Re: [CAD_CAM_EDM_DRO] Hardware step generator idea

Chris Stratton wrote:

> I was thinking last night about the timing problems of PC-based step
> pulse generator (both frequency and speed resolution at high speed)
> and came up with an idea I'd like to bounce off folks.
>
> What about running EMC in servo mode (with encoders) and writing a
> driver that takes the number which would be converted to +/-10v by the
> STG card and instead uses the magnitude to control a simple pulse
> generator and the sign to control the direction bit. (This presumes
> EMC can handle velocity mode servo amps). This might even work with
> the G320 step-servo drivers (would they then behave like a closed
> velocity loop?)

Yes, I've already done this. EMC was originally a servo only program, and
has only recently been changed to work with steppers, too. It works
great, and was demonstrated at the NAMES show last month. I am
planning on making a 4-axis board with optional encoder inputs, stepper
outputs of both step/dir and phase signals, and digital I/O. I hope to make
it available for $200, but I will not know if that will fly until I get the
board designed.

> It seems to me that this card could take the form of an 8254
> programmable timer (3 channels) running from a suitably fast clock
> source and something (74hc373, 8255) to provide bits for the direction
> signal. (Software might also need to control the 8254 gate output when
> zero speed is commanded)

These 8-bit timers do not have enough timing resolution to work well.
I am using a 24-bit counter running at 10 MHz, to give resolution of 100
nS, and a minimum speed of .6 steps/sec (below that, the computer has to
turn the pulse gen on and off).

Jon