Josef Hanzal wrote:

> I am in the process of prototyping an instrument, which moves in two
> axes using stepper motors.  The stepper motors current is controlled
> via PWM using SGS-Thomson L6223 chip.  Motor voltage is about 30V,
> current about 250mA.  The connection between the mechanical and
> electrical unit is done by a computer grade 25 wire cable, about 6 ft
> long.

  Pardon my intrusion at this point, but this sounds like the crux of
the problem.  Hopefully this can be rectified at this "prototyping"
stage.

  Why is the "electrical unit" not riding on the motor chassis?  By not
doing so, you have introduced a significant design problem which may in
the event result in a more expensive design and a major reliability
issue.  What happens for example if the cable, which I begin to suspect
is required to flex in operation, fails at the limit switch connections?

  A 25-wire ribbon cable is certainly quite cheap, but only if it is
immobile in a protected area (cable run).  A robust flexible cable (I
wonder whether "computer grade" is particularly robust) is likely to
be so much more expensive (including to maintain), that it will be far
easier in the "long run" to provide a second PCB at the motor end, and a
heavier-duty 4-wire cable carrying your 30V plus an RS-422 (or...)
communication pair, with regulator for the logic on the "remote" board.

  In particular, this relieves you from the absolute requirement in your
25-wire cable that the sensor and motor grounds were separate, with the
logic grounds flanking the sensor group:

  M1 M2 M3 M4 MGnd LGnd S1 S2 LGnd S3 S4 LGnd     etc...

  OTOH, if you are using a common DB25 extender lead with an overall
shield connected to the connector "shell" (must ground the "shell"!),
you have a problem in that you cannot determine which conductors are
adjacent.  All-in-all, I think you really *don't* want this cable!
--
  Cheers,
        Paul B.