I tried using Mosaic (Framing) mode last night for the first time on M42 (the easiest target in the sky) and was getting about 85% dropped frames (between those that actually errored out and those just not counted).
Tonight it looks like about 55% lost with very few stacking errors. Even with without the errors, the number of seconds between increments to the total exposure time looks like this.. (randomly).. 10 20 10 20 30 10 20 10 30 20 10 20 10 etc.. averaging out to the above mentioned.. 45% stacked exposure time versus real time.
Obviously it's not going to make exposures while it's moving to the next spot in mosaic grid but that means the little continuously running exposure count down timer is just a placebo. It means nothing.
I think I've figured out how it works though by jumping back and forth between live view and atlas which shows the current pointing box in blue (but unfortunately doesn't update live, thus the need to go back and forth).
It's relying on the natural drift of the sky across the frame (at up to .25 degrees / minute) for most of movement across the mosaic. Let's say we've got a diagonal box to fill. At some point it drives itself over on the right side. From there it tracks for 10 seconds while exposing.. Then stops tracking for 10 seconds which causes the sky to drift 1/6 of .25 degrees/minute or 2.5 arcminutes to the right. Then it begins tracking again while taking the next 10 second exposure.. Rinse repeat. It continues doing this.. tracking/exposing, drifting, tracking/exposing, drifting from right to left across the mosaic until it reaches the left edge, then if it's near the bottom of a diagonal frame, it jumps straight up and continues tracking/drifting left.. to the edge, filling the diagonal left edge of the mosaic, then it runs back down to the right side above where it started the first pass, and repeats this process..
The goal is to maximize the number of exposures while minimizing the number of actual purposeful repositions, and to avoid gear lash so it rarely drives to the left, and when it does, I think that's where we see the stacking errors. Any time it drives to the left and then reverses the motor to track right at .25 degree/minute, it'll lose several frames due to crossing the dead zone in the gear lash (motor is running, but scope isn't moving, so it'll get star trails).. Presumably does something similar with the vertical (altitude) motor as well (stopping to drift, and starting to track), to avoid the lash and again loses extra frames when its forced to reverse directions after a major reposition.
