LS3 Knock Frequency

[rrrocketman]

I need to know as accurately as possible, what the frequency of knock is for this engine. I've been trying to find it on google but the best I can manage is a theoretical value based on piston size. I was hoping someone would know the real value by having measured it or finding a knock waveform on the internet by someone else. Has anyone any evidence of what this value is in kHz? Thanks.

[rrrocketman]

Has anyone got any info on this? I'm unable to find any info on the technical specs of the knock sensor used on this car.

[bct10985]

Try hptuners forum

[rrrocketman]

Ok but I'm not sure why they would be more likely to know. The knock sensor is a piece of hardware and is not tunable anyway.

[JANNETTYRACING]

Originally Posted by rrrocketman

I need to know as accurately as possible, what the frequency of knock is for this engine. I've been trying to find it on google but the best I can manage is a theoretical value based on piston size. I was hoping someone would know the real value by having measured it or finding a knock waveform on the internet by someone else. Has anyone any evidence of what this value is in kHz? Thanks.

It varies by engine rpm and load and each cylinder is different.

Take thousands of hours of dyno testing at GM to build the Knock Threshold maps for each cylinder and each engine type.

The sensor is a simple microphone, what they must weed through is what each frequency generated means in terms of real or false knock at each load cell and rpm for each cylinder.

Ted.

[camguynj]

Originally Posted by JANNETTYRACING

It varies by engine rpm and load and each cylinder is different.

Take thousands of hours of dyno testing at GM to build the Knock Threshold maps for each cylinder and each engine type.

The sensor is a simple microphone, what they must weed through is what each frequency generated means in terms of real or false knock at each load cell and rpm for each cylinder.

Ted.

Hey Ted, I have a Dashlogic unit on my LS3 set to display Knock Retard when >1.0
The display occasionally appears for no more than a second under various rpms from 1k on up mostly when accelerating in 1st thru 4th even under light engine load (sometimes just about idle). Never got a CEL. Car seems to run fine. Stock tune with CAI ,PTB, Shortys, HF cats for most of 26k miles. Nothing but Top Tier 93 octane. High Octane fuel table fuse pull reset done at 1k mi. Thoughts?

[JANNETTYRACING]

Originally Posted by camguynj

Hey Ted, I have a Dashlogic unit on my LS3 set to display Knock Retard when >1.0
The display occasionally appears for no more than a second under various rpms from 1k on up mostly when accelerating in 1st thru 4th even under light engine load (sometimes just about idle). Never got a CEL. Car seems to run fine. Stock tune with CAI ,PTB, Shortys, HF cats for most of 26k miles. Nothing but Top Tier 93 octane. High Octane fuel table fuse pull reset done at 1k mi. Thoughts?

The computer is constantly searching for the highest timing value it can run in any cell so it is not unusual to see random knock retard under normal driving.

Ted.

[camguynj]

Originally Posted by JANNETTYRACING

The computer is constantly searching for the highest timing value it can run in any cell so it is not unusual to see random knock retard under normal driving.

Ted.

Thanks for the reply Ted, much appreciated! What would you say is an acceptable degree of
KR in this case?

[JANNETTYRACING]

Originally Posted by camguynj

Thanks for the reply Ted, much appreciated! What would you say is an acceptable degree of
KR in this case?

We see a up to 2 at times in light throttle cruise especially right after a fill up when the Knock learn resets.

Ted.

[camguynj]

Originally Posted by JANNETTYRACING

We see a up to 2 at times in light throttle cruise especially right after a fill up when the Knock learn resets.

Ted.

Good info THANKS AGAIN Ted !

[rrrocketman]

Originally Posted by JANNETTYRACING

It varies by engine rpm and load and each cylinder is different.

Take thousands of hours of dyno testing at GM to build the Knock Threshold maps for each cylinder and each engine type.

The sensor is a simple microphone, what they must weed through is what each frequency generated means in terms of real or false knock at each load cell and rpm for each cylinder.

Ted.

Fantastic... I sort of stumbled upon this conclusion myself when I saw a device for sale that would act as a standalone KR controller. It had what looked like an exponential curve for knock threshold... I don't know if that was amplitude or frequency or both.

PITA because I need a means of arresting my power adders if the car experiences knock but I can't access knock information in the ECU directly so I have to re-invent the wheel using my own electronics. So far my research has led me to believe that 5.6 kHz is a nice center frequency for the LS3 so I have designed a bandpass filter to block out everything +/- 500 Hz away from that. Then using a Fourier transform I can monitor the levels in that range. I'm sure it will pick up knock. The question is, how much will the calibration vary with RPM and load...

[dookie454]

Originally Posted by JANNETTYRACING

The computer is constantly searching for the highest timing value it can run in any cell so it is not unusual to see random knock retard under normal driving.

Ted.

No it's not constantly searching for the highest timing value it can run -
It is running whatever is defined in the High Octane spark tables, and the Knock System is constantly searching for knock and if it see's it then it will retard timing. I think this is what you meant.

If the sum of the tables say 20deg (there are a few of them), then it runs 20deg. If the knock system see's knock, it will immediately temporary pull timing and increment the Knock Learn which is a multiplier table that starts to use timing listed in the Low Octane table. When knock stops, spark retard fades out fast and the Knock Learn will start to go back to 0.00 which slowly put it back to whatever the High Octane spark tables say. 1.00 is the max value for Knock Learn, if it's at 1.00 then it's using 100% of the low octane table (along with any other modifier tables).

So what I have found most useful rather than monitor just knock retard, is monitoring the Knock Learn table. If it's 0.00 then I know there has really been no knock to be concerned about. This is a longer term table which makes it very easy rather than trying to watch the instantaneous Knock Retard table constantly. I actually have a LED light come on when my Knock Learn gets over a preset value of say 0.2 - this means I've had a fair amount of knock recently.

The Knock Learn table can be adjusted to ignore knock retard below any value, stock is around 1.5deg, so if it's showing 0.00 then I know whatever knock I have is less than 1.5deg and that is pretty good really. You will normally get small amounts of knock at random, 0.2deg, 0.7 deg, etc, it is not because the ECU keeps adding timing until there is knock... this does not happen. It is happening either due to mechanical noise or the timing maps are set very close to the edge of real knock and your actually getting just a bit of real knock. I would be concerned if it shoots up to say 5-8deg more than one time... but the Knock Learn table will increment pretty fast of this happens so for times that your not datalogging, this is a great thing to use.

The problem with this is, the Knock Learn table is responsible for making your car feel like a dog at random.. if it's incrimenting and you dont know it, then your running severly reduced timing and it should actually be given more attention than it is. If you are at a higher Knock Learn, then most likely your not getting Knock Retard because your far into the Low Octane Table, it takes many miles and minutes for Knock Learn to start coming down back to 0.0 how you want it. If you constantly have higher Knock Learn value then you need to resolve that either by reducing timing in the High Octane Tables or other tables, or using better gas, mechanical problem, etc.

If I see 0.40 in Knock Learn then I know it's knocking pretty good for a bit and maybe I should make corrections or try better gas. If i was only monitoring Knock Retard then if I blinked, I would miss it.

For knock tuning, I personally prefer to make the High Octane and Low Octane tables both High Octane values, then leave Knock Learn active. This way I'm getting a realtime historic Knock Learn value but if it's incrementing I am still logging accurate knock history in the histograms that are not getting overwritten by knock free data from having timing reduced from running the Low Octane table. When done, I put the Low Octane table back to stock.

[JANNETTYRACING]

Originally Posted by dookie454

No it's not constantly searching for the highest timing value it can run -
It is running whatever is defined in the High Octane spark tables, and the Knock System is constantly searching for knock and if it see's it then it will retard timing. I think this is what you meant.

If the sum of the tables say 20deg (there are a few of them), then it runs 20deg. If the knock system see's knock, it will immediately temporary pull timing and increment the Knock Learn which is a multiplier table that starts to use timing listed in the Low Octane table. When knock stops, spark retard fades out fast and the Knock Learn will start to go back to 0.00 which slowly put it back to whatever the High Octane spark tables say. 1.00 is the max value for Knock Learn, if it's at 1.00 then it's using 100% of the low octane table (along with any other modifier tables).

So what I have found most useful rather than monitor just knock retard, is monitoring the Knock Learn table. If it's 0.00 then I know there has really been no knock to be concerned about. This is a longer term table which makes it very easy rather than trying to watch the instantaneous Knock Retard table constantly. I actually have a LED light come on when my Knock Learn gets over a preset value of say 0.2 - this means I've had a fair amount of knock recently.

The Knock Learn table can be adjusted to ignore knock retard below any value, stock is around 1.5deg, so if it's showing 0.00 then I know whatever knock I have is less than 1.5deg and that is pretty good really. You will normally get small amounts of knock at random, 0.2deg, 0.7 deg, etc, it is not because the ECU keeps adding timing until there is knock... this does not happen. It is happening either due to mechanical noise or the timing maps are set very close to the edge of real knock and your actually getting just a bit of real knock. I would be concerned if it shoots up to say 5-8deg more than one time... but the Knock Learn table will increment pretty fast of this happens so for times that your not datalogging, this is a great thing to use.

The problem with this is, the Knock Learn table is responsible for making your car feel like a dog at random.. if it's incrimenting and you dont know it, then your running severly reduced timing and it should actually be given more attention than it is. If you are at a higher Knock Learn, then most likely your not getting Knock Retard because your far into the Low Octane Table, it takes many miles and minutes for Knock Learn to start coming down back to 0.0 how you want it. If you constantly have higher Knock Learn value then you need to resolve that either by reducing timing in the High Octane Tables or other tables, or using better gas, mechanical problem, etc.

If I see 0.40 in Knock Learn then I know it's knocking pretty good for a bit and maybe I should make corrections or try better gas. If i was only monitoring Knock Retard then if I blinked, I would miss it.

For knock tuning, I personally prefer to make the High Octane and Low Octane tables both High Octane values, then leave Knock Learn active. This way I'm getting a realtime historic Knock Learn value but if it's incrementing I am still logging accurate knock history in the histograms that are not getting overwritten by knock free data from having timing reduced from running the Low Octane table. When done, I put the Low Octane table back to stock.

It is all in how you interpret what I said, but we are in agreement.

The high octane table is the upper limit and the low octane table is the lower limit.

I do log Knock learn for as long as I can remember.

If the vehicle has knock learn it will degrade if no knock is detected searching for the highest value.

I also don't rely on knock sensors as much as the new generation tuners, I look at many other factors learned over my 45 year career as a racer / tuner.

Thank you for chiming in, nice to see some interest in this subject.

[Pro-tuned]

I understood what was implied and a few words were missed “in any cell”, and it’s agreed that timing is limited to assigned/decided table values.While some tuners just mirror the low and high and leave it. Some tune exactly as stated above just return LFTT to stock. I would add a third option that conservative low octaine table values can be tuned and “optimized safely” as well to “reduce” the impact of “the dog” feeling. This being done only after high table numbers are optimized. This can be good in areas where one utilizes consistent reliable sources for fuel quality.