Re: Period setting was EMC error message?

[Fred Proctor <proctor-at-cme.nist.gov>]

EMC stepper motor users,

Ray Henry's post explaining his procedure for setting the PERIOD .ini
file parameter for frequency stepping is right on target. This is the
same procedure I use here.

The idea is to make PERIOD as short as possible, so that there are the
finest possible jumps in frequency between two adjacent frequencies. As
you get near the max, the difference between the max and the next one
down is a large percentage. At the max frequency, you'll see:

10101010...

which is an effective step period of 2 X PERIOD. At the next one down,
you'll see:

100100100100...

which is an effective step period of 3 X PERIOD. So, the frequency
difference between these two is the maximum. For PERIOD = 0.000010, 10
microseconds (pretty fast), the stepping frequency difference is:

1/(2*0.000010) - 1/(3*0.000010) = 16.7 kHz

This means that if you were to jump from the next-to-fastest frequency
to the fastest frequency, you'll get a frequency jump of 16.7 kHz. Since
the output torque of a stepper decreases as the frequency increases, at
some point the stepper can't handle the instantaneous torque generated
by the big frequency difference and you'll stall.

10 microseconds is about as short as you can make the task period. Any
shorter and you'll starve the rest of the tasks and lock up the machine.
Even on really fast machines, there's still the 4 microsecond jitter on
the ISA bus, etc. so buying a faster machine only buys so much.

With hardware assist, you can use a frequency generator running with a
much finer time step so that the frequency difference at runnable
stepper speeds is negligible. This is in progress, and will require
building or buying a board that hook up to instead of the parallel port.

--Fred