This is something i have been looking into based off myself and another member here (as well as others i am sure that i dont talk to) emulating our o2 signals from a better wideband signal.  It is a relatively minor voltage data issue in the grand scheme of the operation of our highly modded vehicles, but none the less i feel the difference is enough to cause other erratic afr symptoms like LTFTs slowly going out of whack because its switch points for 14.7 are indeed wrong, even if by just a little tiny bit...

I am beginning to think there is a voltage offset of 7(?? hypothetical number i measured once at idle) millivolts on the negative line in the stock b5 s4 o2 sensor wiring and or sensor, thus skewing the actual output of values the oem ecu expects to see from the oem sensors, so then there are maps that filter this data in addition to the normal filters (I have seen them listed in FR), they recalculate the o2 sensors actual output data of (hypothetical) .4493 as .4500 (1.0/14.7) when there is a preconceived offset of .0007.

I think for our emulators we need to setup the outputs to account for this offset.  I am not even actually sure if it would be plus or minus this small amount of voltage as I am not sure how the ecu reads it.  The wideband manufactures and basically no one else does this because they dont care about 7 or so millivolts and every car, almost every last engine/chassis could be different when modded.  But i began thinking, if audi/bosch created maps for this very purpose, maybe it does throw of fuel trims.  I am going to try and find more on the maps and also run the car with my fluke to the factory plugs under hood and measure the negative line while driving around.

It is likely not going to affect anything other than cruise and idle but I want this to be spot on perfect data to the ecu.