PDP-8/A repairs at VCF workshop Sept 2019

Summary

Sept 28-29 2019: I went to a Vintage Computing Federation (VCF) MidAtlantic repair workshop, at InfoAge in Wall NJ where their museum is located. I brought my PDP-8/A system and various Programmers Panels I've acquired. I wanted to test the panels and get familiar again with the 8/A. This Web page references the workshop, some other items I worked on, and links to my 8/A work.

Several people were in attendence, to do their own repairs of vintage computers. Evan Koblentz worked on a TC Robotics Lego and Apple II interface. Also the VCF-MidAt museum was open, repairs of their Univac was in progress. I bought some test equipment at a local hamfest in Tinton Falls (Garden State Am Rad Assoc); it all checked out. An Omnibyte brand 68K system I obtained was passed along; it was restored to operation in the course of the weekend. Some details are below.

The photo to the left, by Andy Diller, shows me in red shirt at the table. My 8/A is still in the white box in front of me. The youngster is a participant's son.

Sunday breakfast with my colleagues at a local Jersey Shore diner, OB Diner in Point Pleasant. Also: I visited the Infoage Space Exploration Center and heard a Sunday talk by Frank O'Brian (NASA Solar System Ambassador) about the moons of Jupiter and Saturn.

PDP-8/A photos and commentary

Details of this 8/A's repairs in 2016 are on this linked web page. The 8K DEC core memory G8010 controller in this machine was serial number ending in "1613", which showed some bit errors on the least-sig bit previously.

8a chassis
8a Programmers Panel PC board only
cabling to the Panel
a complete 8/A Programmers Panel - checked OK
a second 8/A panel also OK
Here's a running display.

So my 8/A worked for most of the day. I used a Variac to bring up the AC voltage, as it had not run for a few years. I ran it at low AC voltage (maybe 80VAC?) because at higher AC voltage it shut down (some DC power fault?). After trying to run through the power supply to check voltages, I read the 8/A documents. I forgot that to run the 8/A, the "run" switch on the small operator's panel had to be "on". With that done, the system seemed to read and write the DEC 8K core memory OK; by operations of the Programmers Panel.

DC voltages at 71V AC in at DC regulator fuses and breaker: -17.2, +16.2, +29, -4.98, +7.47. Resistor R20 is 20V; DC at programmers panel is 4.8V. DC voltages at 110V AC were not that much different.

But sometime in the afternoon, after power cycling the 8/A a number of times, it powered up but was not able to load any values in memory. The system acted as though there was no memory-board in place. DC voltages at the power supply seemed to be within reason as above at 110VAC; I tweaked the +5 adjustment for 4.8V to 4.95V. No change. Swapping in either of the complete (but untested) Programmers Panels, did not change things. But other than write, I was able to validate those two Programmers Panels.

I spent some time Saturday evening, reading the 8/A manuals and schematics of the 8K core. But I did not have a spare core board, or an extender card to allow access to the parts of the core board. The manual suggested that for "all zeros" a voltage-reference signal for the read/write amps (output of Q27 transistor) may have failed in some way. Checking that will be my next bit of work on this system.

Follow up

I asked David Gesswein about my core-memory failures:

Herb: I'm contacting you otherwise, because one of my 8/A DEC 8K core boards failed this weekend. Simply stopped responding - all 0000's like the board isn't on the bus. Worked earlier in the day for manual entry, and I ran the 8/A at reduced AC voltage because I was an idiot. Details are on the WEb page, but I doubt running the 8/A at 70V AC or later 100V AC caused any failures. I didn't run anything but little toggle programs, mostly I wanted to test some programmers panels and get the 8/A "hot" again. Even with no memory, the "logic" from the programmer's front panel seemed to work, of course I could not test instructions.

I have other core-boards and fairly soon (this fall) I'll work on the 8/A again with them. But you do have some advice from experience, given these symptoms for this now-dead 8K core memory board? - Herb

David: There is a power good signal from the power supply that will disable the core if its not active. I would check that first. Power OK on backplane pin B2V. I haven't had to do any repair on 8/A core so can't give more specific help. I think I will be going to the next repair workshop if you need help. - David

Omnibyte system

Omnibyte brand OB68K/SYS system. A 68000 based Multibus system with Shugart 851 8-inch drive and Quantum 2020 8-inch hard drive. See the Multibus cage? also the SMS floppy-hard controller board This was acquired recently, and passed along to someone at the event. They powered it up and eventually figured out how to opearate IDRIS, a tiny OS flavor of Unix from Whitesmiths (a software company of the early 80's). The single-board 68000 card has 512K bytes of RAM on it, MACSBUG in ROM, serial ports, and most of the other things needed for the system. A companion I/O card has date-clock and an interface to the SMS drive controller, the large board under the lid. The controller runs both the floppy drive and the hard drive - same interface. Surprisingly, the hard drive was "locked down" for transport! With some futzing, the OS came up. Here's some Web search results on MACSBUG and on IDRIS.

VCF-MidAt Museum: Univac

Repairs of the Univac minicomputer at the Museum

Lego robotics 1986

Evan Koblentz worked on a TC Robotics Lego 9750 Technic Control Center motor-controller and Apple II interface. He has a slide show on the kansasfest Web site.

Herb Johnson
New Jersey, USA
to email me @ follow this for email and business