XFI knock retard: values that match GM setup

[GN SBS]

XFI users or others familiar with the stock ECM's programming: What values are you using for XFI's ESC function, which has three values to enter :

1. Max ESC retard (in degrees)
2. ESC attack rate
3. ESC decay rate

Detailed definitions below

Thanks,
Scott

1. Max. ESC Retard (*)

This is the maximum amount of timing retard allowed if the knock sensor is activated. Even if knock is still detected once this limit is reached, no more timing will be pulled out.

2. ESC Attack Rate (*)

This parameter defines the number of degrees to retard the timing every .1 second when the knock sensor is activated. In other words, this parameter is used to set how fast the timing will be pulled out when knock is detected. In the above form, with the attack rate set at 5, the timing would be lowered 5 degrees every .1 second until the knock sensor deactivated, or until the maximum ESC retard limit was reached.

3. ESC Decay Rate (*)

This parameter defines the number of degrees to remove the timing retard every .1 second after the knock sensor becomes inactive. In other words, after timing is pulled out due to knock detection, this parameter is used to set how fast the timing will be added back in. If the decay rate is set to 0.50 as above, the timing will come back in at .5 degree per .1 second (or 5 degrees per second) until the ESC retard has reached zero.​

 

[GN SBS]

ttt-- never got an answer on this one.

I'm using values of:

• Max: 15
• Attack: 3
• Decay: 2

But am wondering if I'm too aggressive on the attack rate. Anyone know what parameter values GM used (or good general rules of thumb for XFI?

 

[GN SBS]

Pasting in an answer I got as a reference AND to pose a question:

Since you're using 0.1 sec, you'll have to convert the attack & decay values as I'll list them in their native form.

The stock values for 87 turbo are:
1- Max Retard degrees - there are two limits. When not in PE = 15 deg; in PE = 30 deg.
2- ESC Attack Rate - this 5-value table determines the retard to add for each (one) millisecond of knock vs. rpm.
RPM Rate
0400 0.18 deg
1200 0.18 deg
2000 0.20 deg
3200 0.31 deg
4800 0.31 deg
3- ESC Decay (Recovery) Rate - this 5-value table determines the percent of retard to remove per second.
RPM Rate
0400 49%
1200 47%
2000 31%
3200 31%
4800 39%

FANTASTIC data, though ultimately it looks like it won't just be a simple conversion. If GM was pulling 0.31* per milisecond (its time interval) that would be 31* per tenth of a second using XFI's time interval, which won't exactly produce a good result as it's greater than total timing at WOT. eek:

Looks like I'm back to using the general rules of thumb for XFI. Maybe this slow time interval is why so many XFI users don't turn on the ESC feature. Still a must as a valuable source of data and insurance for me, but I am surprised about the difference in time intervals.

 

[Eric Stage I]

Scott,

You're right, you would have to set your attack rate to about 30 degrees to equal the stock ECM calibration. It is much more aggressive than the FAST setup.

Eric

 

[50Bucket]

Do you know the voltages the sensor outputs say when hearing "0" knock and full blast?
Is the output of the sensor a 5V?

What are the contents of the sensor "kit"?

I am trying to use the "knock" feature for something else...

 

[GN SBS]

Bump. It has been a while and I thought I'd see if anyone else has good ideas for using XFI's ESC function.

Right now I'm just using a very aggressive attack rate as (1) a safety valve and (2) a diagnostic signal that I have some knock.

But it's an off/on switch, unlike the stock ECM's ability to set attack rate for each milisecond rather than XFI's tenths of seconds.

Anyone found a better way to do it with XFI?

 

[John at J&S]

Do you know the voltages the sensor outputs say when hearing "0" knock and full blast?
Is the output of the sensor a 5V?

What are the contents of the sensor "kit"?

I am trying to use the "knock" feature for something else...

Knock recording and output from signal processor:
Picasa Web Albums - John - J&S Vampire I...

 

[TurboTR]

If it's a GM knock sensing setup like used on the GN's, the sensor itself is basically just a piezo microphone that feeds an external interface and processing module (the GM ESC module). The ESC module decides whether or not the sensor is hearing knock, and then if so it signals the ECU by pulling the ECU ESC line low. It's a digital (HI or LOW) signal to the ECU.

Trying to make the decision of hearing knock acoustically or not is not a trivial task. The module itself has to integrate the background noise level energy, which makes a sort of automatic varying background noise threshold as rpm changes, etc. So then when the sensor hears knock frequency energy, that knock energy level has to be high enough compared to the current background noise threshold. If it gets high enough, it says we're knocking. The problem is, modified racing engines get noisier or different background levels, making it very hard for the acoustic detection scheme to work properly. Probably an over simpflication, but that's the idea at least.

I think that, poking around in some late model F*rd tuning files for example, the later model stuff also applies windowing to the knock sensing scheme to get it to work better. You can open the window and listen during the part of the combustion cycle where knock would occur basically.

We had a long discussion years ago about whether the (stock) ECU then further processes or filters the ESC signal from the ESC module and the consensus was yes it does, if I recall correctly. I think our resident scientist Carl Ijames may recall the details.

TurboTR

Do you know the voltages the sensor outputs say when hearing "0" knock and full blast?
Is the output of the sensor a 5V?

snip

 

[John at J&S]

Knock sensor data from "BigMoose":

BigMoose's Ramblings: Knock