Ok, this is great, now I can see what’s happening – but it isn’t as simple as you think. <g> There are actually 2 problems here, one from ZWO and one from PHD2. The ZWO driver is reporting bogus values for the mount guide speeds, it is reporting a guide speed of 1800 degrees/sec. The driver is trying to report a guide speed of 0.5x sidereal but is using the wrong units. So it isn’t complying with the ASCOM specification, which requires all mount speeds to be reported in units of degrees/sec. But we’re all in agreement that PHD2 should never crash, so obviously we’re not handling this kind of gross error properly. The PPEC algorithm is unique in many ways including its use of the mount guide speed to keep dithering events from corrupting its internal model. None of the other PHD2 guide algorithms use the mount guide speed to do their job. So your situation is unique in that you have a mount driver that’s returning bad information about the mount guide speed and a guide algorithm that depends on that information without sanity-checking it first. Wouldn’t you know.
If ZWO provided you with a test version of their driver, ask them to fix this guide speed property and you will be good to go with PPEC. In the meantime, I will figure out the best way to handle this sort of problem because it will no doubt happen again. I’ve been able to reproduce the problem with the simulators so I don’t think there’s any need for you to test further.
Thanks for sticking with this and tracking it down.