Normalise Knock

[kevin p]

Hi Everyone,

Having an issue with normalize knock.

The knock threshold drops below what it should be, then goes back the set threshold. I can't pinpoint what I need to adjust to avoid this. 

File and log attached.

Thanks

https://drive.google.com/file/d/151zMRXt4BmGMt_0Uqf_k8eHGoVUaK1sA/view?usp=drive_link

https://drive.google.com/file/d/1xHkl3EKTsY-jX6DClWI3UxiN8Ot_JY3j/view?usp=drive_link

[Adamw]

It is due to your Load Delta Threshold Gain and RPM Delta Threshold Gain tables.  What these tables are used for is to temporarily increase (multiply) your knock threshold when there is a sudden change in load or RPM.  Usually you would have values around 1.5 in these tables so for example if your steady state knock threshold was 2.00, then during a delta event it would be 3.00. 

Since your file looks like it is for an Evo1-3, below are the settings from my evo7 that should be a better starting point.  It works very well on my car, but from memory the early evo's have a different sensor so keep that in mind.  The Evo 7 sensor is a resonant frequency sensor of about 13KHz, but I have found 6KHz gives the best detection on mine.  

With most cars just a single knock threshold works fine, for whatever reason my evo needs a small increase around 4-4.5KRPM, but you could still likely simplify this example.  

[kevin p]

Thank you Adam, btw I'm using a Bosch knock sensor. Can you explain how the knock trim table work. I didn't see it in your screen shot. 

[Adamw]

I generally assign 1 knock trim table per cylinder and set the axes breakpoints up similar to my ignition table so if I regularly see retard in a specific cell in one of the knock trim tables, then I can just go to my main ign table and retard the same cell to stop it knocking in that area in future. 

As an example, say the ecu detected knock on cyl 1 at 4000RPM & 40Kpa, and it had to retard the ign 1.4deg to correct that knock.  You would have -1.4deg recorded in that cell like the example below.  So the next time the engine passes through that same cell/operating condition then cyl 1 would already get that same 1.4deg retard regardless if it knocked again or not.  If the engine later continues to operate in that same cell for more than the "ignition advance delay", then the ecu will start to slowly remove that retard - until either it is back at zero, or until it starts to detect knock again.  

By default these trim tables will be "zeroed out" at every start up.