The disassembly of this defeated me so instead I dumped 36 pages of RAM in 256 byte sections looking for variables that varied in the 0-4 range as the engine was running, taking my hint from DSM disassembly.

I found one that has code related to it which increments on CAS pulses, and varies with RPM, so we count 0,1,2,3 (cyl 1,3,4,2 TDC if same as the DSM which things often are!) for every two engine revolutions.

This will be useful if we want to do something like logging knock sum per cylinder.

 

Very nice...

 

great work! Too bad we couldn't get AFRs per cylinder .

 

O2 Sensors aren't even close to being fast enough. Even in a 2 sensor bank to bank 4 cyl setup above ~4000 RPM its done being able to differentiate which was which.

 

its not so much the sensors but the way the controllers drive, and read them that's slow. Innovate controllers supposedly drive them via hysteresis, and read frequency changes, as opposed to the standard feedback system most controllers use. They say their method can read AFR changes per cylinder.

edit: i dont know that they've necessarily prove the technology yet, but they either have or applied for the patent.

 

John, that is pretty nice. I remember DSMLink had the ability to log individual cylinder knocksum (or knock votlage or something), but the actual cylinder number wasn't known. It was just useful to see if there was one cylinder which was the culprit for the noise/knock.

With what you have found, can you actually identify the exact cylinder number?

 

Not sure how we'd confirm, maybe we could cross reference somehow to known commands to disable certain injectors, but the DSM disassembly seems to have this stuff sorted in the order I posted with no uncertainty expressed.

 

knock ignition trimming already uses this variable

 

It is a bit weird, because I'm getting it reading the octane initialization table depending on the cylinder index in that routine.

 

Whatever, it looks like individual cylinder adjustment of ignition timing might be quite easy now.

 

Before I will forget MUT_CB-CE are dependent on this variable seems knock sum is calculated from them .
MUT_D8-D9 (cylinder index/2) I believe this is actual spark, not sure thou

 

I think you may be right on spark, using the index for 2 coils.

The MUTCB-CE I previously labelled as 2 flags and 2 counters related to knock noise evasion, didn't seem to be in the usual path for knock sum calculation.

 

Gentlemen, you are most excellent!

 

Jscbanks,

The 2g's changed firing order halfway through the generation. I want to say 1997 but may be wrong. Also a 2g ecu will run an evo and and evo 8 ecu will run a 2g. Also a 1g ecu will run a 2g. I don't know if that helps but I'm throwing that out their. Also I have every reason to believe that an evo 8 ecu will run a 1g dsm but have not personally seen it done yet, but will hopefully soon.

 

1g and Evo firing order are 1-3-4-2 I believe, so should be good as I'm using 1g disassembly. I aim to see if the knock sum increments are symmetrical across the cylinders, and if not to allow a global tweak per cylinder (have a candidate table already that could be moved to do this).

Some suggest that the 2g firing order did not change but the cam sensor did? I have no idea, I've never even seen a DSM!