Guiding errors are absolute errors, but the guiding deviation error monitor code does not see that.

For example…. let's say we set a guiding deviation error of 2 arc seconds, before canceling an exposure.
If we get an error of +1.5 then no exposure suspension is done. But let's say the error goes from +1.5 to -1.5 over the course of the exposure. That's an absolute error of 3 arc seconds, and would ruin the exposure.
This would explains why I see some trailed images, when no deviation error was detected.

I suggest EKOS measure the total ABSOLUTE amount of error over the course of the exposure. That would eliminate this problem.

Peter