Re: Re: EMC Running My Controller

Regarding the 5 kilohertz limit on sending pulses to stepper motors via
the parallel port, I am guilty of this figure. I don't know what the
hardware limitation is, but I have also heard (from Matt Shaver?) that
cutting out the capacitor will square up the bits nicely and let you run
faster.

The 5 kilohertz figure comes from:

1. The minimum reliable RT-Linux task period is 100 microseconds. I've
run a test routine at 50 microseconds, and it worked fine, but this is
pretty much the limit.

2. The stepper task outputs either an up pulse or down pulse each cycle,
depending upon the speed of the move. The maximum it can run is up one
cycle, down the next, up the next, down the next, ... so the period of
the full pulse is twice the task period, or 200 microseconds.

3. 200 microseconds/pulse = 5 kilohertz pulse rate.

The rate of the stepper task is entirely determined by the INI file
parameters [TRAJ] MAX_VELOCITY (in units/second) and the greatest of all
the [AXIS_#] INPUT_SCALE values (first one on line is scale in
steps/unit; second one is an offset that's not currently used).
Multiplying the two yields the steps/second required for max feed, and
double this gets the task frequency in cycles/second. If this gets above
5 kilohertz then you're taxing RT-Linux, but try to max it out and see
what happens. Maybe 10 Kilohertz or higher is achievable if you have a
fast CPU, and clip the caps on the parallel port. I'll try this and see
what I can get.

--Fred