KA24DE-T Injector Resize TP Question

[iyarovoy]

Sorry, I've had some bad family news after returning this week, so just trying to keep on top of things

Mr. Matt take your time and thank you for all the time you have put in helping me troubleshoot this.

[iyarovoy]

I came into possession of a manual s13 KA24DE ECU. do you think if I installed the nistune board with the auto base image into the manual ECU, the lean out would go away?

[Matt]

I don't think it will make any difference since I tried the 53F0F ROM (in AT hardware) and still saw the problem

After a good hour repeating on the bench (but was okay in the original 53F0F/53F1F file) your fan control table appears to be affecting things

Try the attached file out (53F0F with maps copied in from original tune). Appears to work fine now

[iyarovoy]

Mr Matt,

No joy with your updated bin, still lean for the first second on the throttle then injection time spikes (See log + bin).

fan_fix_matts_map_2018-05-15_0916_06.csv

(70.72 KiB) Downloaded 226 times

I was trying to see what you modified in the "Fan Control Flags" and both the old map and your new map look identical, but the "compare" tool indicates they are different. How did you go about figuring out it was the Fan Control Map causing issues? Maybe I can repeat the same process on target to help narrow down.

I then copied your "Fan Control Flag" into a tune I have been working on in which I reduced the K-constant to fit the TP into the Factory TP Scale range and updated the Fuel and Timing tables for boost. The issue behavior seems to change, in sense that there is still a lean spot at throttle but not as bad. Still not sure why my ignition timing values are so low compared to my timing map, knock sensor + knock map ? (See log2 + bin2).

MT_Image_KConst_log.csv

(45.26 KiB) Downloaded 270 times

MT_Image_Kvalue.bin

(32 KiB) Downloaded 231 times

Seems the issue is possibly related to:
- TP Rate of change
- Throttle position rate of change
- RPM

[iyarovoy]

Another log to go along with MT_Image_Kvalue.bin attachment that still the lean out clearly visible:

MT_Image_Kvalue_log2.csv

(40.45 KiB) Downloaded 310 times

[Matt]

What I've been doing is taking your existing tune and then gradually copying the MT tune maps into it, until the problem did not show on the bench.

Normally I notice once MAF/RPM are in a certain location, then the injection time is around 5-6ms when it should be 8-9ms and then jumps to this value

I noticed that if I used the original 'fan control' table that this problem did not repeat. I'm not sure on changes with the fan control table, I will need to check with a hex editor since dispalyed values appear the same. Gradually I've been removing changes from other maps, and determined it was probably this that caused the problem

I think the problem occurs with the fuel map added in, and how it ramps up. I've seen similar issues with S15 but the EGR table fixed this one

Also AlphaN table needs to be 255 (not 0) and box not ticked in feedback flags

I saved my files incrementally as changes were made and then put maps back in. Maybe I missed something? What you could try and do is start with the original AT file, and then just copy the basic changes in firstly, so start with VQ/TIM/K/Latency and then check it doesnt happen. Then add in the other maps (TP Max, feedback switch). Save each file as you go. Let me know if that works. I can look at it again if you have a better idea what table change is causing it

[Matt]

You could also load in the original MT file, and start from that too (tables are different between AT and MT version, so only pick one to start from where it does not show the issue, and use 'COPY' button to copy maps in from your current problematic file, upload each time after COPY)

[iyarovoy]

Turns out both MT and AT Feature Pack base bins exhibit this lean behavior, so at this point I dont have a BIN that DOES NOT do this, there fore I cannot narrow down the parameter.

Im not sure if I can use non FP bins in Hardware that has a FP Firmware loaded, nor do I want to try this since the issue might be with the Feature Pack Firmware itself.

From my research, no one has complained about this behavior as far as KA24DE-T tuning goes, however 99% of the tunes I have looked and compared were Non Feature Pack tunes.

At this point I am not sure what else to try and driving the car has become a burden more than a pleasure as it just stumbles every shift for second or two.

Sounds like the same issue here: http://www.nistune.com/mantis/view.php?id=94, viewtopic.php?f=8&t=3596&p=26302&hilit= ... ack#p26302

Ill try running a non FP bin using FP Firmware Board.

[Matt]

Okay I will try your tune out using a non-FP BIN file and work from this at this end and see if it repeats (or not)

Z32 issue was different since it was only when knock sensor fault happened, then it would access knock maps (which were removed, but it had no fuel reference due to a glitch in firmware)

[iyarovoy]

How do you simulate all the inputs into the ECU on the bench? Signal gen for RPM and speed, and variable 12v voltage sources for TPS MAFS and temp?

The EGR map you added does make a difference as leanness is not so severe when the map is set to all 0s, however the injection time still does the step function. Setting this map to all 127, the AFRS stay lean through out the entire rpm,TPS, and LOAD range.

I played around with TPS enrich tables but they all seem to depend on the rate of change of the throttle movement and not on throttle position.

[Matt]

How do you simulate all the inputs into the ECU on the bench? Signal gen for RPM and speed, and variable 12v voltage sources for TPS MAFS and temp?

I have two things
1. Nistune converts the logs into 'commands' which connect to my arduino based engine simulator (so it adjusts RPM, MAF, TPS etc)
2. Arduino itself takes the commands and is connected to other hardware for some outputs (eg temperature simulator chip for coolant tepm, transistors for driving switch based inputs) and other items like MAF signals generate PWM output which is smoothed to make 0-5 volt input to the ECU

I can also direct drive the arduino itself to control MAF, AFM etc.

Anyway looking at this again now

[Matt]

Maps showing current changes attached

[Matt]

53F10 ECU with 53F00 firmware (no FP) and copied in above maps only. K is standard, standard also

S13_KA24DE_53F00_jump.PNG

(27.35 KiB) Downloaded 2592 times

Orange (inj time) shows jump. Ignore the timing (knock timing was 0's)

Setting EGR from default (factory) to 255 also has the same jump (no difference in timing)

[Matt]

Setting EGR offset table to 0, stops the jump up (keeps the injection time 'lean' as you would say)

S13_KA24DE_53F00_nojump.PNG

(67.21 KiB) Downloaded 2591 times

[Matt]

Below with just changes to TTPmax (set 255 from 3200rpm) and AlphaN (set 255 whole table) and EGR (from 53F10) using factory fuel map (smaller jump in injection time)

S13_KA24DE_53F00_jump2.PNG

(27.28 KiB) Downloaded 2590 times

The above does not happen (no jump) if EGR table set to 0's