08/98 323i w/ M52TUB25
new Bosch 13477 O2 sensors
EU2 MS42 DME flash (7528736)
I've had headers and EU2 before I started tinkering. The car ran good except the exhaust leak and poor fuel mileage, which I attributed to unmetered exhaust leaking before the precat O2's. My ebay headers had flanges warped to hell and leaked, so I took them off and put them on a Bridgeport to mill the mating flanges flat. New studs and nuts were used on assembly to ensure no leakage. This fixed my initial reasoning why I started this, but a can of worms has been opened.
Upon first startup, car ran with no issue. After a few trips, it started to stumbling and running very rough and lack of any acceleration. Codes were set for "precat O2 min value undershot." Was able to stop and clear codes and adaptations (for s&g's) and get home without issue. To me, classic symptoms of swapped O2's, no problem, swapped them.
Now that O2's are swapped- for first ~3 minutes on startup, the car runs very rough. Feels like very (very) bad misfires and cannot maintain idle, sometimes dies out. Throttle input gives no response, sometimes bogging the engine. After this first ~3 minutes, car runs fine and drives like a bat out of hell.
Diagnostic tools available to me are INPA and ScanMaster-ELM. I know, INPA isn't good for what I am up against, but its what I have in the moment. Initial findings with INPA show (while idling) my fuel trims remaining constant. (Error b/c unplugged, but those are the exact values I was seeing.) I am unsure if there are X number of drive cycles before fuel trims start setting after adaptation reset, but in hindsight, I really shouldn't have reset it.
My next step was to look in ScanMaster-ELM. I plotted the precat O2 data and I was stunned. My precats were reading constant (0.022v and 0.025v respectively). If I unplug the MAF to force open loop, I see the sine waveform I would expect, but MAF plugged in I have constant, flat-line readings.
Currently, I am going to go verify that O2's are in correct position by unplugging one and seeing if it loses connection on the software end (engine off of course). I have OBDFusion adapter on the way, but I won't receive it in time before I leave back to school, so this will likely continue into May.