Ken Roach
Lifetime Supporting Member + Moderator
they're inside a periodic task that allows time for the IO to update before next scan
That's what I call "working by coincidence"; because the task is executed less frequently, then the probability that the I/O timeout occurs between the Start and Stop bit logic execution is much lower.
I first encountered this phenomenon on a PowerFlex 40P installation with ControlNet, on a corn chip packaging line. Because the stop/start action of the drives was very frequent, the opportunity for them to "not start when requested" was also frequent.