Re: glitchy step signals

> To recap: Use 'HC' (not 'HCT') logic between your PC and stepper-drives! Do

The different logic families have different ideas of the diving line
between what the consider a zero and what they consider a one. This
will undoubtably affect their noise immunity, although I'd be hard
pressed to figure out which is going to be better without really
thinking about all the specifics of a particular case.

> On my successful PC-controlled machines, I have always put a 74HC541 (octal
> buffer) between the PC printer-port and "next", and then that "next" is a
> couple of 74HC4538 monostables set for 4 uS, each, and, for driving the
> opto-isolator inputs of most stepper-drivers, I use a GOOD O.C. "sink" driver
>
> IC, last. The opto-isolators work WONDERS! I know, as I had built a
> still-working home-brew lathe (run by a long-defunct CBM "PET"!!!) where the
> parallel port is directly connected to the "TTL inputs" of my stepper drivers
>
> (the old unipolar L/R kind), and "where did THAT come from?" steps show up
> now and then. But NEVER once, since I have been using opto-isolator input
> drivers.

Part of the problem with direct connect signals is that the computer
and drive can disagree on where 'ground' is once you start running
substantial pulses of current through the power supply leads, which
are both resistive and inductive. With an optocoupler you are making
a current loop with the current flowing through the LED, which is more
noise immune than 'voltage' signalling to ic inputs that draw less
current, and more importantly removing any dependence on the
relationship between the computer's and drive's idea of ground.

If a scope is available, looking at the voltage of "ground" at the
drive vs "ground" at the power supply while triggering off the chopper
or pwm signal may prove quite interesting.

The Gecko's have optoisolated inputs for step and direction.
Something I had not realized until talking to Mariss about this was
that if I wanted to run the encoder signals into the computer as well
as the G210 I would need to optoisolate them, to keep the motion power
(and encoder power derived from it) seperate from the computer power -
otherwise the computer might miss-count ground bounce as encoder
motion.

Chris Stratton

-------------------------------------------------
Christopher C. Stratton
Engineer, Instrument Maker, and Horn Player
ccs@... 617 628 1062
http://web.mit.edu/~stratton/www/brassbuild.html