The guide logs show a somewhat different story than what you are remembering. The same problem happened on both nights at the same sky pointing position. And guiding didn't "stop" - it ran continuously but was completely unable to recover or control the huge tracking errors you had. PHD2 tried multiple times to alert you to the problem, but you had chosen at some point in the past to suppress that alert. That said, it's hard to identify a specific source of the problem just from the data we have. Here's a look at the events on the two nights - RA is in red, Dec is in green:
September 5:
September 6:

Notice the timestamps on the two different nights - very nearly the same - and the y-axis values to the left. These are gigantic tracking errors. When they occurred, PHD2 continued banging away trying to correct them but it was a hopeless situation. If we measure the rate of displacement in these two cases, we see tracking errors of 186 arc-sec/sec and 166 arc-sec/sec. Compare that to normal sidereal tracking which is 15 arc-sec/sec. So even if the mount had stopped tracking altogether, it wouldn't have produced this result. And of course, since you were running PPEC for RA guiding, all of this got baked into the model it was trying to build. I think you probably have some substantial mechanical instability in your guiding assembly that is showing up in this sky location. It could be the guide scope/camera moving around or it could be a cable that is pulling on the assembly, something along those lines. I suggest slewing the scope back to that location and looking very, very closely for something that is disturbing the guiding assembly or is not allowing the mount to track smoothly. The pointing position of interest is side-of-pier west, Dec = 36 degrees, and RA 1.7 hours west of the central meridian (about 25 degrees)
Hope you can track it down quickly,
Bruce