You must interpolate values not do that big jump to minus values.
Better effect You can make by modding fueling on deceleration. (CWSAWE, KFNWEGM, KFTVSA)
Actually no, this is not an interpolation issue at all. And claiming this shows lack of understanding of the torque structure.
First of all load should be logged with throttle released.
The ignition should be changed only at this load and below (so the axis values in the ign maps might have to be modified, but usually not).
The reason for waves in power delivery is not changing KFZWOP/2 to suit, because as you transition in the load curve, torque calculation goes completely negative very suddenly (and even if you interpolate some things, which will actually introduce a TON of driving issues on low part throttle).
Because of the jump in calculated torque, you trigger LSD, sometimes even DASHPOT a second time and ARMD.
The result is ECU messing with torque - increasing it, then it sees that suddenly torque sharply increased, etc etc you get a feedback loop.
If KFZW and KFZW/OP are tuned exactly the same then as you pass over the boundary the calculated torque does not change and there are no issues.
None of this is a problem on newer ECU's where ZWOUT gets set to ZWSPAE on overrun anyway.
