SMT techniques (was Re: group buy for homebrew CPUs?)

Dave McGuire mcguire at
Sat Aug 12 12:00:10 CDT 2006

Ethan Dicks wrote:
> I _have_ thought about that for replacement -8/L and -8/i boards -
> design a simple board with, say, 4 to 6 16-pin SO pads and decoupling
> caps and a dual row of jumper pads to cross-connect the edge fingers
> to the various chip pads to be able to replicate, say, an M111 vs an
> M117.  The "problem" is that I doubt you could get real TTL (not
> LS-TTL) parts in SO.  I haven't experimented with replacing TTL with
> LS parts in any of my -8s, so I don't know if there would be any
> issues or not.  So far, I've always had enough of the right things on
> hand when I go to replace a chip on an M-series module.  I would think
> that the dimensions could be on the order of 2 x the size of just the
> area of the fingers.
> The advantage, of course, is that it would be easy to make a large
> number of these on one PCB panel, reducing per-unit costs.  Using one
> (or maybe two at most) base designs would also help quantity issues.
> You could sit down one day and make a stack of a dozen M111s, then
> make two dozen M113s the next day with the same PCB; just add the
> right SO-parts and configure the jumpers in the right order, and there
> you are.

   A variation on this theme might be to build a very small number of 
different types of boards, but with a CPLD on each one...a small, cheap 
one like a Xilinx XC9536 (something like $3/ea qty. 1)...then have a 
JTAG connector on each board, and squirt the "personality" into the 
board that way.  I've not had a chance to work closely with 
FlipChip-based machines (unfortunately) so I don't know how much 
consistency there is on things like power pins and such, so I don't know 
if this'd be practical, but I think it's an interesting idea.


Dave McGuire
Cape Coral, FL

More information about the cctalk mailing list