Re: [CAD_CAM_EDM_DRO] Wish list

Eric Keller wrote:

>
> Jon,
> in reality this approach makes the most sense, but i don't like it
> anyway:).
> The interesting thing to think about is that Plxtech has a board where
> you
> take your isa board and plug it in to their pci board in preparation
> for
> porting the isa board over to pci. So any work you do on the isa
> board
> would not go down the tubes if you could use this approach.

I'm aware of this.

> I have come to like the idea of replacing the us digital chips with
> pics,
> but i'm not sure how many axes one could read reliably with a pic.
> are you
> planning an 8 axis board?

You can't do this. The PICs will be sampling the encoder lines, and
also trying to
listen to requests from the computer, which they have to reply to in a
couple hundred
nano-seconds. There will be lost encoder counts unless the encoder
update rate is
very low, or there a PIC interrupt on a change in encoder signals. I
just don't see
the advantage of replacing a cheap hardware solution that is totally
deterministic
with a sampled solution that is non-deterministic, and obviously prone
to
synchronization problems. My personal view is to replace all 8 encoder
counters
with one Xilinx chip, with all the necessary logic in it, and solve a
number of
problems with the US digital chip at the same time. (The way the index
pulse
is handled in the US digital 7166 is pretty poor.)

Jon