Okay, a deeper dive inside the JB Victory.
I finally pulled the cover to check and see if my flange bolt was off center as suggested by A Pukeko. Sure enough it was. I also noticed that my belt was riding at the very edge of the flywheel pulley into a pseudo āgrooveā, actually beyond what I imagine are the six grooves to mate with the belt. I also noted a bit of wear in that last, outer āgrooveā.
So, I pushed the belt back towards the flywheel shaft while slowly spinning the drive wheel to put it back into position.
Unfortunately, as soon as I started turning the drive wheel clockwise, the belt would jump back into the outer āgrooveā. So, I went ahead and pulled the belt off to check runout on the flywheel and see if the off center pulley flange bolt was causing some wobbling that might be bumping the belt outwards. I am getting about 2-3 thousandths of an inch (~0.05 mm) which is fine by my standards.
I went ahead and unbolted the flywheel pulley (10mm flange bolt) and pulled the flywheel assembly out. Everything looks good there. I inspected the optical target and coils, and didnāt see anything out of sort. I also didnāt see or feel anything majorly wrong with the flywheel bearings. The construction and design quality seem quite good actually.
Also, as I noted previously, the pulley flange bolt hole is strategically positioned slightly off center of the flywheel shaft to give more clearance for the key groove needed to lock the shaft and pulley together.
But, check out the premature wear on that outer āgrooveā. Iāve got about 600km so far on my trainer.
That didnāt inspire too much confidence for belt longevity, so I went ahead and shimmed the pulley with a thin 15mm ID washer during reassembly. That has kept the belt in the 6 inner grooves, and Iām happy to report smooth riding after about an hour of testing.
But, the power accuracy problem is still an elusive issue. At this point Iām convinced itās a software/firmware issue, not a hardware issue. I suspect aliasing or perhaps the āIN REAL LIFE RIDE DYNAMIC (IRLRD) ā responsive momentum to simulate riding outdoors.ā Is the IRLRD changing resistance in response to cadence to simulate the pulsing feeling when riding outdoors? If so, there might be too much latency at higher cadences.
Itās a bit beyond my paygrade at this point, so Iāll stick with my power meter for power and cadence till this gets sorted out.