. That is the reason for watching the signal from the ESC to the ECM. I have checked over the car pretty well, not to say I did not overlook something. My curiosity is what kind of noise level is being picked up vs what is the ECM recognizing as knock.
It's not the noise level. The knock sensor measures a certain frequency that has been determined from the factory. Anything that happens to be at that frequency near the knock sensor can cause the signal to be picked up and sent to the ESC. The strength and duration of that frequency is the determining factor.
 

RmvBfrFlght

Well-Known Member
My curiosity is what kind of noise level is being picked up vs what is the ECM recognizing as knock.

The knock sensor is just a microphone. Usually they do have an area that is more sensitive to a particular frequency.

The real brains is the knock filter (ESC module). It is an acoustic filter that checks for two things. First is a level of noise (energy) at a particular center frequency. For the '87 filter (HKP) this is at 5900 Hz.

The second thing that the filter does is to note a sudden change in the frequency of the noise. That too will cause the filter to report knock to the ECM.

It is best to get a small audio amplifier and a speaker. Then tap into the knock sensor signal that goes to the knock filter. This way you can hear what is being picked up by the sensor.

RemoveBeforeFlight
 

Nigel

Well-Known Member
The knock sensor is just a microphone. Usually they do have an area that is more sensitive to a particular frequency.

The real brains is the knock filter (ESC module). It is an acoustic filter that checks for two things. First is a level of noise (energy) at a particular center frequency. For the '87 filter (HKP) this is at 5900 Hz.

The second thing that the filter does is to note a sudden change in the frequency of the noise. That too will cause the filter to report knock to the ECM.

It is best to get a small audio amplifier and a speaker. Then tap into the knock sensor signal that goes to the knock filter. This way you can hear what is being picked up by the sensor.

RemoveBeforeFlight
Thanks for that info, interesting. So no real value in monitoring the signal to the ECM? The signal out from the ESC would not look any different if it was due to a change in noise frequency vs a noise spike from detonation? I guess tapping on the block sets it off due to the sudden frequency change?
 

Jerryl

Tall Unvaccinated Chinese Guy
It is currently about 10 ft-lbs.
Retorque the sensor to 14 ft-lb with a calibrated wrench, and clean threads (ie. Not gooped with tape)

Loosening the sensor is an old skool trick that can have the opposite and adverse effects, because the ground circuit is compromised.
 

Mike T

Well-Known Member
After the rebuilt of my current engine it showed minimal knock at odd times. Knock would show up at mid load and relatively low boost (maybe 12 psi).
This totally puzzled me because the same basic combo had ran as much as 22 psi on straight 93 and had zero knock.

Looking back I believe it was due to oil consumption. It didn't smoke but for whatever reason it took a long time for the rings to totally seat.
It used a quart in the first 300 miles and it wasn't until roughly the 5000 mile mark that oil consumption went totally away.

IIRC knock took care of itself between 1000-1500 miles. Some will argue that rings will either seat right away or not at all....my finding were different.
 

Mike T

Well-Known Member
"rebuild" not "rebuilt". Corrected that twice and it wouldn't stick, now edit timed out.

I see the benefit of not being able to edit but it's rough for us that are grammatically challenged.
 

RmvBfrFlght

Well-Known Member
Thanks for that info, interesting. So no real value in monitoring the signal to the ECM? The signal out from the ESC would not look any different if it was due to a change in noise frequency vs a noise spike from detonation?

The signal from the filter to the ECM is the same no matter what. It just pulls the signal line low to report knock to the ECM. The duration that the signal is pulled low can and does vary. This is why the actual level of retard can be low or high.

There are parameters in the calibration that set the maximum allowed retard along with attack and decay rates.

I guess tapping on the block sets it off due to the sudden frequency change?

Most likely, same as tire spin.

RemoveBeforeFlight
 

Nigel

Well-Known Member
Doing a little goofing off, just to get an idea of how the system works. I do not have any serious results, but just doing some investigating.
I picked up a cheap frequency generator and a small device to turn a laptop into an oscilloscope. I've also recorded the knock sensor with the laptop using some acoustic freeware. Overall, all this shit is painful. It involves acoustics, electronics, controls, and knowledge of how an engine works(which I know just enough about all of these to do some damage). Ultimately, the stock system is not going to save a high performance engine, but if you are logging you may get some info if you it blow up to maybe know why. I have been thinking about a J&S safeguard mainly just to be able to get some confidence in recording knock and staying away from it.
I have seen enough information to think that I am getting false knock. No need to go into did you check this or that unless it is something really esoteric.
I run conservative timing, a reasonably conservative A/F (I actually worry about being too rich). My goal is to simply make the stock system work, or to go aftermarket. I also understand some people run no knock detection system, and with good fuel and a dyno tune that monitored knock during the tune I would be OK with that.

First, an acoustic snap shot of WOT with stock knock sensor:
View attachment View attachment

View attachment
View attachment
Absolutely no spikes or sounds of knock, frequency spectrum is interesting but nothing on many runs even suggested knock. Ear sensor also cannot detect knock after multiple playbacks on many runs.

I did a simple mapping of the ESC, nothing to report here either. A step change from no volts to peak to peak volts provided as to what will cause the ESC to send a signal to the ECM. Above the blue line, ESC sends signal, below no signal. This chart is very simple, and I want to try and expand upon this to get more understanding. i.e. If I am at a steady voltage(noise) input which my cheap junk can't simulate, and have the same size step, will it still set off the ESC.
View attachment View attachment

A snap shot of the oscilloscope, and a step frequency input in red at 5900hz and the ESC output response in yellow as an example.
View attachment View attachment

Last piece, the aparatus, some serious NASA shit, the components in the bread board are not used yet other than to directly wire the ESC, freq generator, and oscilloscope.
View attachment

Anyway, not sure where this is going, but my goal is to pick apart the knock system. Last bit of info, the knock sensor will create a large voltage spike when struck with a butter knife handle, like 48V instant spike.
 

grass doctor

Dream Catcher
Doing a little goofing off, just to get an idea of how the system works. I do not have any serious results, but just doing some investigating.
I picked up a cheap frequency generator and a small device to turn a laptop into an oscilloscope. I've also recorded the knock sensor with the laptop using some acoustic freeware. Overall, all this shit is painful. It involves acoustics, electronics, controls, and knowledge of how an engine works(which I know just enough about all of these to do some damage). Ultimately, the stock system is not going to save a high performance engine, but if you are logging you may get some info if you it blow up to maybe know why. I have been thinking about a J&S safeguard mainly just to be able to get some confidence in recording knock and staying away from it.
I have seen enough information to think that I am getting false knock. No need to go into did you check this or that unless it is something really esoteric.
I run conservative timing, a reasonably conservative A/F (I actually worry about being too rich). My goal is to simply make the stock system work, or to go aftermarket. I also understand some people run no knock detection system, and with good fuel and a dyno tune that monitored knock during the tune I would be OK with that.

First, an acoustic snap shot of WOT with stock knock sensor:
View attachment 359876 View attachment 359876

View attachment 359877
View attachment 359877
Absolutely no spikes or sounds of knock, frequency spectrum is interesting but nothing on many runs even suggested knock. Ear sensor also cannot detect knock after multiple playbacks on many runs.

I did a simple mapping of the ESC, nothing to report here either. A step change from no volts to peak to peak volts provided as to what will cause the ESC to send a signal to the ECM. Above the blue line, ESC sends signal, below no signal. This chart is very simple, and I want to try and expand upon this to get more understanding. i.e. If I am at a steady voltage(noise) input which my cheap junk can't simulate, and have the same size step, will it still set off the ESC.
View attachment 359878 View attachment 359878

A snap shot of the oscilloscope, and a step frequency input in red at 5900hz and the ESC output response in yellow as an example.
View attachment 359879 View attachment 359879

Last piece, the aparatus, some serious NASA shit, the components in the bread board are not used yet other than to directly wire the ESC, freq generator, and oscilloscope.
View attachment 359880

Anyway, not sure where this is going, but my goal is to pick apart the knock system. Last bit of info, the knock sensor will create a large voltage spike when struck with a butter knife handle, like 48V instant spike.
my man.....
 
Top