I have a 2001 Honda Prelude SH (Honda H22A4 engine). I was trying to determine how the reported short/long term fuel trims fit in to the fuel control loop (in particular I was wondering if the long term and short term values were added together to provide the actual applied value -- they aren't).
The ECU uses the short term adjustment to alter the injector duration, and therefore the mixture, in order to make the oxygen sensor voltage swing around 0.6V.
Over time the ECU will look at the average short term oxygen sensor adjustment and determine if the engine is running rich or lean overall. The ECU will alter the long term oxygen sensor adjustment based on the average value of the short term oxygen sensor adjustment.
According to What exactly are fuel trims?:
Short-term fuel trims (STFT's) refer to the instantaneous fuel trim correction being applied by the engine management in response to rapid changes in throttle and load.
Long-term fuel trims (LTFT's) refer to corrections that are "memorized" by the engine management. This will factor in operational deviations like an leak in the intake tract or degraded MAF sensor.
And some random readings from my car over a period of about 12 minutes while driving:
The descriptions combined with the graph seem to confirm that the reported long term value is pretty much just a running windowed average of the short term value.
The short term value seems to be what's actually applied. The long term value is just the short term average over time.
So my question is this: Does the ECU actually do anything with the long term fuel trim value and, if so, what does it do? Or is it merely diagnostics info for humans?
If the long term value was being used, at least in closed-loop mode, then I would expect short term readings to be centered around 0, but it seems like all action is represented entirely by the short term values, and the long term values are just a report.