I was running a sequence tonight and had both 1 degree temperature based auto focus enabled as well as auto focusing every 30 minutes. During my sequence, the 30 minute point came each time, before the temperature difference was large enough. Each time it did, PHD was not paused during auto-focus (even though I had that enabled) and it killed tracking and my sequence.
Luckily I was at the helm watching, and noticed this issue. I corrected tracking, and let the sequence resume assuming it was some anomaly. When time based auto focus kicked back in, PHD again was not paused and the sequence tanked.
I attempted to reboot, load up SGP and PHD again, and the issue reproduced.