Weird mixture adaptation values...

[Albert850i]

As I often use INPA I can confirm that the initial mixture is too lean (as measured by the Lambda sensors).

Thanks, I will investigate further. But with the diamonds on the adaptation scale all the way to the left (= negative I guess), the workshop manual clearly suggests too rich a mixture.

[Timm]

Unfortunately I only have experience of INPA - and your INPA screenshots above show a positive adaption value - which means the initial mixture is too lean.

[Albert850i]

Update: I put back the original injectors: no change. Then I changed out the original MAF's with aftermarket Ridex 3926A0165 MAF's. On bank 7-12, this made all the difference, and the adaptation values went to around zero after some learning time. See first 2 images.
Both banks run in closed-loop.
Bank 1-6 responded erratically: idle adaptation initially improved somewhat (from 127 to 71, still bad), but started to behave like crazy after some idling time. I swapped the MAF's, but the problem remained in 1-6. Occasionally lot's of blueish smoke is produced, and sometimes the bank is cut off or the entire engine just suddenly stalls. I disconnected the (new) PCV hose from the header (plugging hose and nipple), no difference. During a short test drive, the smoke from pipe 1-6 was embarrassing. After some idling time, even 7-12 started to act up again: see both INPA screen shots, both showing addaptionvalue additive 67. Huh?
So the problem is not fuel pressure (measured earlier, new regulators), not MAF's. Vacuum leak in 1-6? Unlikely as I already re-sealed the intakes to the head, and even tried a smoke machine before.
I tend to believe the problem may not be fuel injection related but something else.
Mmmm.

IMG_8861.jpegIMG_8875.jpegIMG_8877.jpegIMG_8876.jpeg

[Albert850i]

Still struggling with this problem. Part-load, the engine runs fine, power is there, no DME faults. Idling, the engine runs poorly, still no DME faults or check engine light.
DIS is very clear about the idling problem: mixture way too lean. Both DME's reach the limit (+127) in making the mixture richer. INPA seems to confirm this: "adaptionvalue additive !127".
Interesting thing is that this happens on both banks, exactly the same.
Options:
1) air leak: intakes have been resealed, and smoke machine reveals no obvious smoke leaks
2) insufficient fuel: new regulators, fuel pressure checked OK
3) mechanical problem: both compression and leak tests OK
4) MAF problem: have been checked by Extra Classics; oxygen sensor problem: new sensors installed

I will continue to investigate (measuring vacuum pressure, EVAP and PCV valves, check with wide-band oxygen sensor) but am running out of options.
Suggestions please.
Thanks in advance, Albert

Screenshot 2022-05-01 at 11.59.33.jpgScreenshot 2022-05-01 at 12.07.59.jpg

[Timm]

The question is, is this a measurement problem, or is the mixture much too lean? Pop a plug or two, what do the plugs show, black and sooty or white and clean?

[Albert850i]

Thanks Tim. I agree that first I need to confirm the lean mixture when idling.
Spark plugs (few years old) all looked pretty bad, and -apart from one- measured infinite resistance.. So I put new ones in recently, which solved the check engine (oxygen sensor) and ignition errors (CID sensor) but not the mixture issue.
Screenshot 2022-05-02 at 09.13.39.jpg

[R Shaffner]

These systems can pull in a lot of unmetered air through the CCV. I had high trims once that were caused by worn timing cover gaskets. I found the leak by smoke-testing the crankcase directly.

[Albert850i]

caused by worn timing cover gaskets. I found the leak by smoke-testing the crankcase directly.

Thanks: do you mean leaking valve cover gaskets?
I will try and inject the smoke from the machine into the oil filler opening directly. See image of my non-automotive smoke machine ;-)
Screenshot 2022-05-02 at 15.57.37.jpg

[Timm]

Those plugs do not show a lean condition, they are sooty in the main (rich), and some others are sooty and oily (rich and misfiring) - certainly not lean! If they were faulty plugs then it would be interesting to see what the working plugs are showing, but so far, very rich and misfiring.

[R Shaffner]

Yes, of course Timm is right. I didn't look at the picture.

[Albert850i]

@Timm: I must add that the engine burnt its last tank of petrol by idling in my workshop only.. Which may explain the bad spark plugs.

[Albert850i]

Update:
I bought a wide-band oxygen sensor, and installed it in place of the original sensor of bank 1-6. This bank switched to open loop, and I could measure the actual air-fuel ratio produced by the engine.
Results: 13.7 idling, 12.8 @ 2000 rpm. Hence too rich a mixture. Results for bank 7-12: 14.2 idling, 13.6 @ 2000 rpm. Also (slightly) too rich.

Then I re-installed the original lambda sensors and measured air-fuel ratio at the tail pipe using the wide-band sensor. Result for both banks: 16.7 = way too lean.
I'm lost: mixture entering the (new) cats is too rich, mixture leaving these cats way too lean.

Note: for reference I also used this wide-band sensor at the tailpipe of my E30 M20 (with cat), and the gauge showed 14.7 being spot on.

What is going on if a cat turns a rich mixture into a very lean mixture (with associated emissions I guess)?
Screenshot 2022-05-08 at 10.15.43.jpg

[Timm]

Wide-Band Lambda sensors have a 1.47V output at 14.7:1 air/fuel ratio - and the output goes higher the leaner the mixture.
Narrow-Band sensors have a 0.5V output at 14.7:1 air/fuel ratio - and the output goes LOWER the leaner the mixture

So, sticking the wide-band sensor in will have skewed the adaptations in the wrong direction making the mixture more and more rich - once the narrow-band sensors were re-fitted, if the adaptations were not reset then the mixture should have stayed rich until the adaptations pulled the mixture back.

Did you reset adaptations?

[Albert850i]

Thanks Timm, I did not connect the wide-band oxygen sensor to the DME's, which were running open-loop during the measurements.

[Albert850i]

SOLVED

As the engine was misfiring and limping all the time when I bought the car, I installed two Aliexpress digital air-fuel gauges on the dash years ago. To monitor which bank was actually running, and in open / closed loop.
Because I recently compared the readings with my new wide-band lambda sensor, I noticed that the gauges were showing a lean mixture where the wide-band was showing rich at the same time. So finally, after 4 years, I did the following: measured the oxygen sensor bias voltage with the Ali gauges connected (0.35V, showing ±16 A/F) AND DISCONNECTED (0.45V = correct). Conclusion: the gauges were pulling down the oxygen sensor bias voltage by 30%, making the DME's erratically believe the mixture was too lean. Which explains the excessive positive adaptations and fuel consumption!

Result: mixture adaptations LL (idling) back to around zero, perfect. Bank 2 still shows too high (+26) adaptation during TL, which may be a clogged injector or so, will check. Cannot be a vacuum leak as adaptation 0 during idling.

Getting there...
Screenshot 2022-05-22 at 11.02.14.jpgScreenshot 2022-05-22 at 11.02.06.jpgScreenshot 2022-05-22 at 11.01.57.jpg

[Timm]

Ahhh, it was a trick question right from the start! The hardest thing to mend is something that has been fiddled with......

[Albert850i]

Thanks for your help Timm.

[Albert850i]

Checked all original injectors with a garden hose, see photo. Result: one of bank 7-12 was stuck closed. Replaced all 12 injectors with the new (blue) Bosch model, bought cheaply on Ali.
Engine is running perfectly. Still I need to give the DME's time to learn new adaptation values now I guess.
Note that I measured intake vacuum pressures: bank 1-6 = -550 mbar, bank 7-12 = -470 mbar. Too big a difference, so I will again refurbish the throttle bodies.
Screenshot 2022-05-26 at 15.29.20.jpg

[BMSman]

-550mbar/-470mbar... was this measured at idle? seems a bit low if so.

[Albert850i]

Correct: idling, warm engine, measured at fuel pressure regulators.