Sunday, February 20, 2011

Shift/Jump unit control board in progress

I've started working on the shift/jump unit control board.  Today I implemented the MODSEL/!MODSEL logic and the LR selectors.  For reference, here's the schematic:


Back side of board -- point-to-point wiring being used in this case.

Front of the board -- the sharpie labels will get replaced with proper printed ones when finished.

Sunday, February 13, 2011

Interface

I went ahead and soldered the headers onto the perfboard that the control bits will end up on...

 

shift/jump unit backplane PCB is finished

All done: the shift/jump unit backplane is now complete.  I think it turned out pretty darn good, all things considered.

Solder-side view.  The gaggle of wires at the bottom is the
B register bus.  Since the pinout of the 40-pin socket is
such that the rightmost pin must go to the leftmost breakout,
there was no way to run the traces without using vias, which
I am not able to do.  So a little point-to-point soldering finishes it off.

Component side

I got these plastic snap-in feet from Jameco.  Must simpler than trying
to scrounge up brass standoffs and screws.

The worst section of etching on the board.  Surprisingly I only had to
re-route two of the traces.  Also notice the gummy stuff between the
pins -- I still haven't perfected my flux application, and I tend to use
too much, resulting in a gummy mess between all the pins.

I'm very happy with how the silkscreen came out.  It's surprisingly
legible, and I'm shocked it lined up as well as it does, given
that I had to align it pretty much blind.
The next steps will be building the four data transfer modules (each board will just be a pair of 74245 tristate buffers on an etched PCB), the control module (perfboard with point-to-point connections), and an A/B register bus simulator.

The A/B register bus simulator will just be a board with 32 DIP switches (16 for bus A and 16 for bus B) and a little control logic to set the outputs to hi-Z when the write signal is not asserted.  This will let me test whether the board actually works, before having actually constructed a working register file.