Log Replay ... Ignition timing out of sync?

[Torque]

Hello ...

I recently did a log of a short run with my car.
While replaying the thing the log player window states 13 BTC, whereas the mapcursor dwells in a
different area.

IMHO 13 BTC is plain wrong ...

So what is the actual timing?

[bachig24u]

Does your ECU use a knock map?

[Torque]

Hello ...

Yes, it does have a knock map but the area the cursor is in (load level/rpm)
does not consider knock since the knock flag is off in this region.
(you can tell by the shaded cells)

I further think that the log does not know about knock map switching.
(would be great if it would)

Cheers from foggy London (fog=good for the intercooler )
Andrew

[Matt]

This is not a strict rule since knocking can pull timing out regardless of in the map area (not necessarily jump maps)

Just to check if you make both maps the same and then check that

Also log in 'stream mode' so you have better resolution to work with

[Torque]

Hi Matt,

Hmmmm

So what you say is that knock detection and subsequent timing changes are reflected in the log somehow?

But in my case the knock flag is not set at the 'cell' the engine is operating at this moment.
So the timing should not be subjected to any knock aversion tactic ... right?

I will sync the knock and ignition maps and have another go.
(This time stream mode)

Andrew

[Matt]

Timing reported by consult (and in the log) is the result of reading the current timing maps

Which ones depends on TPS switch position, and subsequently the cells in that table dependent on RPM, TP and temperature indexes

Then any variations are applied to that. If there was knocking for example then timing would be pulled back depending on the amount of knock and counts required. If you have the R32 ECU and knock sensing panel up then the knock count would be recorded

I know that you are using the Z32 ECU and we dont record or report knock for this ECU since I havent been able to identify the register which reports the count, subsequently you will not have any knocking reporting in the log file for this particular ECU

Knock reporting only active on Z31, VLT, S13 CA18, HR31 RB20DET, HCR32 RB20DET vehicles currently

[Torque]

I can not get into stream mode.

If I activate it all readings are messed up!
At least I have solved the energy problem of this planet since I can drive 40 km/h without any rpms ...

Any ideas?

[Matt]

Restart Nistune and try again. Consult table looks like it got screwed up during register detection and things are out of wack

[PL]

Hi Andrew,

These ECU's run various knock avoidance algorithms. Sometimes it almost seems like "If in doubt pull timing"!

It doesn't matter if you're working where the knock cells are active or not. If it detected knock when operating within the active knock cell area then it'll jump to the knock maps and stay there.

They will also pull extra timing *on top of* the values in the knock map if they get really upset (usually if a knock sensor fault code is raised). So if your main timing map shows 18 degrees and your knock map shows 15 degrees, you could see 10 degrees (or less) if you have a knock sensor fault.

Consult reports correct timing. When tuning it's a common mistake to just look at the IGN timing values in the main timing map. You must also monitor the timing values reported by consult. If they are less than you expected (like what happened to you) then find out why - it'll nearly always be knock sensor related. This has caught me many times....

Same deal with injection times. Always use "knock copy" - but the bottom line is the value reported via consult. If in doubt then do a log and inspect the values reported. Be sure to use Stream Mode to improve log resolution.

PL

[Torque]

Hey ... looks like the Fab Two are online

Many thanks .. will sync all maps to the same values and then try to get into stream mode and do a log.
I think I have restarted Nistune and then went into stream mode directly (but will double check)

Many thanks .... very interesting.
That could explain why the car is not pulling in the upper range as I think it should.

I was under the impression that if the knock flag is off (in the cells) that should be it!
(as with the O2)


Will be back tomorrow since the car is now locked up in the garage ..

[Matt]

I did testing on HCR32 ECU with a pulse generator putting knock in around 2000rpm low load where the flags arent and it pulled the timing back to -2 or something silly because the ecu thought the engine was pinging its tits off!!

Will rerun the VCT on the bench and see what is going on there...

[Torque]

That is interesting indeed...
So what are the knock flags for then anyhow?
This is confusing! I thought I was on the safe side with the flags being off.

---------------

basically I injected a post filter signal from the knock sensor board to the ECU for the R32 this flags in address 0x007F bit 1 clear = knock

When this occurs and the cursor is in a knock flag enabled area on the ignition timing map it will decrease the timing rapidly as pulses are fed in. Once the knock is not longer occuring the timing will gradually increase

I dont see any activity on the 14AA address to indicate knock count is occuring, even though the routine appears to increment/decrement the address

Had a further look through the code and to enter that routine
R32 $10X0054 must be some other kind of flag dependent on input switches (currently clear on the bench)

--------------

Is pulling back the timing and using the knock fuel map always done simultaneously?
Or does the ECU vary on what to do?

As for VCT/VTC that would be just great since the logging of it would be very intersting and useful.

Cheers,
Andrew

[Matt]

From what I'm left to believe, some timing is pulled back and then switches maps and then pulls more timing.

Knock sensing code is quite complicated and no one has really gone through it to know what it really does.

What I've seen so far is the knock register is monitored and when active will increment a knock counter (which decrements in absence of knock signal). There is a separate knock counter which is monitored before determining whether to turn on a flag to use main/knock maps

Earlier ECUs like VL turbo / Z31 only have single map and just start pulling out amounts of timing whilst there is knocking with no switching (easier to grasp)