Need help diagnosing mild lean at WOT issue

[jnshk]

I'd like to pick your brains on a technical issue I'm having. My car recently started getting a bit lean at WOT, and though I suspect it's possibly a fuel pump issue I'm not 100% convinced that it isn't some sort of electrical issue. I'd like some more opinions. Here are the pertinent details:

The engine is a 1997 1.8L which is (for all intents and purposes) "stock" in all the ways that matter. I have a wideband O2 gauge as well as access to OBD2 data (primitive though it may be on the early cars). The fuel pressure regulator is stock. The fuel pump is a Deatschwerks DW200.

I've run some tests on fuel pressure and can state the following:

Deadhead fuel pressure on fuel pump is ~85. (**See note at end of this post!)
At idle and vacuum cruise fuel pressure is about 35-36.
At WOT (atmospheric MAP) fuel pressure shoots to 40-41.
If running through the RPMS to redline at WOT, fuel pressure stays solid at 41.
All of these are normal/expected values for a '97 NA8, based on the published specs that I've seen.

When engine is shut off fuel pressure drops to 29-30 quickly.
I don't know if this is normal?
Long term fuel trims and short term fuel trims are all close to zero and tend to be pretty consistent/stable and on the negative side (in other words, pulling just a small amount of fuel).
This is within the expected tolerances.


However, despite fuel pressure staying at peak while WOT...


Once above 3K RPM at WOT, the AFR starts to lean out from 13ish to between 14 to 15 as the RPMs climb. The exact value is somewhat inconsistent, but generally the peak averages around 14.7. Things get slightly richer again once it reaches 5K+ RPM, as it gets into low 14s and high 13s, but this should normally be in low 12s and high 11s at this point, based on prior datalogging.
Although I considered that possibly the wideband gauge is misreporting due to some error, the narrowband signal data from the OBD2 backs up my observations from the wideband gauge. The car is definitely leaning out under those conditions.
Below 3K RPM, the engine is sometimes in closed loop so maintaining ~14.7 AFR, but if not in closed loop then it tends to be just above 13 (when it should be slightly closer to 13).
If running up through the RPMs at slightly less than WOT the AFRs tend to be correct (starting around ~13 and then dropping to ~12 by redline).


So it seems that the only time I'm not getting enough fuel is during open loop operation at the highest loads. Since the ECU seems to ignore O2 feedback during open loop, maybe it's possible that commanded injector pulsewidths are expecting more fuel to be available and thus running slightly lean? I would expect to see a pressure drop if it were insufficient fuel flow though, right? Could the injectors be failing? :dunno:

**NOTE ABOUT FUEL PUMP: The DW200 should be capable of 100 psi (according to documentation, or 95 according to at least one real-world test) when deadheading, but I'm seeing only 85 psi. This may be due to slightly lowered voltage because the test was performed without the engine running, but I'm thinking that even at 12 volts (as opposed to 13.5 as specified) it should still generate closer to 95-100 psi, right? The stock fuel pump is only specified for 75-90 psi, so even if the DW200 were not at max capacity I would have thought it would still be up to par with stock, but perhaps it's not able to generate enough flow in addition to the lack of pressure?

[jnshk]

Additional Data (generated as response to a question on Facebook):

The fuel filter was replaced last week (shortly after I noticed the symptoms, but before I started testing fuel pressures) and the battery was disconnected for a few hours so that any fuel trims in the ECU would be reset. Although there was a minor improvement (and I suspect it had more to do with resetting the ECU fuel trims than replacing the filter), the behavior is largely the same.

The car is running on the stock ECU at present (1997, CA-emissions, manual transmission). The injectors are stock. The only thing that is non-stock at the moment is the fuel pump. The O2 sensors have been replaced recently as well. At first I put a brand new Bosch sensor in the front position, and then after reading some reports that the Bosch sensors don't always play nice with our cars I moved it to the rear position and installed a known-good slightly-used Denso sensor in the front position. Data from the OBD2 shows no significant change in signal behavior of one sensor vs the other, and wideband O2 confirms that the fueling remains the same as it was too.

I have been running the car on an MS3 in parallel for the past few years as I've been trying to create a tune that so closely replicates the factory tune that it will "play nice" and pass all of the OBD2 emissions readiness monitors and run without causing any CELs. I've been able to get it to run without causing CELs, but I was still having difficulty getting it to pass the self-tests, so I switched it back to the OBD2 for a while to do some comparison testing. The OBD2 is not clearing the self-tests either, but otherwise seems to be running fine. Then after about of week or two of running on the OBD2 computer I noticed the odd lean behavior at WOT. It's possible that it's been that way this entire time and I just didn't notice because I wasn't holding WOT through that range very frequently.

I suppose I might need to hook up the MS3 again and see if it's running leaner than when I disconnected it--which would indicate a hardware issue and not an ECU issue--but if it doesn't run any leaner then I'm still no closer to a solution (as it may have been tuned to compensate for whatever is happening, if whatever's going on has been a problem for a while that just went unnoticed since I could make minor adjustments for it as I tuned).