Ran the Tau test earlier today, turns out not much difference between next, myNext without triflight, and myNext with triflight. Next was 0.0429, myNext without triflight was 0.0474, and myNext with triflight was also 0.0474. The computed gains were also very close among all three. The only difference was in some of the bias numbers, and I think this was likely due to triflight biasing the tail to the trim position, where as the other tests the tail was just trimmed vertical.
Each test started with loading the firmware, erasing all the settings, and then setting up from a clean slate. Each autotune was done on a fresh battery. The tests done without the triflight algorithms had the servo feedback wire disconnected from the FC.
The test rig was my Armattan Tri, Cobra 2204 1960 kv, LittleBee 20 amp ESCs, 5/4/3 blade bull nose props, DTFc flight controller.
Here's the test rig:
Results from next:
Results from myNext without triflight:
Results from myNext with triflight and a prior ground and airborne tailtune: