KA24DE-T Injector Resize TP Question

[Matt]

Please try the attached with your board (53F1F)

53F1F_Fix_First_Tune_ClosedLoop.png

(22.6 KiB) Downloaded 2542 times

Record another log file showing the jump, when it happens and I can check further

[iyarovoy]

Mr Matt,

Thank you for working with me on this. I really appreciate it.

Full throttle run:

53F1F_WOT_Run_1to4_Gear.JPG

(135.19 KiB) Downloaded 2529 times

log:

53F1F_Fix_First_Tune_ClosedLoop_1-4th_Gear_Pull.csv

(37.94 KiB) Downloaded 203 times

Notes:
-At WOT, the issue is not present as in the past however, some strange oscillations can be felt as highlighted in the screen capture (highlit are TP, Injection Time and RPM)

Partial Throttle run:

53F1F_Partial_Throttle.JPG

(126.35 KiB) Downloaded 2529 times

Notes:
- The issue is still there and can be felt. Seems like when the engine is lugged or light throttle, the issue is worse (shifting from 2nd to 3rd quickly at a low RPM or releasing throttle and lightly applying it again while in gear).
- Looking at the log, the step is almost time based, as if it is a count down counter somewhere before injection is restored.
- The issue seems worse when the engine is cold.

[iyarovoy]

Log for partial throttle run:

53F1F_Fix_First_Tune_ClosedLoop_Partial_Throttle_LeanStep.csv

(36.96 KiB) Downloaded 322 times

[iyarovoy]

Looking more through the logs I was surprised to find that the "Injection Time" and "Injector Duty Cycle" are not a derivative of each other (aka Injection Time and Duty Cycle slopes/graphs are often a bit different than each other).

It seems like the AFR follows more closely to the Duty Cycle and not Injection Time. Now Im trying to wrap my head around how you can vary injection time and duty cycle of PWM output. Seems like the ECU is continuously varying the pulse width and the frequency/quantity of pulses per second, as it doesnt make sense to have 5ms injection time at 50% Duty Cycle vs 5ms injection time at 10% duty cycle, unless you vary n-pulses per second.

Ex1 5ms Pulses at 10% Duty Cycle = 10 x 5ms Pulses in 1 Second

Ex2 5ms Pulses at 50% Duty Cycle = 50 x 5ms Pulses in 1 Second

[Matt]

Duty cycle is just a simple calculation which takes injection time and RPM. So the variance you see may involve RPM. Will be looking into this soon. Thanks for the updated logs

[Matt]

- The issue is still there and can be felt. Seems like when the engine is lugged or light throttle, the issue is worse (shifting from 2nd to 3rd quickly at a low RPM or releasing throttle and lightly applying it again while in gear).

Logs do not show AFR so I am unable to determine if the loss in power where the problem happens is due to AFRs, timing, boost or combination

The oscillation is happening because of the initial drop in RPM which affects TP, then affecting injection time. It is not due to the initial problem, but another separate issue where you get the initial drop in RPM triggering the oscillation

Calculations are TP = RPM/MAF(VQ) x K
Inj time calculated from TP x enrichments x fuel map

As the MAF voltage increases, the oscillation smooths out

These notes based on (1-4th gear pull log)

Initial:
09.250: RPM 3525, MAF 3.5V, timing 20 deg, inj 10.1ms

09.344 RPM drops to 3075, MAF same
=> TP increases, resulting inj time increasing 12.33ms, timing hits last column 18 deg BDTC

0.9443 RPM increases 3425, MAF increasing at same rate
=> TP increases further, so does inj time, timing same

0.9547 RPM drops and continues but smaller steps


Try some stuff here:
Keep your fuel values the same for the last two columns, because it is fluctuating between them. Is this a turbo or non turbo setup? Appears you are using around 740CC injectors?

Same for the timing (in the meantime), last three columns set to 17 degrees

See if it does the same thing then. Why is the RPMs dropping initially after acceleration?

06.547: We see a big drop in timing from 18 to 8 deg BDTC. This appears knock related maybe? Try disable knock feedback in the interim to see if that drops pulling timing like that

- Looking at the log, the step is almost time based, as if it is a count down counter somewhere before injection is restored

Those oscillations are a result of RPM decreasing whilst MAF is slowly building, and it then affecting injection time. Flatter maps should not repeat this issue. Normally turbo setups have flat fuel maps at top end, since as TP increases, so does injection time so you don't need to 'step' the maps up. Larger injectors have flatter maps compared to smaller injectors (740CC will have 1/2 values of that compared to 370CC fuel map)

- The issue seems worse when the engine is cold.

Cannot comment on that without comparing both

Let me know how the above goes

[iyarovoy]

Try some stuff here:
Keep your fuel values the same for the last two columns, because it is fluctuating between them. Is this a turbo or non turbo setup? Appears you are using around 740CC injectors?

This is a Non-Turbo KA24DE with a SR20DET T28 Turbo fitted to it (Basically NA ECU being tuned for Turbo). The Injectors were 275CCs which I upgraded to Subaru 550CCs.

53F1F_Fix_Rev1.bin

(32 KiB) Downloaded 195 times

This bin contains the following:
-I flattened out the Fuel Map
-Resized KConst to use the whole Fuel Map and Timing Map
-Adjusted TIM for fueling (i think it still off, as my "Estimated AFRs" do not match the actuals in the high load areas)
-Pulled in all the changes you had me try in the previous 3 Pages of this thread

Issue is still there.

I did logs of Partial and Full throttle for your review which both include AFR measurements.

53F1F_Fix_Rev1_Full_Throttle.csv

(53.64 KiB) Downloaded 219 times

53F1F_Fix_Rev1_Lean_Partial_Throttle.csv

(136.09 KiB) Downloaded 235 times

The full throttle run, fueling seems off as the numbers in the fuel map do not match the numbers measured, even though at cruising everything is around 14.7-15.1 and the Fuel Trims are near 0s.

The partial throttle still has the step, even though TPS and MAF measurements seem flat. Check out the Short Term Fuel Trim right at the edge of each step. Happens 99% of the time

I really hope we can figure this out soon.

Thank you for the help

[iyarovoy]

Issue is still there.

Lean_Partial_Throttle.JPG

(210.63 KiB) Downloaded 2422 times

[iyarovoy]

I may have narrowed down the map that changes this behavior.

"Sync RPM Factor" in expert view seems to change characteristics.

[Matt]

Those logs look a lot better than before, but still leaning to 16:1

Okay, does that sync table take away that lean spot we are seeing?

[iyarovoy]

Those logs look a lot better than before, but still leaning to 16:1

Okay, does that sync table take away that lean spot we are seeing?

It happens significantly less, and typically now at throttle tip-in it goes pretty rich, to ~15.3 then back to 14.7 like its supposed to. At very light throttle, it might do it, it might not.

If you could play around with this table on the bench and characterize its effects, that would be very big help! It might be throttle based or RPM based, or the delta-RPM(derivative aka change in RPM) based or deltaTPS (rate of TPS change) or both.

I am attaching a bin that I have been working with. I am viewing nistune in Expert Mode. I copied every map that is under the "Throttle Enrichment" section from the B13 SR20DE bin, as they are very similar Bins and use the same MAF between the SR20DE and KA24DE. I then started copying back the KA24DE enrichment one at a time until I stumbled on one that makes the issue come back.

[iyarovoy]

If you could play around with this table on the bench and characterize its effects, that would be very big help! It might be throttle based or RPM based, or the delta-RPM(derivative aka change in RPM) based or deltaTPS (rate of TPS change) or both.

Any updates on this? Just curious if you had time to put this on the bench yet.

[Matt]

Sorry missed this post. Will have a look

[Matt]

You could try richening up columns 41-68 if this is a turbo application, since running lean

InjTime_vs_AFR.PNG

(30.56 KiB) Downloaded 2231 times

[Matt]

Here we see a lean area but the injection time (orange) is steadily increasing, so that part does not make sense

Lean_spot.PNG

(20.52 KiB) Downloaded 2230 times