Re: [CAD_CAM_EDM_DRO] Re: time for microprocessors!

Alan Marconett KM6VV wrote:

> >
> > Why would you want to recompile the part? Why can't you just get
> the
> > PROM
> > bitstream from me, or just by the prom from me?
>
> That's fine, no intention of duplicating or competing with your
> efforts. Just a way to evaluate the scale of such a task, and whether
>
> it is something I would want to investigate (a step up from PIC's?).
> Shucks, I might even learn enough to redesign the data acquisition
> system for an embedded system we build here at work (I know, leave it
> to
> the EE's). But that's what it's all about, learning.

yes, I might put the schematics out in the public domain.

> > No, I am stuck using the SMALLEST Spartan, at 68 pins, because I
> don't
> > want
> > to get into 208-pin devices!
>
> Looks like it may be hard to wirewrap! Do you go straight to PCB
> layout? Ouch! But then, it's all a simple matter of programming! ;>)

You could do this. But, a wirewrap prototype leaves me one step away
from a
prototype board. So, since I have the gear, I make a PC board for it.
Without
plated through holes, it is a bear.

> How about acceleration tables, such that the steps generated would
> follow an acceleration profile? How would endpoint correlation be
> handled, for compound axis moves (if that's what you call all that, I
> may not be using the correct terms)? Sounds like that would be tough!

No, the CNC computer is MUCH better at handling this. Yes, in theory it
could be
done. Remember that the very first NC systems had no computer, and did
all the
interpolation by hardwired logic. but, why would you want to do that?

Having a single CPU handling ALL the axes allows them to stay
coordinated
at all times.

If the CPU is updating the velocity info at 1000 Hz, everything can be
VERY smooth,
without doing any acceleration/deceleration profiles in the hardware.

Jon