I’ve had this machine on the shelf for quite a while. Why on the shelf? because it didn’t work.
I’ve not really documented this one but it’s been so long since I posted anything here, that I thought anything was better than nothing 🙂
The first problem was no output at all. This was down to a mod by a previous owner who had put a switch in the output. Easy fix.
Next looked like a stuck bit in the video memory. This one took a bit of tracking down. It turned out to be a broken pin. A quick fix with a piece of wire and we are back on the road.
Sorry, a bit of a terse post but I’m back in the room.
Way back in the mid 1980s I worked for a small software house that used NCR hardware. Their work ranged from mini systems to micro based stuff and before the World went completely PC focused, NCR made the Decision Mate V, my employer bought them and I rather liked them.
The machine in the photo above is a recent addition to the collection and scratches an itch I’ve had for some time. As you can see it has a half hight 5.25″ floppy and a Winchester (size currently unknown).
This is a Z80 machine running CP/M-80 as with lots of machines of the day. Is slightly unusual in that the expansion slots are accessible from the back of the machine without removing any covers.
When I bought this machine I was warned that some capacitors had been removed from the power supply. These are 10,000uf, 35V electrolytics and easy to find. I’ve fitted three new ones but sadly this machine is not running yet. The 12V line is dead at the moment and the diagnostic LEDs at the back are saying RAM fault. so some investigation is needed.
Part 2. Looking for clues.
Looking at the schematic of the power supply it looked logical that Q1 had failed. It’s a big transistor that looks to be doing the heavy lifting. Whenever I work on old machines from the late ’70’s onwards I am always impressed by the fact that components’ fittings and shapes haven’t changed much. A quick trip to eBay and a new BDV64A Darlington transistor was in my hands. Sadly it didn’t help.
Sadly it didn’t help
That should have been it really but no. Disconnecting the PSU from the motherboard however did help. The PSU came up, the fan started spinning and the 12v line was good and healthy.
Capacitors! I remember hearing once that electrolytic capacitors tend to fail open circuit and tantalum capacitors tend to fail closed circuit. The theory now was that a tantalum cap had gone short and was crowbaring the PSU but the PSU was a high quality item and hadn’t blown up or blown the cap up.
That’s an easy one to check. The mother board only had three tantalum caps on it and sure enough the one across the 12v line had gone short. A quick replacement later and we have the fan running and the motherboard still complaining about a RAM fault.
The RAM is made up of eight 4164 DRAM chips. Sadly they aren’t socketed and so I decided to try and remove them. This is tricky and my attempts to do it elegantly failed. Without know which chip(s) it was (were) I took a punt and removed four by cutting the legs and desoldering the pins. I cleaned the holes and fitted four turned pin sockets and four new 4164s.
Bingo.
RAM fault gone. It seems I got lucky.
Now it’s complaining that it has a keyboard error. Understandable as I haven’t got a keyboard yet!
So as you will know, I have serviced and cleaned the various power supplies and PCBs that make up the PDP-8 clone.
It’s time to get it back together for the first time. I expect to strip the machine down again before too long when I’ve decided how I want to treat the rusty 19″ rack.
First of all, I cleaned out the card cage and vacuum cleaned the inside of the rack. I refitted the three power supplies and the fan tray and turned on the mains.
No smoke or fireworks and the fans came on and span quietly. Result.
The three power supplies after refitting in the rack. Note the power lights are on.
Next, I fitted the regulator boards in the right hand side of the cage as we are looking at it and powered on again.
Still no smoke.
The card cage after its scrub up.I have made no attempt to remove any rust yet. Just removed the dirt and grime.
Finally, I refitted the rest of the boards and turned on again.
Still no smoke.
I was hoping to see some activity on the front panel but there was very little. The LED that indicated the “Fetch” cycle was lit but that was about it. The buttons had no effect and the 7 segment LED displays didn’t light.
The front panel showing the illuminated “Fetch” LED
I am going to have to start to understand how PDP-8s work now to find out why mine does not.
The CCI has a lot of cards in it’s card cage and the power supplies are massive and so, there is going to be heat. To get the heat out of the way there is a large fan tray mounted just beneath the card cage.
Bottom view of the fan tray.
It runs from the front of the machine all the way to the back and draws cool air from the bottom of the rack and forces it up through the cards an out.
As with everything in this machine it was very dusty and needed a good clean but that’s all it needed. After 20 minutes with some foam cleaner, a brush and a cloth it was looking pretty good.
There is some corrosion on the blades but nothing to worry about.
Top view of the fan tray.
I fitted a modern UK plug and fired it up.
Tadaa. It works beautifully. It’s quite (for an 80watt fan tray) and the fans run smoothly.
On this machine, the PSUs rectify the power but don’t regulate it. The regulation is done by a number of boards at the right hand end of the card cage.
As before, I’ll just show the photos here and perhaps give more details, if needed in later posts.
-5v regulator board. Top view.5v 20A regulator board. Top view.Regulator board 3. Bottom view.
Having been through the CPU and memory boards it’s time to have a look at the next batch of boards which are the ones that look after the peripherals.
There isn’t much to say about these at the moment as I’m just in the process of cleaning them up and looking for damage.
I’ll just throw the photos of them in here and perhaps in later posts I’ll go into more detail.
Top view of the Comms controllerLine printer interface. Top view.Teleprinter interface. Top view.Remote VCU DMA interface. Top view.VCU terminal control. Top view.Universal MTT controller board. Top view.Terminator. Real time clock board. Top view.
Having taken a look at and a brush to, the first few boards, I have come to the memory boards. This machine has 32kwords of 13 bit core memory in two memory modules.
There is a memory bus controller board but I didn’t take a photo of that. I did get photos of the memory modules themselves though.
Top view of the 16kw x 13 bit core memory module.
The memory modules were made by Ampex and have a customer part number containing the initials CCI. This must mean the boards were made specifically for CCI and were not off-the-shelf-items just bought in.
Looking from either side of the module it is not overly obvious that there is any core memory at all. It’s well protected between the two PCBs that make up the module.
Bottom view of the 16kw x 13 bit core memory module.
Identical in every way?
When I was cleaning the first of the modules I noticed a broken resistor. I struggled to read the value bands so I took a look at the other module, one I hadn’t cleaned at all, and it had the same broken resistor!
Two view of the upper memory board showing the broken resistor
I had a word on the VCFED forums and the view of the team was that the resistors were 510 Ohms. A quick raid of the parts bags and two new resistors were fitted.
Top view of the upper memory board showing the replacement resistor.
Having to part the two boards that make up each memory module gave me the opportunity to take a look at the core memory itself.
The core array is behind a steel plate and a sticker warning me that I was about the void the warranty.
View of the core memory array.
I was very careful when removing the metal plate and I’m glad that I was. This is the smallest core I have see I think. Sadly I don’t have a standard banana for scale but trust me, the cores are tiny.
Having worked my way through the power supplies, it’s time to take a look at the circuit boards in the card cage.
The original PDP-8 had a number of “Flip-chips” that held basic logic circuitry made from transistors. These boards were plugged into a backplane that routed the signals between the appropriate gates on the flip-chips. Later PDP-8s had the processor built up of 3 large PCBs and the later PDP-8 compatibles from DEC, such as the DECMate had a single chip with the processor.
The CCI machine is none of those. Instead it has six boards that make up its CPU. Each is clearly marked with its function, “Major registers”, “Accumulator” etc.
Here, I’m going to describe working my way through these boards, visually inspecting them and giving them a clean.
Please keep in mind the size of these boards. They are huge by Today’s standard. 40cm x 25.5cm.
Front panel / operator console.
Standing on the front of the machine, the front panel contains the switches and LEDs needed to load simple programs into the computer. PDP-8s of this vontage didn’t have any ROM code so do nothing when first powered up. By using the toggle switches and push buttons, it is possible to load and run simple programs.
This machine is a bit unusual in that it uses 7-segment LEDs to show the register contents rather than the more usual one-LED-per-bit arrangement of genuine PDPs. It also uses push buttons instead of spring loaded toggle switches.
On the right you will see the board after I have removed the red tinted acrylic sheet. There is nothing much wrong there, it just needs a good dust.
Com Seq Gen board
The second board in the cage is labelled “Com Seq Gen”. I’m assuming this mean command sequence generator and I’ll continue to believe that until other evidence comes along. It is clearly labelled Sep 01 1976. Most of the other boards are either 1977 or 1978 with chips with dates anywhere in between.
Interestingly, this is the only board in the first six to have any lacquer on it and it’s only on one side.
Just needs a good clean.
Sequence Input Generator
Another sequence generator. This on is labelled “7807” so I’m thinking July 1977. By now CCI must have given up on lacquering their board as this one is raw fibreglass.
Major Registers
I love the idea that all of the parts that make up the CPU are spread around a number of boards rather than under a plastic or ceramic lid.
Extended Memory Control
The basic PDP-8 architecture can only address 4k words of memory. This machine has 32k as we’ll see later. The original PDP-8 got around this limitation with a K8ME board and I’m assuming that this povides the same functionality.
A small about of patching on this board in red wire (centre left).
Accumulator
Accumulator board top view
Here is the brains of the outfit. The accumulator. Nice isn’t it.
Memory bus terminator
I think this is the last board of the CPU. Alternatively, it’s the first board of the not the CPU.
I wonder if there are enough cards on this page to prove that the CPU is alive?
So we are two power supplies down, one to go. This one is a bit of a monster with three main voltages and two extras.
Top down view of PSU 3 before cleaning.
As with the previous PSUs, this one transforms the mains into lower voltages, rectifies them into DC and smooths them out but like the others it does not regulate the voltages. That’s done by a number of regulator boards at the end of the main card cage.
As with the other PSUs, this one is also filthy and with questionable capacitors.
PSU 3 chassis split into three pieces before being cleaned.
My first job was to split the two halves and see what we are dealing with. As you can see it’s a mucky mess with dirt everywhere and a lot of rust on the capacitor clips.
The capacitors were removed and tested and just like the other supplies, they were fine. They don’t make them like that any more.
The rear of PSU 3 showing the PCBs and fuses.
The three PCBs were removed and cleaned up. I took a wire brush to the fuse holders and blade connectors, cleaned the fuses and gave the boards a general spruce up.
One or two of the spade connectors had perished and failed when I was removing them so I made some new cables with new spades and replaced some of the other, dodgy connectors.
I replaces a lot of the nuts and washers as they had corroded quite badly. I didn’t replace the capacitor clips as I want to keep as many original bits as possible but I gave them a good sanding to get the rust off before re-assembling the whole thing.
It all went back together and is looking good. The voltages seem OK too!.
Here is the finished unit. You can see the capacitors have been cleaned and refitted. As have the capacitor clips and the cables.
The second PSU is the same model as the first and I went about its refurbishment in the same way.
Firstly, I removed the capacitors. Here was the first problem. One of the smaller capacitors was leaking. Straight over to RS components to see if I could find a replacement. Bingo, next day delivery! Amazingly I found a capacitor of the same diameter with the screw terminals on the same centres with the same screw thread. Marvellous.
The faulty capacitor from the second CCI power supply. The new capacitor is on the right.
I tested the other caps with my capacitance meter and they all looked feasible. I then brought the up, one by one on my bench suply with the current fixed at < 20ma. As each one stabilised at it correct working voltage, I noted the leakage current and this was typically around 0.6-0.8ma. That’s not bad at all for 40 year old caps.
Next the whole unit got a good dust, the terminals and fuses were cleaned with a wire brush and everything was reassembled prior to testing.
The second CCI power supply after refurbishment. The new capacitor is on the left.
On of the wires running from a rectifyer was showing signes of bad corrosion so I made a new cable up with new connectors and fitted that.
I went through the same tests with car tail-light bulbs as I had for PSU1 and they lit up beautifully.