f21 pcb or computer at UT

[jfox@xxxxxxxxxxxxxxxxxxx (Jeff Fox)]

Dear MISC readers:

Is there anyone in the list with experience at and ability to
get PCB for testing F21? 

Three people volunteered to make boards.  However they didn't want to
design test boards.  They always did their boards the same way and
they knew the application that they wanted.  Their boards would
work if all the things on the chip worked and all the firmware
for the F21 and all the peripheral PIC chips all worked together
perfectly the first time.  Unless everything was perfect and you
made no mistakes in the design or firmware programming the boards
would work the first time.  They would be useless for
incremental testing step by step which had never been done.

I was also concerned that not only would none of these designs for
finished application level products work the first time on untested
chips, but even if they did they would limit testing to the tiniest
percentage of what needed to be tested.

I could not understand why people would want to tie up or 10 multi-
megahertz general purpose I/O lines to talk to a PIC to do 9600 baud.
I could not understand why no one wanted to test any of the F21's
capability but rather design products to use 10% of it that could
only work if everything was perfect.  I just assumed that they felt
it would be less work if we just skipped the testing phase and
went right into their application.

I did however eventually just say, ok, ignore the stuff I have 
published about what I want in a test board, ignore the diagrams
that I posted and the outlines in html and pdf and do the board
you want.  (when it doesn't work I can still cut the traces to
all the application circuitry on the board and test the chips.)
But I never heard back from anyone about any progress on their
boards.

Chuck says that wire wrapping is about a 1 day job.  All you need
is some sockets, some components, memories, wire, tools, etc.
Sounds simple except when you are waiting to sell the first chip
to be able to buy a meal.

It appears that I will soon have sold a chip and can now afford to
have a pcb done.  My idea of a test board is small, simple and
good for testing.  Small size, short traces, clean, and jumpers
for optional stuff and headers.  I layed one out and stared wire
wrapping but ran out of wire, solder, gas money, food, phone etc.

I liked the idea of a small clean board with a P21 socket inside
of the surface mount f21 socket so either chip could be used.  I
wanted it two sided, surface mount and simple.

some people wanted a 1Mx32 simm socket to use 1Mx20 the way Dr. Ting
did.  some people wanted sockets or pad for 5x 1Mx4.  some wanted
2x 1Mx16, or a 1Mx16 and a 1Mx4 with matching page sizes and speeds.

I should now be able to contribute money towards that project. Who
in this list is familiar with such things.  I was programming manager
at iTV not board designer.  there are plenty of examples published
of the information needed to lay out a board and it might be good
to let other people review the design for errors or whatever before
sending it out to produce pcb.

Minimally we need a 2 layer 1.5" x 2" board.  That might be a little
too minimalistic.  I don't like going to 4x6 however.  

I do see a new problem today.  I thought we might lose the last 
link to the outside world for the last couple of weeks, the ISP
account, web site and email.  But the next best thing will be
happening.  I think tommorow we lose the computer on this end.
Maybe I should buy a computer instead of having pcb designed.
What do people think can be done?  It is a tough call.

Jeff Fox
-------------