The pure *hell* of restoring a PDP-8/S desktop.

From: John B <dylanb_at_sympatico.ca>
Date: Mon Feb 7 16:23:45 2000

Two weeks ago I *had* restored it.... and about every half hour it would
break down with a different problem. I am posting this message in the hope
that if anyone else ever has to restore such a nasty transistor mini then
they can use this as a reference for *everything* that can go wrong with a
minicomputer (excluding the power supply).

First, the best methods I have found to eliminate the main problems in the
most stubborn transistor minicomputers:

1) Bad contacts on backplane - Spray it with "Engine Gunk" (seriously, an
automotive engine degreaser/cleaner), then stick a plain extender board with
cloth wrapped around contacts and insert/remove a couple of times. Then use
contact cleaner.

2) Weak Transistors - Lower the power supply voltage enough that poor
transistors start to fail.

3) Thermal Transistor Breakdowns - (these are the Fairchild transistors that
*always* fail due to the epoxy inside expanding and breaking contact). I
recommend leaving the minicomputer on for 5 hours and let the poor
transistors completely fail. Nothing worse than an intermittent one. BTW, I
had 18 Flip Chips filled with this problem :-( (thanks for the info
Allison!)

4) Intermittent Flip Chips - These occur in areas where the mini is running
hot (ie: The MB register) and cause poor solder contacts which *might* be
visible. I recommend pulling these altogether and replacing them. 100s of
ticking timebombs. Get rid of all of them before you go insane. Use a
digital thermometer and check for areas in excess of 115 degrees F (35 C).
In the last 3 transistor minicomputers I have restored over the past 5
months I have found you will get a lot more intermittents if the mini was
used *a lot* (ie: 12 years) and was stored in an area that varied greatly in
temperature.

5) Expect your prints not to contain a complete schematic of the mini
computer - I have found many changes to the straight-8, PDP-8/S that were
not listed in FCOs but instead new prints were issued. (1960's)

COMMON PDP-8/S Problems I have found:

Since the speed at which data is switched through the PDP-8/S is *much*
faster than the Straight-8 expect problems and replace/resolder the major
registers and *ALL* pulse amplifiers. Most pulse amplifiers are switching at
the maximum speed for such a Flip Chip (R602,603) so the slightest addition
of resistance caused by a poor solder contact will cause endless problems.


This PDP-8/S had the following problems (beyond the ones I mentioned before)

Saul Dinman decided to add a *new* interrupt called "LPC", Low Power
Condition. It will look like a negibus device interrupting the CPU but in
fact is a few gates wired to the power monitor board. Of course, it only
affects *PARTS* :-( of the CPU like... Parity! ,ION, and other nasty *bugs*.
As the original 8/S did not have this installed it was not included in
original prints :-(. I now found a third set with these mods... after
spending a couple of hours pulling my hair out. After polling for interrupts
and a memory parity error it is safe to conclude it is a low poer condition.
Thanks Saul!


Of course, everything else *only* failed when the CPU was running a
program..

 It's OLD instruction set was:
=========================================

ION when an external device draws *just* enough power when being accessed to
cause an LPC rather than an IO Interrupt :-(.

Clear Accumulator Sometimes - (external device control) when warm and only
when running a program and oscilloscope is not connected.

Move AC->MB , 99 out of 100 times asked to.

Stall CPU Clock 1mS - no longer or it might be seen without scope.

and my favorite...

If Zero, Branch Sometimes - BUT branch always in SI mode.

=========================================================================

The PDP-8/S has now been running for over 6 hours and every major "bit"
junction has been checked for a poor signal - all Fairchild's removed and
buried. The desktop 8/S is a micro-sauna which means I will probably install
fast boxer fans in place of the two eggbeaters currently installed in the
back. My rackmount 8/s has 4 fans along the sides of the boards + 2 blowers,
one above and one below.


I am installing a 20ma->RS232 converter and will get this mini connected to
Telnet with FOCAL loaded as quickly as I can. This mini needs a good month
of continuous running to make sure it is finally... happy.

PAPER TAPE: If anyone here is using a PC0 with a PC01 controller
(Straight-8, or 8/S) I found internal mods used at DEC to increase the read
speed from 300CPS to 650CPS... error free! 8/S *really* loads FOCAL fast
now!


I hope when I get a KA10 it isn't filled with the nasty problems this one
had.



john



PDP-8 and other rare mini computers

http://www.pdp8.com
Received on Mon Feb 07 2000 - 16:23:45 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:32:52 BST