Knock system

SCOOBY DOO

I'M NOT A MONSTER, I'M JUST AHEAD OF THE CURVE!!!
The weather here is deteriorating quickly, so I don't expect to get a whole lot done this year. My plan is to get things set up so I can hook the oscilloscope up to the car and record some WOT runs. I will want to record sensor and ESC output. I think I am going to get a simple screen recorder to grab the scope output. I know better than to say stuff with certainty, but I am confident that I am not knocking.

Like I mentioned earlier, I am also seriously considering a J&S Safeguard. By all accounts this is a quality product. I have no plans to push timing to the knock limit, it would be nice to have a quality product to know that I am not knocking, and it could possibly react quick enough to pull timing if a malfunction occurred.
Nigel. Contact Norbs here on the board. He can help you with the J&S Safeguard. Tell him Brad said hello.
 

Nigel

Well-Known Member
What is J&S Safeguard?
It is a stand along device that detects knock and will pull timing. You can google seach for it and get details. I believe in a "safe" tune, so I would not try to run to the edge on timing and have the device pull timing. I believe though, once you get it set up you can get an acurate indication of knock or lack there of. It also offers protection that the stock system can't match, so it would be extra insurance in the event of something else failing.
 

mommasGN

Active Member
Have you looked into any modern knock sensors? Say, post 2000? Even though a knock sensor still operates the same, have there been any improvements in one designed for a newer car versus one from the 80's? I'm sure it wouldnt be compatible with the buick ESC, but maybe for Fast/holley/ECUgn guys?
 

Nigel

Well-Known Member
The stock ESC is limited to a matching resonant 1 wire sensor. I am sure some of the aftermarket stuff can use a modern wideband sensor. That said, you have to know how to set up the ECU for your particular engine, bore size, sensitivity .... Stakes are high if done wrong.
 

Nigel

Well-Known Member
Minor update:
For some of the testing I purchased a cheap function generator where you can set the frequency and it has knobs for peak to peak voltage. That is what I used to chart the frequency vs volt in an earlier post. I rigged up a button and second potentiometer so I could have a steady state input and then step the peak to peak voltage. Essentially, I have the same results as going from zero to some set voltage. So the earlier chart essentially represents a step change from a quasi-steady state. See shot below. Red is the input to ESC and Yellow is the output to the ECM.

So far from what I can tell (plan to do some testing with the car come spring) using the test set up:
The ESC does not care what the steady state voltage (noise) input is. No steady state noise, no matter the level, will set it off.
Only a sudden step voltage (noise) will set it off. How long it takes to establish the steady state baseline I do not know, test apparatus is not sophisticated enough to tell, but it is on the order of 10s of milliseconds. Once the baseline is established a sudden impulse will set it off and send a signal to the ECM. It is most sensitive at 6K hz and sensitivity exponentially decreases as you move away from 6K hz as shown previously.

View attachment
 

Nigel

Well-Known Member
Try a sudden change in frequency without changing the voltage. That too should set it off.

RemoveBeforeFlight
So I cannot generate a change with a frequency shift, I believe and I may be wrong, but you simply have to exceed the delta in voltage required for the given frequency. The ESC appears so far to be a very simple requirement. But, I have hammers and chisels where a scalpel may be required.
View attachment
 

Nigel

Well-Known Member
So just to summarize, because I think my last post was worded poorly. I was able to simulate a frequency shift at roughly a steady voltage and saw no reaction from the ESC. Frequency of the input does play a part, but just how sophisticated the ESC is when it comes to frequency is still a mystery to me, other than what I have posted so far.
So based on what I can tell, the ESC will accumulate an input and develop an average signal input in terms of voltage. This seems to take anywhere from 100 to 200 milliseconds (hard to say exactly). If a sudden increase in the input occurs within a few milliseconds the ESC will signal that as knock. This sudden increase has to occur anywhere from 0 to roughly 20ish milliseconds (I don't really have the equipment to define) and has to exceed the threshold to set off the ESC. Any sudden shift in noise from the baseline average will trigger the ESC if it is large enough. For instance, my guess is one noisy lifter or valve that hammered hard enough would be enough, but if the RPM was high enough the noise would be averaged out and not set off the ESC.
My thought is still that a filter could make the ESC useful for a diagnostic, it could never be made good enough for protection on a high-performance build.
My goal in the long run is to figure out a good filter to not see false knock, but if I do get knock at the very least I want to be able to see it in a log even if I do damage or blow something up.
 

TurboTGuy

Gray Beard Member
So just to summarize, because I think my last post was worded poorly. I was able to simulate a frequency shift at roughly a steady voltage and saw no reaction from the ESC. Frequency of the input does play a part, but just how sophisticated the ESC is when it comes to frequency is still a mystery to me, other than what I have posted so far.
So based on what I can tell, the ESC will accumulate an input and develop an average signal input in terms of voltage. This seems to take anywhere from 100 to 200 milliseconds (hard to say exactly). If a sudden increase in the input occurs within a few milliseconds the ESC will signal that as knock. This sudden increase has to occur anywhere from 0 to roughly 20ish milliseconds (I don't really have the equipment to define) and has to exceed the threshold to set off the ESC. Any sudden shift in noise from the baseline average will trigger the ESC if it is large enough. For instance, my guess is one noisy lifter or valve that hammered hard enough would be enough, but if the RPM was high enough the noise would be averaged out and not set off the ESC.
My thought is still that a filter could make the ESC useful for a diagnostic, it could never be made good enough for protection on a high-performance build.
My goal in the long run is to figure out a good filter to not see false knock, but if I do get knock at the very least I want to be able to see it in a log even if I do damage or blow something up.


It's a lofty goal, for sure.

Way back when this thread started, I said I disconnected the stock knock sensor because I was getting 15 plus degrees of false knock on my newly built stroker, roller engine. We've (my builder and several very well versed TB guys) come to the conclusion that it's just a plain noisy engine.

The sensor is still disconnected. It's still doesn't knock at 23lbs plus boost. I don't think there is a solution for my particular issue, one that I'm certain many face, but I would love to see one.

Good luck!
 

Nigel

Well-Known Member
It's a lofty goal, for sure.

Way back when this thread started, I said I disconnected the stock knock sensor because I was getting 15 plus degrees of false knock on my newly built stroker, roller engine. We've (my builder and several very well versed TB guys) come to the conclusion that it's just a plain noisy engine.

The sensor is still disconnected. It's still doesn't knock at 23lbs plus boost. I don't think there is a solution for my particular issue, one that I'm certain many face, but I would love to see one.

Good luck!
I'll keep at it and repot back if I figure anything new out. I can definately see how a noisy engine would set off the ESC.
 

RmvBfrFlght

Well-Known Member
It's a lofty goal, for sure.

Way back when this thread started, I said I disconnected the stock knock sensor because I was getting 15 plus degrees of false knock on my newly built stroker, roller engine. We've (my builder and several very well versed TB guys) come to the conclusion that it's just a plain noisy engine.

The sensor is still disconnected. It's still doesn't knock at 23lbs plus boost. I don't think there is a solution for my particular issue, one that I'm certain many face, but I would love to see one.

Good luck!

You can desensitize the knock sensor by mounting it on a short piece of iron pipe (or a street elbow). I did this as a manual 1st and 2nd gear would continuously set off the ESC system (high line pressure == noisy trans). This may not be the best way to go but at least the ESC system is still in place.

RemoveBeforeFlight
 

Nigel

Well-Known Member
You can desensitize the knock sensor by mounting it on a short piece of iron pipe (or a street elbow). I did this as a manual 1st and 2nd gear would continuously set off the ESC system (high line pressure == noisy trans). This may not be the best way to go but at least the ESC system is still in place.

RemoveBeforeFlight
Did you hve trouble with fitting it with the wire harness? I thought about that, but not much room with the factory harness.
 

RmvBfrFlght

Well-Known Member
Did you have trouble with fitting it with the wire harness? I thought about that, but not much room with the factory harness.

It is a custom fit harness based on the factory harness. Not sure how tight the stock harness is. Picture in next post shows that it doesn't lift the sensor up too much.

RemoveBeforeFlight
 

RmvBfrFlght

Well-Known Member
Innovative fix.......

Do you have a picture? :)

Got one, note that this isn't the stock location, the pipe plug is the stock location (bottom/center of pic). Previous owner would swap between a modified stock ECM and a '7749 ECM. So had two different knock sensors installed. One for the stock external knock filter and another for the '7749 internal filter.

As for the stand off, it is a short piece of iron pipe with tapered pipe thread on each end. Then a pipe threaded coupler between it and the sensor.

View attachment

RemoveBeforeFlight
 
Top