opto-isolator pitfalls

Whenever you insert opto-isolators into a signal chain be sure to
read the data sheet. The main thing to watch out for is propagation
delay. The turn-on and turn-off times can be substantial. A lot of
optos have delay times in the order of 100us, which limits them to
about 5kHz. This is certainly too slow to be used to isolate shaft
encoder outputs and might even be too slow for step/direction signals
on some systems. Of course much faster opto-isolators are available.
The other thing to be aware of is that turn-on and turn-off times are
usually not identical. If you use optos to isolate step/direction
signals this will add the requirement of longer setup/hold times for
the direction signal as the active edge of the step signal gets
shifted in relation to the direction signal. This could cause
problems of lost steps in one direction like Boris described in msg
32835.
Another problem associated with opto-isolators is that the output
rise/fall times can be fairly slow (depending on model). This can
cause the signal to "linger" around the switching threshold and makes
it more likely that noise gets picked up causing a double clocking. A
short connection to the following stage or the use of active output
opto-isolators will improve things a lot.
Don't get me wrong, opto-isolators are great tools when used right,
but like most things in engineering, it's a good idea to look at the
details. Read the data sheet and know what to make of it!
Klaus