Re: [CAD_CAM_EDM_DRO] Logical point to split EMC/Control logic?

With all the problems related to timing issues, I wonder
if it would not be prudent to split the machine control
logic along a clean point, is there one.

I thinking about stuff like the fact physical control
processes need to hit a chink of code say 5 or 10,000 times
per second leading to obcene computing power necessary
to balance that with user applications.

Can this come down to something as simple as an intelligent
controller that will take something like a (Next X,Y,Z) position
and a speed, and then deal with it.

I would think plain old liner iterpolation would be good enough,
and if that ends up being a bunch of lines generated by the
computer based on an arc, so be it.

I'm not talking about any real inteligence, and stuff like look-ahead
or any other logical processing would need to be done elsewhere.

It just seems like the stuff that is on one hand doesn't
take a lot of brains, are the very ones that are such a
pain in the butt as far as timing.


This could be pushed off to a dedicated CPU, I have literaly
thrown away 386 PCs that should be up to it as long as they
are doing nothing else.
Something with a floppy and a chunk of tight running code
that gets commands thru a serial port.

The same thing lends itself even more to embedded controllers.

Is there a clean logical point to do this?

Is everything reduced to a number of linear moves as fixed
velocities at a given level of the code?

Marc