pcmhacking.net

I believe it makes it easier to see actual knock counts.... most of the knock counts are normally missed due to logging speed being far slower than ECU processing speed. so VL400 made the knock counts in 11p like a latching counter... if you see it steady, there are no knock counts... if you see it move, there are knock counts.

Ok thanks. That makes sense. It's the sudden change you look for. Strange that it was level at 205 till it knocked then it levelled at 130. Anyway it's obviously something I don't need to worry about.

goes to 255 and then resets to zero ?

If you look at the log watching the knock counts it displays a count of 205 continually untill it did knock and you can see it jump up to 229 which coinciding with a timing drop then after the knock stopped it dropped down to 130 and remained at that level for the rest of the logging. As Jayme said he thinks it picks a steady number and it's only when it jumps from there that it actually knocks. I suppose it's easy to follow a deviation from a stable line but the fact it went down to 130 instead of returning to 205 is a strange one with me.

yeah it counts 0->255 then 0->255 again (8 bits of data = 1 byte of the ecms very limited ram = numerical range of 0 to 255). The knock events happen so quick you register heaps of them between frames of data. This is why its implemented as a counter. In your case 205 -> 255 is 50 events, then 0->130 is 131 events, so 181 knock events detected. Its just the speed at which everything happens and the speed of the program running on the ecu. If it was implemented as a momentary state 50 knock events could be detected between frames and you would miss it in the logs. Note that this number does not equate to combustion cycles. The knock filter implements PWM (I think...) and faster PWM is louder knock. Exactly how that turns in to knock counts Im not sure. Presumably louder/harder knock = faster pwm output from knock filter = more knock counts.

Thanks antus I figured it was something like that but having no knowledge on how it worked was curious if it was a problem or a bug at my end. Thanks for the piece of mind now I can go on happily tuning and enjoying 11P.

not sure if it's like the ecotec routines on VS-VY but the knock events is when the code runs, a routine watches the amplitude of the knock signal and after a certain level if it's held high for 2 loops i think it then runs the knock code, when the code runs it increments the knock register by 1. This doesn't mean you have knock it just means the trigger level has been passed for 2 loops and the knock routine has been executed, the knock code then looks at how much ignore time or addon time you have and adds that in before retarding the spark advance.

Knock control on the early ECU/PCMs is very different to the later models, its far simpler.

The knock filter board is on the memcal and the ecu just gets a square wave signal indicating knock, the filter board does the signal processing for bandpass filtering and signals intensity by how long the square wave signal is low for. The ECU measures the amount of time the signal is low during 12.5ms windows, it is not gated to a particular engine position like later models and just continually samples (which is why engine noise gets picked up as knock).

So, the ECU uses a 16.384KHz counter and samples over 12.5ms. With 16.384KHz counter each knock count is 61us of time (intensity) from the the knock board, so approx 16 counts is 1ms of knock intensity. On a side note, with a 12.5ms sample period the maximum counts the ECU can measure is 205 (which is what you see when no knock board is fitted to standard $11 or 12P code).

You will have seen the knock attack rate table column is shown as deg/ms, this is degrees of retard per ms of knock. So with your example above of 181 counts, if the table was 0.51 deg/ms then 181 * 61us = 11.04ms of knock = 11.04ms * 0.51deg/ms = 5.63deg of knock retard. In the log it looks like you had multiple knock events very close, which adds to the total knock retard, up to the maximum knock retard scalar.

so interesting question came up from a mate of mine... can we have a "spark only" limiter?
I think 11P with spark cut enabled, cuts fuel AND spark. can we have a setting for crazy people that want 4 foot flames and expanded mufflers to leave the fuel going during the spark cut limiter, for burnout comps, etc?

i think the problem would be bypass mode if you cut EST.