PC based CNC system architecture/Black Box
Posted by
Doug Fortune
on 2000-12-09 15:18:10 UTC
hi Jeff, just some comments on your comments:
Jeff Barlow wrote:
We use it because it does the job cheaper than any other alternative.
Also, you would not use a generic OS for Real Time control, just as
you would not use a hammer to replace a screw driver!
RTL working vs being turned off), with a suitably sized cpu.
the parallel port Step & Direction works quite well. It could always be
better, of course. Depends how much you improve the situation vs
how much you complicate (and spend).
and if I had any time, I would put a number of timers and bit I/O
lines onto a DIMM stick that fits into a free memory slot (and emulates
memory I/0, but of course is Not memory) where the cpu has the absolute
minimal latency to respond to interrupts, and do I/O to the outside world.
Doug Fortune
Jeff Barlow wrote:
> Multi-axis motor control is clearly a real time process. Generic PCThe original design intention of the hardware is of course irrelevant.
> hardware and OS's were not designed as real time processing tools.
We use it because it does the job cheaper than any other alternative.
Also, you would not use a generic OS for Real Time control, just as
you would not use a hammer to replace a screw driver!
> EMC (for example) is proof that a PC can be successfully pressed in toYou bet.
> service as a real time controller.
> Among the costs I see are: The installation and system integration isagreed
> messy enough to drive many folks away;
> the RT subsystem negatively impacts the performance of the user interface;not with Linux/EMC - I would challenge you to notice any change (of the
RTL working vs being turned off), with a suitably sized cpu.
> the performance of the stepper drive version is rather limiting.all the world is a compromise ...
> We seem to be in general agrement that hooking motor drivers directly toI think the general agreement is the contrary - for the reduced complexity
> the PC printer port doesn't really cut it.
the parallel port Step & Direction works quite well. It could always be
better, of course. Depends how much you improve the situation vs
how much you complicate (and spend).
> It is tempting, for many of us, to try to go off by ourself and designcost benefit. I actually see the biggest problem as hardware timers,
> this "black box"....
> To put that another way, I would like to see us more fully define the
> functional role of this "black box" before we get too carried away with
> figuring out just how it works.
and if I had any time, I would put a number of timers and bit I/O
lines onto a DIMM stick that fits into a free memory slot (and emulates
memory I/0, but of course is Not memory) where the cpu has the absolute
minimal latency to respond to interrupts, and do I/O to the outside world.
Doug Fortune
Discussion Thread
Doug Fortune
2000-12-09 15:18:10 UTC
PC based CNC system architecture/Black Box
Wally K
2000-12-09 15:42:16 UTC
Re: PC based CNC system architecture/Black Box
ballendo@y...
2000-12-09 22:27:38 UTC
PC based CNC system architecture/Black Box
Ian Wright
2000-12-10 06:09:03 UTC
Re: [CAD_CAM_EDM_DRO] PC based CNC system architecture/Black Box
Chris Paine
2000-12-16 19:15:29 UTC
Re: [CAD_CAM_EDM_DRO] PC based CNC system architecture/Black Box
Jon Elson
2000-12-16 22:12:17 UTC
Re: [CAD_CAM_EDM_DRO] PC based CNC system architecture/Black Box