Any clock change will cause PLL to have to re-lock on my with my Nvidia Shield Video Streamer which I am currently going in AES from USB. I've also tried USB in and Optical on that source and it doesn't fix the issue. The AES DDC is a Singxer SU-2 KTE so it should clean up any clock problems but it doesn't. This is a pain, because the clock pretty much changes every time you pause or stop a video for about two seconds. Sometimes I turn off PLL so this doesn't happen with this source. The quality of the video source material is 90% of the time inferior enough that you wouldn't notice the difference with PLL on or off.
But with my Roon Endpoint that is very high quality, I don't get that kind of delay, except with when switching to DSD it is about 1/2 a second and a light pop.
When I first got the May, I tested it with PLL on and off (just very quick switching between states) and I didn't hear enough of a difference with it on or off so I left it off. I've done a lot of work with cables, power delivery and grounding since then. And I recently did a comparison with several day comparisons, and the PLL gave it a leap forward in terms of liquidity, spaciousness, tone etc. So now I keep it on pretty much all the time.
I talked to Tim at Kitsune about this problem. I suggested that a improvement would be to allow you to switch PLL on and off for each input. This would solve my problem. He said it was a very good idea, but not something that could be fixed in firmware. The PLL issues are by far the biggest issues users report with the May, I'm sure future versions of the product will deal with this somehow.