Re: [CAD_CAM_EDM_DRO] Re: step pulse timing resolution

Art Fenerty wrote:

> Sorry mat, I misunderstood your question. I think that using 8192
> interrupts/second and then always interrupting at that frquency, I
> should be
> able to make a table of how often to pulse. This gives me a
> granularity of
> 122 us (roughly) and then by using a common algorithm (similar to that
> used
> for drawing lines in x,y axis) pulse at the algorithm calculated
> times. It
> won't be perfect , but it just might work to make a very smooth linear
> ramp.
> On the
> other hand I could be wasting my time, but thats what programming is
> all
> about.

Yup, the Bresenham line routine looks good, until you realize you need
to accelerate
smoothly at the beginning and decelerate at the end of the move. That
makes it a
lot more complicated. Also, moving at 8192 steps/sec is unattainable
(with a simple,
as opposed to computerized microstepping drive that smooths out the
rate) because
the next lower speed is 4096 steps/sec. That is a 2:1 jump, and no
stepper
motor could follow it. Even the next lower jump , from 2730 Steps/sec
to 4096
won't work. Regularly scheduled interrupts won't hack it, unless they
are going
at a MUCH faster rate, or you resign yourself to about a thousand
steps/sec
as the maximum rate.

Jon