PCI card

[Mark Pictor <mpictor-at-yahoo.com>]

I'm new here, thought I'd post now that I found some
info that should be useful.

In January there were a number of postings about
better ways to interface EMC with the hardware, to
avoid problems with things becoming obsolete (i.e. the
parallel port).

Someone suggested a free design be developed, maybe a
PCI card.

I was poking around the 'net for free PCI interface
designs and found something that could be very useful
at http://www.opencores.org/projects/pci - this is a
VHDL core to put in an FPGA, along with whatever else
you want, which will then work as a PCI card.  This
looks to be the way to go, though it would still
require someone familiar with the VHDL language.

FPGA? core? VHDL? An FPGA (Field Programmable Gate
Array) is something like a computer: you can change
how it behaves by re-programming it, and you can plug
things in for features that programming alone won't
give you.  The core is the program that goes in it. 
VHDL is the most common language used to make the
core.

Disadvantages: Very hard to solder together by hand
(fine lead pitch on FPGA: 208 pins on 1-inch-square
package); circuit board would be difficult to
home-brew due to fine pitch; 32-bit/33MHz PCI will
become obsolete eventually.

Advantages: If the hardware was done right, the
difficult part of the design (the PCI card) could be
reused with many, many different feature variations:
logic changes are in firmware, hardware changes (such
as more A/Ds and D/As) could be on a breakout board,
external to computer, with much coarser, and etchable,
traces. When 32-bit/33-MHz PCI (and PCI itself) become
obsolete, only part of the design would need changed -
the card and the PCI interface logic - any
EMC-specific logic programming would stay intact. 
Re-programming should be trivial, and there is no
limit to how many times it may be done, a cheap memory
chip could wear out but it's easily replaced. A large
enough FPGA could be used so that some features could
be migrated to hardware, such as timing and
interpolation, so that the system could be less
dependent on real-time software (for example).

I have a bunch of ideas on features as well, I will
post after I see what you think of this.  I have
little experience with VHDL, etc. so I can't really
help out too much.  I assume Jon knows about VHDL from
his servo controller, maybe someone else does too.

Mark

__________________________________________________
Do you Yahoo!?
Yahoo! Shopping - Send Flowers for Valentine's Day
http://shopping.yahoo.com