Hondata

I've got an odd issue and can't seem to find a fix to it.

I thought my problem was originally just a poorly tuned map, as it was a bit rich, but now that the map is tuned relatively well, (I like my map values to be a tad rich, and use o2 to lean out the map to stoich) I am still having this issue.

Even with closed loop enabled, but S.Trim values set to +/- 0% (so effectively off), S.Trim will still kick in and screw up my tune in very specific scenarios;

Scenario 1)
Hot starts. The car will start up, use the cranking fuel table, then the fuel table and will be right at 14.7, and then S.Trim will kick in, and pull 11% fuel from the map for about 10 seconds, causing the car to lean out to nearly 18.0, and then it goes back to 0% correction, and the car is back at 14.7. Most annoying thing in the world, because I can't drive my car for those 10 seconds, otherwise it will run poorly and very lean.

Scenario 2)
Between EVERY shift, S.Trim will pull that same 11% fuel right at throttle tip in, but only if I get off the fuel entirely, this can be prevented by hovering my foot on the throttle while shifting and can be replicated by decelerating and then getting back on throttle. For about 1 second, S.Trim will kick in, pull 11% fuel, make the car run lean, and then return to 0% correction, where the car runs fine again.

This will happen regardless of what Short trim settings are set at. I just set it to 0% tonight to confirm that it was S.Trim causing this issue. It should not have corrected the fuel AT ALL, since the S.trim settings are set to +/- 0% currently, however despite that being what I'm commanding, it's still wanting to pull fuel.

Here is .cal that I'm currently using plus a datalog of both scenarios.

There is a learned AF trim value that the ECU uses. This basically is to quickly set the short term trim to a value after running in open loop and switching to closed loop. eg after a gearshift. The solution is to tune to fuel tables so that the AF is closer to the target.
There also is a min and max for the learned value (look at a 2006 Civic Si calibration). I'll add this in for the next release for the S2000.

Wonderful news,

I got the answer "The tables need to be closer to stoich" on other boards as well. I normally tune my stuff a tad rich so I'm not running lean with weather changes, but this may not be a viable method with this learned S.trim value.

Hopefully once the values are tuned closer to stoich (and the tune is complete), the learned value will decrease.

Also, a min/max table would be graciously appreciated! The fact that it was a hidden function in the ECU is what was driving me insane.

Thanks