Re: [CAD_CAM_EDM_DRO] DeskNC Windows version

Hi Art,

Good deal on the joystick! It's good enough for CNC jogging? Mine sure
aren't (old C&H)!

OK, then I must be thinking of something else being 20K pulses or so. I
can see now where you get the 8K rate. 8K could give me 60 IPM on the
Sherline, fine for now (later on to bigger and better machines)!

This P6 ASIC sounds like something to check out. I don't think I have
any, just an old Pentium 180, a 300+ Celeron or two, and my latest
acquisition from my son, a 400 Mhz AMD. Any of those do? You mentioned
legacy 133's, do they have it? I was "fast track" writing diagnostics
for new PC hardware about 8 years ago, but I'm out of the loop now (no
more free machines to play with).

How smooth is smooth? How can one measure it? Something like
distortion products or sub-harmonics on the step pulse frequency (over
my head again)? Should I be able to see/detect "un-smoothness" (sic)
with a 'scope? I was going to hook up some speakers to the step lines,
and listen to them! I also thought about using a program called
"Spectrum" to "look" for birdies!

Thanks again for your replies and answers! This is interesting
material!

Alan KM6VV


Art Fenerty wrote:

>
> Alan:
>
> Actually 8000 pulses per second is pretty good. EMC for example, at my last
> reading was limited to 4096 steps per second. Even DOS or normal Linux is
> hard pressed to exceed 8192 pulses per second and maintain any real-time
> capabilities. This is due to the fact that you can olny interrupt the system
> at a rate of 8192 times per second without using some kind of timing
> generator. The last time I checked on the EMC kernal, it interrupted at 8192
> but because it only set a pulse on one interrupt, then cleared it on the
> next, its final frequency was 4096 steps per second. (I'm sure I'll get
> corrected if I'm wrong :) )
> The P6 chips with a built in ASIC can be programmed to interrupt at
> any rate you wish. Since it uses the CPU Clock to achive this, It uses a
> timing of 1.2GHZ on a 1.2GHZ machine. And with multiple timers on the chip,
> it means you could set an X,Y and Z Timer at different and dynamically
> changing frequencies to step the motors. This gives a much higher speed, and
> an increased resolution in terms of the asyncronous nature. The problem is
> legacy, even though its possible to probably beat the performance of those
> black box pulsing units, most people are not using a 1.2GHZ machine for
> their CNC work. (But we'll get there eventually). Using the CPU clock for
> timing makes things much better. My 500Mhz pentium can time to about
> 2nanoseconds for accuracy of the pulse. It makes Master5 smoother than most,
> though you can never really achieve linux real-time smoothness due to its
> pre-emptive interrupt capability.
>
> PS: Just got a wingman extreme joystick hooked up to Master5, kinda neat,
> throttle for feed overRide, Joystick for multiple speed Jogging.....kewl..
>
> Art
> ----- Original Message -----
> From: "Alan Marconett KM6VV" <KM6VV@...>
> To: <CAD_CAM_EDM_DRO@yahoogroups.com>
> Sent: Wednesday, April 18, 2001 2:49 PM
> Subject: Re: [CAD_CAM_EDM_DRO] DeskNC Windows version
>
> > Hi Art,
> >
> > Thanks for the reply. I'd heard that parallel was impossible, and
> > serial possible under windows. Is there any EPP/ECP support under
> > windows? Might that work better, similar to serial? But then, it
> > wouldn't be a "simple" parallel port, I guess!
> >
> > By 8000 PPS, are assuming a 133Mhz Pentium class machine on Windoz? I
> > thought I'd heard higher numbers, must be for Linux or DOS.
> >
> > I'd also like to hear about the P6 instruction(s) that aid in
> > timing/step generation.
> >
> > Alan KM6VV
> >
> >
> > Art Fenerty wrote:
> > >
> > > Alan:
> > >
> > > When sending to a serial port from windows, you don't hang the
> system
> > > as much. You simply send a series
> > > of commands and let the intelligent box do the rest. Most windows
> software
> > > suffers greatly when sending to the parrallel port due to the time
> involved.
> > >
> > > ""Sounds like a FlashCut!"" -- It is, thats just how flashcut does it
> for
> > > the most part. It tends to make it very smooth and is a good solution
> for
> > > the money.Without a black box, you really can't get higher than 8000
> > > Pulses/Sec.
> > >
> > > Getting Master5 to be smooth under windows was a real challenge and
> it's
> > > easy to see why Black box idea's are starting to proliferate. You just
> can't
> > > go faster than 8000 pulses per second without such a black box, although
> > > that may change under the P6 series chips. My experiments show that
> variable
> > > asynchronous velocities are possible with P6 processors at quite a
> speed. I
> > > may try to implement such a system in a year or so. But it will take
> awhile
> > > for most peoples computers to catch up with the latest motherboards. I
> have
> > > enough problems with the number of people out there using 133's.
> > >
> > > Art