So, autotune wouldn't work for me. And then it would. And then it wouldn't. Never figured out why. I was checking the temperature signal, and the map settings, and so on. The corollary when it started working was that the coolant temp was high enough for it to turn on, the engine was off, I modified a map to turn it on (among other things, was thinking now that it is MAP based autotune, no harm in leaving it on all the time, even if it only works sporadically), uploaded to the unit, disconnected USB, and then started the engine. It has been working as advertised ever since.
Glitches in "normal" operation: When set to only be active at higher than a very high throttle position, it will sometimes add fuel to the first couple map cells of very low load on lift throttle from boost. If set to anything quicker than 300 ms or anything higher than 0.5% adjustment increment, it will sometimes go crazy rich in a few random cells (with adjacent cells slightly lean). IMHO, 0.5% is too much, or 300 ms is too short. It would be better if it took cumulative data for several seconds before making a 0.5% adjustment. That's basically how the stock computer works, I guess, with STFT and LTFT.
I had a plan in mind where I set up auto-tune to run in parallel with closed loop stock ecu operation, to seek something slightly richer than stoich. This fight with the stock computer would have the stock computer trims full lean everywhere at stoich. My plan was then to remove auto-tune from the closed loop part of the map and then manually remove the stock trim amount of fuel across the board in the closed loop parts of the map (actually everywhere, and then let autotune add fuel back in under boost). Flaws in this plan include the percentage in EMU maps not lining up in a nice linear fashion with the stock fuel trims, but not having a way to sync fuel trims with EMU cells, this was still my best plan for getting the maps close without someone to drive for me.
Well, I tried 14.7 setpoint, and the stock computer went full rich, with the auto-tune fighting it the whole way down. Decided as a failsafe plan, going the opposite way would be better. So I tried 14.6, figuring stoich is between 14.6 and 14.7.
No drama.
Huh?
Did it break again in a new and interesting way?
No....
Check STFT1/STFT2/LTFT1/LTFT2. All within about 7% of 0, generally with STFT and LTFT with opposite signs.
So does this mean auto-tune can actually be used to zero out stock fuel trims, and then be turned off in all the closed loop cells?
I expect this happy coincidence is only valuable for a wideband-derived simulated narrowband output, where the EMU AFR and stock ECU AFR line up nicely, and the EMU makes adjustments faster/out of sync with the speed of the stock ecu. More as I explore this further.
Glitches in "normal" operation: When set to only be active at higher than a very high throttle position, it will sometimes add fuel to the first couple map cells of very low load on lift throttle from boost. If set to anything quicker than 300 ms or anything higher than 0.5% adjustment increment, it will sometimes go crazy rich in a few random cells (with adjacent cells slightly lean). IMHO, 0.5% is too much, or 300 ms is too short. It would be better if it took cumulative data for several seconds before making a 0.5% adjustment. That's basically how the stock computer works, I guess, with STFT and LTFT.
I had a plan in mind where I set up auto-tune to run in parallel with closed loop stock ecu operation, to seek something slightly richer than stoich. This fight with the stock computer would have the stock computer trims full lean everywhere at stoich. My plan was then to remove auto-tune from the closed loop part of the map and then manually remove the stock trim amount of fuel across the board in the closed loop parts of the map (actually everywhere, and then let autotune add fuel back in under boost). Flaws in this plan include the percentage in EMU maps not lining up in a nice linear fashion with the stock fuel trims, but not having a way to sync fuel trims with EMU cells, this was still my best plan for getting the maps close without someone to drive for me.
Well, I tried 14.7 setpoint, and the stock computer went full rich, with the auto-tune fighting it the whole way down. Decided as a failsafe plan, going the opposite way would be better. So I tried 14.6, figuring stoich is between 14.6 and 14.7.
No drama.
Huh?
Did it break again in a new and interesting way?
No....
Check STFT1/STFT2/LTFT1/LTFT2. All within about 7% of 0, generally with STFT and LTFT with opposite signs.
So does this mean auto-tune can actually be used to zero out stock fuel trims, and then be turned off in all the closed loop cells?
I expect this happy coincidence is only valuable for a wideband-derived simulated narrowband output, where the EMU AFR and stock ECU AFR line up nicely, and the EMU makes adjustments faster/out of sync with the speed of the stock ecu. More as I explore this further.