This is pretty unusual and I can only guess at the cause. Here's what things looked like:

This doesn't look like the usual, sine-like periodic error you might expect which is why the PPEC algorithm isn't really addressing it. Instead, about every 10 minutes (probably the worm period), you encounter a period of very poor tracking for about 4 minutes followed by a a roughly 6-minute interval where tracking is pretty good. So I think it's pretty clearly a mechanical problem that you will need to dig into. Assuming that you're still using a gear-driven drive setup, one possibility might be that the worm mesh changes slightly depending on its rotational position. In the time period when you're getting the poor results, perhaps the mesh is too tight. This could be caused by something moving slightly in a mounting block of some kind or perhaps there's something slightly out-of-round. Or it could be something easy, some accumulation of crud or dried grease. As I said, I am simply guessing but I think you need to open up the drive system and try see what changes from one part of this worm phase to the other. You can also get some help from other sources who are experts in the mechanics of your mount and have specific ideas about how to approach this problem.
Good luck,
Bruce