[Bassnut (Fred)]

OK, just a few things in case. You must have the reverse guiding direction on flip checked or the guideing would not work on flip. Yes DL pauses guideing on image download, but you can pause imaging to allow guiding to settle after download before imaging recommences. If you dont have this delay set, stars can look not round due to the 1st few seconds of erratic guide settling of an otherwise good guide graph.

Are you useing auto darks on the guide cam?. If not, and the guide cam image is noisy, even though you slew back to a previously well guided object, it may now lock onto a hot pixel or a different dimmer guide star. Is the reslew coming back to the same guide star?.

If guiding is correct in every way and in every case, then think about this. The guide cam is the only thing the mount/DL guiding knows about, the OTA is outside the control loop altogether, its just extra weight on the mount. Any image defect in the OTA image (assuming good, no problem guiding) can only be a mechanical problem, nothing to do with the mount or DL at all.

So your left with mechanical flex/movement in scope mounting, focusers, draw tubes, mirrors, cameras (all either with the guide or OTA).

Is the guide cam solid on the back of the guide tube?. Movement of the guide cam on its scope/draw tube/focuser alone can cause OTA image problems. Remember ONLY the guide cam itself is in the guiding control loop.

Im banging on about the guide setup particularly, because I spent a long time sorting drift which ended up me bolting the guidecam to the mounting rail, that made the biggest improvement (at long FLs.) Even though your images are OK most of the time, and the inconsistancy is puzzling, there should not be the slightest dought about the whole guide set up stiffness.