300ZX Z32 - One side richer + knock, long story

[nickopok]

I've noted baseline 'noise' knock coming from the knock sensor which is calculated per each cylinder. On top of that baseline the count will increase during the event when knock actually occours.

The difference between the baseline 'noise' count and event triggered 'noise' is effectively your per cylinder knock count.

Thanks for the precision Matt.
It worry me a little because if the ECU measure the background noise, and deduce it from the knock spikes noise, there is probably something wrong with my engine.
I had the hope that my problem could only be a higher than normal background noise.

The knock reporting that your are developing is a very exciting feature, even more if we can see counter for each cylinders.
Or at least if we can see which cell was active when counter increase.

Meanwhile I will do more runs, reducing the knock limit cylinder by cylinder.


Do you think that 15AFR is a good measured value for 40-52TP ? Maybe a little high ?

[RomChip200]

14-15 AFR looks ok for TP=40/52, you are still in closed loop up to 3000rpm.
But but you have to reach 13.5 AFR very quickly for TP going beyond 52

Few years ago, I had an area where knock was detected.
It was only 1-2 degrees but annoying enough because it affected my consecutive pulls, and my timing at full boost.
It didn't make sense because I run E85, even if the load in this area is pretty high (turbo pressure build-up)

(In the following, I transposed values from my maps to the oem ones to ease the discussion)
To trigger it, I had to floor starting from 2000rpm.
It was in the area TP 56-64 / 2500rpm where the stock timing is pretty high (27-29° BTDC).
But I was unable to hear it.

I tried various things:
_enrichment (to go with pressure build-up), even if it can make the things worse as you inject already more with E85
_retard timing, didn't help
_increase knock limits for all cylinders by x1.3
_I finally attributed it to false knock (I assume E85 burning when it is injected in avalanche effect is different) and disabled the knock flag area.

No problem since.
I don't recommend doing such when running unleaded.

[nickopok]

Thanks for your feedback Romchip.
It seem that my knock occur at the same rpm range, but even with only 60% accel.

On your advice I reduced this zone, but without luck.
Setting Knock Cyl limit to 90 on the whole range is maybe hardcore as it's x2 to x2.5.
I should now refine this setting.

From the last run, made with this 90 setting, it seem that ECU it still playing with timings, slightly reducing or advancing them.
About advance it's not suprising, but reducing mean that sometimes he measures more than 90

Knock_limit_90_run1.jpg

(376.28 KiB) Downloaded 1664 times

Knock_limit_90_run2.jpg

(392.96 KiB) Downloaded 1664 times

[RomChip200]

I don't see any retard in your screenshots. Remember the timing is interpolated on 4 cells and your trace screenshots inherited from the TP scale of the fuel map (bug in Nistune !?). Timing map has different TP scale.
Your average values look ok.

When I had retarded timing in the 2500rpm/TP56 area, it was 1-2° on the first pull, 3-4° on the second, 6° on the third, then finally to switch to knock timing map ...
The ECU has to detect pulls without "knock" to decrease the retarded timing, otherwise it accumulates across the successive pulls. Once you reached the knock timing map, there's no way to go back to the regular timing map, only engine restart.

[Matt]

Input trace always uses the fuel map scale (TP + RPM), and then places the cursor based on the cloest actual RPM/TP for all parameters (including timing). Normally I recommend making timing map using same scales as fuel map, which makes the tuning process easier to work with when both maps are on the same scales

[nickopok]

Sorry for late reply.

Romchip, on the trace selected cell you can see min, max, ave values.
For 25 into timing table for TP56, we can see that ECU used 23 to 25.

Into the TP64 column, timing is 24 between 2400 and 3200, while on trace, average value is 22.25 to 22.92
Into TP64/2500RPM values are :
Min 21
Max 24
Average 23.3 for 9 Count

So I suppose that ECU is still detecting some knock, but succeeded to go under 90 threshold with 21°
Having a Compare button could be usefull to quickly see the difference between timing table and trace value.
I think I will try to put values into an excel file for a better view.

The ECU has to detect pulls without "knock" to decrease the retarded timing, otherwise it accumulates across the successive pulls. Once you reached the knock timing map, there's no way to go back to the regular timing map, only engine restart.

Yes you right, I think that if no knock is detected, the ECU will add 0.1° each second, maybe related to the "Timing advance factor" setting.
I think we can avoid to switch to knock timing map by increasing "Knock retard limit" table
There is so many settings for knock, I didn't understood all the factors.


Matt : Thanks, I will update the timing map to get same TP than Fuel map. It will ease reading.

[RomChip200]

Taking into account the interpolation, I see no problem regarding the timing values reported.

I partially trust the min/max values because of the transitional effects and interpolation: when you floor the throttle, you move quickly across the cells and the ignition timing reported is close to the current TP/RPM but may be not perfectly accurate.
for TP=56: (24+24+22+20)/4 = 22.5
for TP=72: (22+20+19+20)/4=20.25

So 21 is not enough far away from the cell highlighted.
In your run, when do you get 25° as "max" value and 21° as "min" value ? Is the value recorded really decreased across the time ?

The only way to accurately know if retarded timing is involved is to monitor these:
timing_offset2 EQU 0x1614 15
timing_offset2 reflects instantaneous knock and 254 value means 2° of retarded timing (decrementing from 0).
knock_value EQU 0x1610 11
knock_value reflects repetitive knock situation and when the value reaches 248 (decrementing from 0), you switch to the knock timing map.

[nickopok]

Oh right I forgot interpolation, you are surely right as I didn't found a min value lower than a close cell.
Anyway, nothing comparable to when using standard knock cyl values.

It seems that few cells of Run2 are lower than Run1 by 1 or 2°, especially in the 2000RPM range.
But I don't have enough data to be sure.

I should do a new run with a lower treshold value to see if I get more difference without goind to safety.

Wow I found your old post about monitoring timing_offset but I must admit that I don't know how you monitored them
viewtopic.php?t=1280

I modified the consult registers table @EA00 to log some other registers instead the standard ones, with log player.

Where is the consult register table ?

[Matt]

This is good to know:

Code: Select all

timing_offset2 EQU 0x1614 15
timing_offset2 reflects instantaneous knock and 254 value means 2° of retarded timing (decrementing from 0).
knock_value EQU 0x1610 11
knock_value reflects repetitive knock situation and when the value reaches 248 (decrementing from 0), you switch to the knock timing map.

I saw 1614 copied from 1610 and 146B in various locations

Code: Select all

;; Z32 locations
knock_count	equ	$0083
knock_timing	equ	$146B
knock_timinga	equ	$1610
knock_timingb	equ	$160E
knock_timingc	equ	$1612

knock_amount1	equ	$1578
knock_amount1b	equ	$157E
knock_amount2	equ	$1579
knock_amount2b	equ	$1580
knock_amount3	equ	$157A
knock_amount3b	equ	$1582
knock_amount4	equ	$157B
knock_amount4b	equ	$1584
knock_amount5	equ	$157C
knock_amount5b	equ	$1586
knock_amount6	equ	$157D
knock_amount6b	equ	$1588

[Matt]

The register table can only be modified if you are using an emulator. With the above definitions, I am planning to add another stream to report the knock stuff

[nickopok]

I see, It will be too complicated for me

Finally during a trackday the car triggered safety boost after several mid / full load pulls on 3rd gear
So I'm back to the start or almost.

I need to know if the knock is detected only at a specific rpm or on a wider range.
If only at a specific rpm like RomChip, I seen I can toggle detection with "K" key

[Matt]

Monitor your timing and see where that is getting pulled back. There are several zones where knock is detected, but as to where it pulls timing I'm still to investigate, as Nissan doesn't document how it works so we need to reverse engineer it