First, make sure you’re starting both tracks at the same time. The song could have its own timing fluctuations. Test both with keylock and without. The video posted does seem to show an overheating motor problem, and they have the Ms right up against a hot mixer and turntables right up against the outer sides of the Ms. Mine have never done that, but I’ve never had mine in a hot environment or with equipment crowded against them like that. They do have fans built in, though, but you would think overheating should not affect the ability of the motor’s timekeeping to work effectively. Perhaps setting them to low torque mode also makes overheating less likely? I usually keep mine on low torque. I would be curious if this is also only happening in places with higher voltages than the US, which would indicate a power supply issue on the Ms.
When keylock is on, are you hearing a triggering of the scratch algorithm and a sudden wowing of audible tone pitch? The Ms do not keep time based on constant record (the top plastic platter, not the bottom aluminum one) monitoring like DVS. Usually, the physical motion is being ignored until a threshold is reached, then you have a very narrow region where you can’t hear the wowing when keylock is on but it’s still barely possible to drag and nudge (even harder to do than drag gently), and soon after this (maybe too soon) you reach the threshold for the scratch algorithm which will disable the keylock. So I’m wondering if your motors may be having issues where it’s triggering just the bend threshold but not so much to turn off the keylock with the scratch algorithm.
You could try playing a timecode file on the Ms (keylock off, obviously) and using DVS software, even free Mixxx, and see if the stated speed % from zero is changing on the computer screen. When you’re not dragging or nudging the edge (of either platter) it should be just as stable, perfect actually, with the motor running as with the motor off or the non-Ms. If everything seems stable with DVS, then you know it’s not happening with keylock off. Of course, you can’t use this DVS method to check with keylock on. I think I have noticed a subtle, counterintuitive speed fluctuation that briefly occurs seemingly only when changing the speed on the Ms when keylock is on, but no one else has confirmed it, I can’t check it using DVS, and it is not the issue you state.
You could be having another issue with Elastique, too, or random file decoding/playback timing glitches as occurred early on with Prime and the Gemini MDJs. In the case of the latter, I suppose you would rule that out with other players or software and checking file types. So, it’s not necessarily an issue with your motors, though that’s also a possibility if yours are having motor speed issues so bad you get into those trigger thresholds. If indeed with the motor off you never see this issue, though, it’s leaning to a motor problem, but, again, if you don’t hear wowing then it’s not triggering that second threshold, so it’s very subtle. Then we’re back to whether it’s happening only with keylock on or not. An actual motor problem should be independent of keylock state.