It lives! Imlac PDS-1, serial #2 is running again.

From: Bob Shannon <bshannon_at_tiac.net>
Date: Wed Sep 4 00:47:00 2002

Hey,

After a small battle, and several hundred hand-soldered decoupling caps
(the original engineers figured they didnt need any), and a few dead
bugs, I've gotten an Imlac PDS-1 running once again.

The original Imlac hardware design divided the main CPU and display
processor up onto about 65 small cards with a handful of TTL chips on
each. No ground planes at all, and the 'decoupling' consists of a 15 uf
bulk cap on each board along with a ferite bead inductor. No high speed
decoupling caps, with the exception of 3 chips on the clock generator board.

With a few decades of age on the linear power supply filter caps, the
logic supply noise was amazing, often over 4 volts peak to peak on the
+5 rails. Small wonder the machine would crash and hang, its kind of
amazing it ever ran this way at all.

So a good deal of the work I had to do to get this machine running again
was to hand-solder several hundred decoupling caps, one for each chip in
the machine, more or less. I could have re-capped the main supply, but
given
the inductance between that filter cap and all those un-decoupled TTL
chips, adding the decoupling caps at the end of those long power supply
leads made more sense. After all, the nose at the power supply filter
caps was nearly zero, while the ground bounch on each small logic board
was quite large. Without a ground plane, the TTL chips had to seek a
common ground through long thin runs of etch, so I added the caps as
close to the power and ground pins of each package.

While this is hardly original to the machine's original design, I know
of no other practial solution to getting this rare old machine running
properly, so out came the soldering iron.

I still have a massive ammount of cosmetic restoration to do, but
getting to see it run its "HELLO" test program with green
vector-generated characters was well worth a few days with the iron and
scope in hand.

By the way, during my debugging, I came across a bad 7473. Its a odd
little JK flip flop, and I'd like to find a replacement. Note that the
74LS73A is NOT compatible, its got a different truth table! Should
anyone have one of these kicking around, please let me know.

And now, its time to sleep....
Received on Wed Sep 04 2002 - 00:47:00 BST