Re: [openppc] Routing PPC mobos

[Alasdair Ferro <aferro@lsil.com>]

As this was sent to this list, here is my reply - apologies if you didn't want it :-) Feel free to
comment.....


> As I am reading what you wrote, some questions arise. G3 and G4 would then
> be absolutely impossible to route on a PCB by hand ? What is then the
> meaning of OpenPPC ? Nothing more than a dream ?

See Tom's previous answer :-)


> And about what you said (133MHz tracks, crosstalk, ...) : I understand what
> you are talking about, but is there somewhere I can find this ?

I'm not aware of somewhere myself, but try somewhere like epanorama.net - it's a very good links
site, or a search on Google.


> Speaking of the PCB routing tools, maybe I should try to find out whether my

> school could get some for educationnal purposes (CAD soft vendors usually
> offer such options to show how powerful their products are and hope that
> future engoneers will choose 'em...).

You can get trial versions of OrCad, go to their web site. Check out other vendors.... The only
thing is they tend to be limited in some way (try to make you by the software!!)


> This brings me to my last question. Nobody really knows how expensive
> producing a proto PCB is... Should I understand that only big companies have
> tried the OpenPPC concept ??

The problem is, as you say, nobody knows, cos nobody's tried :-) Judging by PC mobos, you're looking
at 4 layers, and some serious signal integrity issues. This is not easy - have you tried complex 1
layer bopards - they're not nice by hand, or with tools. To do the sort of layout required does need
a good knowledge of PCB design, along with the right tools, and support, hence only large companies
can try it.


> Thought it was something like the Linux
> community...

The problem comes down to the cost. To produce good software you need a computer, a text editor and
a compiler. The last two are free, and the first, well (nearly) anything will do :-) To produce
hardware, you need a nice (decent PC / workstation) computer, some expensive software, and some
expensive hardware.... Until we can resolve these issues, open hardware as a concept has some major
concerns. That said, there is an effort to produce GPL'd EDA tools (do a search on freashmeat.net
for EDA, or similar), but the cost of maufacture, be it PCBs (for anything above 2 layer) or silicon
is still high.

That said, however, there is still a lot that can be done. What can be fitted on an FPGA is
increasing in leaps and bounds, and home-PCB manufacture is becoming more popular. There are also
companies who will maufacture PCBs for one, and populate them, but it does cost, and there aren't
many of them :-(

Sorry to be so negative, but sadly that's the way it is at the moment. If you are interested, keep
on the mailing list, and follow developments - I'm hopeful :-)

Alasdair

--
To unsubscribe from openppc mailing list please visit http://www.opencores.org/mailinglists.shtml