• PDP-8/M, 8KW core, M8655 serial UART, very clean machine in working order
• PDP-8/E, 4KW core, M8650 UART, requires a full rebuild
• PDP-8/A, 16KW core, extended memory, serial, paper tape drive and controller (non DEC). Very clean and in working order
• RX8E + RX01 drives
• PC8E + PC04 paper tape
• DECmate III
• WP78 (aka a VT78)

Email offers to toby@pdp11.co.uk . Machines are located in UK

The main sticking point with getting this machine back was a failed +5V and +15V rail on the PSU. Being no expert on PSUs and having been mildly ‘warning’ zapped by it when I decided to see if anything obvious was failed, I had a hunt for a company to perform the repair.

I opted for Electronic Repair Centre Limited after some internet searches. I was very impressed with their service, which was fast and very well priced. I thoroughly recommend them to anyone looking to get a power supply repaired.

After getting the PSU back, I reassembled the machine and with a bit of CPU board swapping (3 of the originals turned out to be failed), I now have a mostly working machine. The one gotcha I currently have is that on entering the DEC accumulator counting test, the accumulator counts at full bore and cannot be seen neatly increasing on the console. My suspicion is that the line time clock might be the cause of this.

The investigation continues…

[Show as slideshow]

The /L is complete with the (somewhat important) exception of its core stack which has been lost at some point. If anyone has or knows the whereabouts of a spare stack (I believe they were common to the /I and PDP-12), please let me know. If I can’t find one I will have to resort to fabricating an alternative using NVRAM chips,although I doubt this will be trivial.

The /E is generally clean although its plexy is pretty scrappy. I quite like the dog eared look and so I will leave it this way. The console itself is missing a couple of keys although I have replacements for these. The machine came with no cards, but I have spare CPU sets and so I’m only left needing to find memory which should be achievable.

Completed

1. Perform a general appraisal of machine
2. Photograph machine in detail
3. Remove console and cards
4. Remove power supply
5. Remove wiring looms and any other remaining items attached to chassis
6. Remove backplane
7. Clean and sand chassis
8. Spray chassis to primer
9. Rub down primer
10. Respray chassis (to matt black)
11. Clean backplane
12. Clean wiring loom and other chassis furniture
13. Take detailed pictures of power supply
14. Dismantle PSU
15. Clean power supply components
16. Remove and reform capacitors
17. Reassemble power supply
18. Bench test power supply and confirm operation
19. Reinstall power supply (but do not reconnect AC or DC at this point)
20. Reinstall backplane
21. Reinstall chassis furniture
22. Power on and confirm fans are operational
23. Reconnect AC to power supply and confirm operation of PSU
24. Reconnect DC from power supply to backplane and confirm voltages on backplane
25. Install minimal card set (console and CPU) and test console operation

To Do

1. Add further earth straps – they appeared to be non existent when I got the machine and I’m far from convinced the chassis is earthed at the moment
2. Tidy up DC wiring loom
3. Configure and fit core memory
4. Debug console to establish why address bit 5 isn’t working
5. Confirm correct operation of the console
6. Enter basic test programs and confirm operation
7. Solve problem of some console keys catching on console plexi
8. Repair 3 console keys that having broken pivot shafts
9. Fit console and confirm operation
10. Fit RX8E and confirm its working

The minimum board set was then installed consisting of:

• Slot2: M8330 KK8-E timings
• Slot3: M8300 KK8-E major registers
• Slot4: M8310 KK8-E major register control
• Slot 20: M8320 KK8-E bus loads

The machine was then powered on and LEDs on the console sprung into life. Some basic operation from the console was possible (I was able to set the PC) although bit 5 of the PC appears to be stuck off. I will have to investigate whether this is a fault on the console or not.

I also installed an 8KW core memory set and retested with this in place. With these boards in place, the PC was all bits high and could not be changed.

So still some way to go! The next step will be to establish if the KK8-E is fully working and track down the issue that is causing the bit 4 of the address bus to appear stuck low on the console.

[Show as slideshow]

To rebuild, I stripped the supply down and gave it a gentle clean with electrical cleaning fluid and a clean cloth. A paint brush was used to dislodge any dust in hard to reach places. The electrolytic capacitors were then removed and reformed.

The power supply was then reassembled and plugged back into its mains connector in the chassis. Power was then applied for short periods of 1 second, 5 seconds, 10 seconds, 30 seconds, 1 minute, 5 minutes, 10 minutes.

After each application of power, I carefully touched the cases of the capacitors to confirm there was no build up of heat. Care should be taken as these beasts hold a lot of charge for a long time, believe me you really don’t want to find out how much it hurts to have one discharge through you.

Once I reached 1 minute, I also tested the +5V, +15V and -15V DC ouptuts to confirm all was well. Luckily for me it was.

The final step was to hook up an oscilloscope and to check each of the power rails to check that they were stable and to confirm all of the control signals were working correctly. The line time clock (LTC-L) is the only signal of note when checking with an scope as this should give a 50HZ (or 60HZ in US) square wave.

Having passed all checks, the power supply was reinstalled into the chassis and connected back upto the OMNIBUS backplane. The consists of 4 connections, -15V, +5V, +15V and the PWR OK L signal (spade marked DV2)

[Show as slideshow]

Useful Links:

• H740 Schematics (PDF)
• Article on reforming capacitors

Unfortunately I could not persuade the build label to come off of the chassis cleanly so this had to be sacraficed. Before I removed it I carefully took a scan of it by placing the chassis on a flat bed scanner. This will allow me to replace the label with a replica at the end of the rebuild.

The chassis was resprayed in 3 coats of grey primer and then 4 coats of matt black paint. I’m really happy with the results, although in some areas I should have spent longer sanding to get a really perfect finish.

[Show as slideshow]

An appraisel of the machine

Before commencing, I will carry out a thorough appraisel of the machine so we can get a measure of the job in hand:

Front View

The Backplane

Inside, the box is mostly clean, a few rust blisters but nothing too worrying. The backplane looks in good shape.

---

The build label

The build label reveals that the machine was once fitted with:

• KK8E Processor
• KL8E Serial Console
• 3 x MM8E core sets (giving 12KW of memory)

When the machine arrived it had no cards. I’ve since aquirred a KK8E, KL8E and RX8E. For memory I plan to use one of the MM8E sets from my other 8/M

Power Supply

The power supply arrived loose in the box. It isn’t too bad, although has received some somewhat rough patches during its life. One area of concern is the two missing power transistors at the front of the supply.

1. Power supply strip and rebuild
2. Clean and respray of chassis
3. Clean and repair of console
4. Reassembly of chassis/console/PSU and backplane
5. Commissioning of power supply
6. Installation and test of KK8E and MM8E
7. Installation of RX8E and connection to RX02 drive

The first machine is complete and pretty well spec-ed having an RX8 controller in it and 24K of core memory. The second machine is basically an empty chassis which is in pretty good shape bar a few missing console switches.

My intention originally was to keep the empty chassis as a parts bin for the more complete unit, however I have since managed to locate a full board set and assuming I can get them for an affordable price I will ultimately return this second unit to service.

Restoration of the more complete M will start early in the new year once I have finished off my current project of cataloguing my collection.

Here are some pictures of the machines:

Machine 1

[Show as slideshow]

Machine 2

[Show as slideshow]