I also noted that non guided dithering happens more on one side than on the other. At the beginning I thought it was the mount drifting, but then I realized that such a drift over the small amount of time (10s) between exposures (30s) would result in extremely elongated stars, which is not the case. I have still to understand where the wrong guide pulse is generated, if in kstars or in the ioptron driver. I may be wrong, but I seem to remember that this behavior started when I updated my GEM45 to the v3 command set (which, by the way, introduces a new set of commands for guiding).
Nevertheless, I searched a bit in the forum, and found out that a similar behavior was seen also for guided dithering, and then Jaseem fixed the "random dithering not being not so random" (github.com/KDE/kstars/commit/6e4d0c91846...202a1ef28432c08de4fd). Maybe a similar fix is needed also for non guided dithering? I analyzed the logs, and even if on the long run the direction and amplitude of dithering are uniformly distributed, there are often sequences of 5-10 pulses with the same direction and almost same amplitude.