NefMoto

During closed loop AFR, I'm getting strange AFR behavior that I have been having a lot of issues solving. I've posted this same issue a while ago without coming to any solution.

I fixed some hardware issues and ran many checks on many parts and now I am back where I once was, still with closed loop AFR issues.

This data was recorded not under load with the car sitting in the garage after installing Fuel Injector Clinics flow matched 650cc injectors at 3 bar. I applied the data they provided me with to TVUB and calculated KRKTE and KVB. I have not yet touched KFKHFM or FKKVS yet (although I do have a properly scaled MAF). I have messed with those back when I had 550cc genesis injectors and they had no effect on my closed loop issues.

Anyone have any other suggestions? Messing with the idea of moving to a slot style MAF,I've heard they deliver pretty good, smooth, robust data.

If you can't even get this running with a stock MAF, you're insane if you think an aftermarket MAF is going to make things easier.

Any ideas?

Look into hardware issues?
Air leaks?
Bad maf?
Bad O2?
Can a tiny pinhole weld leak result in unmetered air? Im unsure how sensitive these system sensors are.

Software issues?
MLHFM?

Study deeper into a specific portion of GGLSU?
Anyone ever deal with closed loop desired vs actual discrepancies?

KRKTE or MAF scaling is wrong, or intake leak post maf

What are your trims

Hmm, this is new.

STFT’s and LTFT’s seem to be pinned at 0.0% after several days of driving in multiple conditions. I suppose that indicates a bad O2 and open loop conditions (open loop does not use O2 if im correct?)

I'd expect at least a few DTCs if you are seeing trims pinned to 0% and a ton of lambda correction. Why not adjust KRKTE accordingly and see what happens?

Log the trims bits to see if they are being set to activate atleast while driving.

I used MasterJ's calculator then did it by hand, came out to the same value.

.05236789

656 was the average flow between the 4 using D2 fluid by Fuel Injector Clinic. Called them and they said D2 flows the same as gasoline, no need to scale based on N-Heptane.

TVUB is set to FIC's provided latency chart of:
2.03
1.39
1.04
0.80
0.64
of the correct corresponding voltage between my BIN and their chart.


*edit:
shit, left NOLRA=7 by accident. will correct and report

also, unsure of the bits but the log shows STFT's active.

ITT:
People learning that injector specs given with injectors are usually useless.

Welcome to tuning. Have a nice stay.

Oh and ... every time I see anyone who describes some physical parameter about the engine as "smooth", I instantly know this person has no idea about anything. Prove me wrong ;)

P.S.
There is KFLF, KFKHFM, FKKVS. At least KFLF and FKKVS need to be zeroed.

No it doesnt. It stays at 1, that is the actual trim value. Log B_frau, b_frao , b_lra.

Maybe you need to look closer, because fr_w is changing, meaning STFT is active.

KFLF, KFKHFM and FKKVS are all 1.

Trims with my prior posted values were -1.3% and -9.4% after a 20 minute drive.

New values are:
KRKTE: .05727600
TVUB:
2.003
1.371
1.027
.789
.632

Trims after: -1.9% -16.4%
Lambda (partial) self-adaptation (-16.4%) kept climbing after a 40 min drive, most often jumping up in value after closing/opening of the throttle.
Either air leak or interuption of MAF (turbulence) from DV?

Also, i feel like i went the wrong way with trim scaling...

No codes stored.

yes negative means rich, not lean.

I didnt see, my bad.

Got it reacting predictably.

KRKTE: .04750800
TVUB:
2.056
1.408
1.053
0.811
0.648

-0.8% and -1.6%
Will scale once more and log on my way home and check the results. If it looks promising, i’ll try FKKVS again.
I did not get these results with ev14 550’s even with NOLRA=4

Tuning any injectors with known MAF:

1. Set the MAF parameters correctly (or if it's stock, leave as-is).
2. Zero FKKVS and KFLF
3. Set KRKTE to something reasonable
4. Set TVUB to something reasonable.
5. Test

To set TVUB:
6. Check requested vs actual at idle and revving from 1000 RPM up all the way to 5000 rpm:
If fuel gets richer as the revs go up - decrease TVUB (multiply entire table).
If fuel gets leaner as the revs go up - increase TVUB (multiply entire table).
Repeat until it's decent.

To set KRKTE:
7. Do a WOT pull from low RPM to high RPM, check lambda correction.
On wideband - just multiply KRKTE by the average fr_w.

To finish up:
8. Make a matrix graph, which is like FKKVS, but values are fr_w.
Multiply FKKVS values by the values of the matrix graph until fuel is decent around the entire range.

9. For TIPIN/TIPOUT - try tipping in at different RPM's with different strength - if there is a lean spike, increase KFBAKL, if there is a rich spike, decrease KFBAKL.
10. Try tipping out at different RPM - if there is a lean spike - decrease KFVAKL, if there is a rich spike increase KFVAKL.

Wait I am confused, how do you log lamda corr at WOT, when its 1'd out ? Am I missing something?

Thanks Prj, i tried this last time with my 550’s to no effect. Went pretty extreme with the changes to KFVAKL/BAKL with no effect whatsoever, but i’ll give it a go this time.

Trims are -.6% and 0.0%
Havent had time to log yet.

Very clear instructions.  Should be pinned or added to the Wiki.

Obviously on a wideband ECU it's not!

It's like you have to spell everything out - on a narrowband you would use external afr controller compare actual with target, divide them and get the same theoretical fr_w...

Everytime I read one of your comments, I just feel like smacking my head against a wall....

LOL for some reason i completely forgot about wideband lmao, prolly because below that you said "on wideband do x" , so I got confused on how you got lambda corr on wot on narrowband without UEGO wideband.

If you only EVER read like ANYTHING before you posted. Read the original post, it SPECIFICALLY says "ON WIDEBAND".
The car in the topic is also WIDEBAND.

LMAO relax, he's just upping his post count :D

Yeah right  :D .

It probably does not help that when I posted that it was basically midnight after a few long days, but whatever.

Dialing in TVUB, it's taking a while. I noticed my idle's hunting.

adaptation seems to be correcting as well.

*Edit: yea i dialed my trims down to 0.0% and 0.0%. Took a while. Idle hunting is still there, randomly, most times it idles well, sometimes it hunts.

Also, i shifted KFVAKL up and down .2, .5, and finally, 1. It had no effect. As stated, that is the tipout enrichment? (Thats where i have a .75 lambda enrichment spike.)
Anyone have ideas?

PRJ, Im a bit hung up on this. Aside from my decel enrichment spikes and the hunting, my actual afr looks to me to be pretty equal to requested +/- 0.05 lambda aprox. Could you provide me with a log example please?

I’ve been working on this issue for some time now.

I’ve hand editted KFKHFM and FKKVS (separately and independently) which produced no effect to AFR’s.
I used the fixxer to edit FKKVS (which produced very close values) and that produced no effect to AFR’s.
I multiplied TVUB in both directions from theoretical values which produced poorly running results. Bucking, heavy backfires at all rpms. In data, lots of idle and cruise oscillation and load change erraticism mainly on the enrichment side.

I returned it to theoretical (provided spec sheet) values

I editted KFVAKL/KFBAKL. Raising KFVAKL (decel enleanment) reduced load change enrichment spikes. (It eliminated them under steady cruise load change (shifts, throttle close) but had no effect in rapid load change scenarios (vehicle standstill clutch engagement in 1st gear, 2nd gear, 3rd gear until steady cruise. Mainly, steep, abrupt rpm changes)

Lowering KFBAKL (accel enrichment) has had no effect on accel enrichment spikes under any conditions.

KRKTE was refreshingly successful and predictable. Zero’s PT trims (checked with VCDS), reduced by multiplying the displayed percentage until zero’ed. Took 3 passes and felt realy good.

Everything else has been a massive entangled unsuccessful struggle. My logs still look like the past ones posted.

Can anyone help?

Still in the weeds.
Modifying FKKVS makes the idle hunting dramatically worse, even if i scale FKKVS ms axis to address idle areas. KFKHFM makes it worse as well.

There are too many issues to figure out and i’m just reading the FR randomly, understanding nothing and getting no where.
Car still runs like shit.

Anyone have ideas of what to variables to log?

It seems load, rpm, afr and corrections oscillate in time with eachother and ign timing oscillates inversely. What drives what? I’m not sure whats at the top of the pile here. Like, if rpm was stable would it trickle down and the other variables will straighten out? Where do i begin?