A mug's eyeful.

There used to be a reason for vintage computer collectors to head to car boot sales. Tons of stuff going for pennies used to adorn the tables of the happy 'booters' (well, maybe not adorn but there used to be a fair bit). An Amiga A500+ for £3, a CD32 controller for a fiver, Commodore 64 (boxed) for a tenner etc etc.

Those days are long gone. Anyone who has anything remotely vintage either sticks it on eBay or heads to the car boots with ridiculous prices in mind. Things are so bad now that there are even Facebook groups dedicated to the delusional people of eBay and the whacky and bizarre prices that they think their stuff is worth. My favourite was a chap in Scotland who for several months put his boxed Amiga A500 on eBay for the sum of £1000 starting bid. If that had been serial number 1, signed by Jeff Porter and hand delivered by R J Mical, it might have been worth it. In reality, they're worth less than 1/10th that.

I digress. 

A few weeks ago, Crashed Jr(1) let me know of a 'PC' that was at a local fete for a very reasonable price. He sent me a picture and it turned out to be an Amstrad PCW9512. 

I know this! It's a CP/M system..

For those of you not in the UK, the Amstrad PCW9512 was one of a range of low cost PC like machines that dominated the early low-end small business computer market. I actually used one of these many, many years ago at an estate agents I worked at. It uses CP/M and has a primitive word processor and, originally, came with a printer too - no printer with this one but that's fine.

So, does it work? Well, the thing with these machines is that they are very much built to a price point so to actually get this thing up and running requires either the word processor disk or the CP/M disk. Switching on with no disk in the drive gives a beep and a white screen and nothing else. Without the disks, this is a lighter than you'd imagine paperweight.

And on the subject of disks... The more astute of you might have spotted that this thing actually uses 3 inch floppy disks rather than the more usual 3.5 inch disks. The story is that Alan Sugar managed to blag a boatload of 3 inch drives dirt cheap and so they went into every machine Amstrad generated that used floppies (CPC6128, PCW8256 etc etc). Not sure if there's any truth to that but anything's possible

Anyway, this unit did exactly as expected when I got it home, i.e. it has no disks so it just shows a white screen. I did have a 3inch disk for a PCW8256 but I didn't realise at the time they aren't compatible so when I put it in, nothing happened. But not exactly because of the incompatibility..

The floppy units in these machines use a drive belt to turn the spindle motor that makes the disk spin. And we all know what happens to rubber belts as they age. They turn into a thick, black goo that is horrible to remove from everything. To check the state of the belt I had to take the back of the machine off, then dismantle the base from the CRT so I could get the drive unit out. After taking the back off I realised I wasn't the first person to do this.. 

There are two controls on the side of the unit to control the brightness and contrast. The contrast knob had been hot glued. Oh dear. So as I took the case off, both of the potentiometers for the controls basically fell apart. Dangnabit. I managed to collect the bits and then desolder the bar that they were both attached to. 

Broken. Darnit.

After a bit of head scratching I realised that all that had happened was that the protruding control knobs were held on by plastic stakes that were melted to hold the front cover plates on. By gluing the plates back on - and trying to avoid sticking the knobs too - I could probably get away with re-installed these original ones. So that's what I did. I gave the knobs a twist to make sure they didn't stick and then let the glue overnight to cure. And the knobs were stuck solid the next day. Solved that by a firm twist with a pair of pliers. 

Fixed! Yay! Don't play with the knobs
or they may drop off.

Back to the floppy drive. The belt had, as predicted, become a sad strip of goo and needed to be replaced.

Belt has started to disintegrate

Gooey belt stuck to the main flywheel. Yuk!

In the course of moving the little green circuit board I heard a slight 'plink' of something hitting the floor. Unbeknown to me, these drives, also clearly built to a price point, have a small metal pin which drives the write protect function. This pin is not secured and simply drops into a tiny hole in the drive body. And simply drops out if you're not watching carefully..onto a small piece of patterned carpet...with the debris of many hours of soldering and other vintage computer restoration activities. 

Bugger.

Of course, at the time I didn't realise this was the case. But somewhere in the depths of my slightly confused middle-aged brain an Emergency Action Message was triggered that told me I should check the floor for something that I had never laid eyes on before. And sure enough, after half an hour of crawling around in the dust, solder blobs and occasional annoyed spider, I found it.

Dropped pin. Gah!

After checking the circuit board I located the write protect switch and made sure the pin was dropped into the correct hole. 

Back to the drive belt. After cleaning away the remains of the sticky old belt, which involved lots of scraping and cleaning with isopropyl alcohol, fitting the new belt was simple. The PCB was screwed back into place, which also then secures the write protect pin (grrr) and everything was ready to be re-assembled.

New drive belt goodness.

Controls re-installed

With the unit back together all I need now is a disk with the correct software on. I have a disk image for CP/M for PCW9512 but there are a couple of problems.

I have an additional floppy drive which is 3 inch but when I plug it into power, the drive makes 'chunk chunk' noises and the power supply (courtesy of the Cifer) goes into meltdown. A closer inspection shows that there are leaky caps just at the power connector.

Ohhh, another 3 inch drive.

 

It's an EME-01 (apparently)

Track damage just above the orange 
wire (mid left)

More track damage under the brown wire

Although I didn't have a full set of capacitors I did have the two values next to the power socket and a couple more on the main board. And changing them solved the problem. Bizarrely, that's the second time that changing the caps has fixed a fault and both times it was a floppy drive.

This 3 inch floppy drive is fully compatible with the Shugart standard. This means that I can use the awesome Greaseweazle with it. I managed to get a record of the disk that actually came with this drive - which was for the PCW8256 - using the Greaseweazle software with the intention of writing the the CP/M image for the PCW9512 back to that disk. 

But there are two major problems. This is only a single sided drive. The drive in the Amstrad is double sided. 

Second, the additional floppy drive only supports 40 tracks per inch whereas the PCW's drive is 80 tracks per inch. 

Arse.

Looks like I'll be heading to eBay for a CP/M disk then. Darnit.

It's worse than that, he's dead, Jim.. - Torch Triple X Part 3

Before anyone panics, the Torch isn't dead (I think). Rather, my ATX power supply is. While I was probing around the board, looking at signals, I noticed that the cyan display had actually turned magenta. This was different so I switched off the ATX supply, waited 20 seconds or so then switched it back on. 

And I got nothing.

After a little panic I realised that the green LED on the ATX supply hadn't lit up meaning that the supply was off. The red LED was lit though, indicating that it was in standby. Hmmm. I initially thought I'd popped a thermal fuse so I left it overnight and tried the next day.

Red standby LED is still lit
(centre bottom at the end of the black box)

Still dead.

So I tried checking the internal fuse and that was OK too.

Bugger.

Now I am in the situation where I'm not sure if the power supply died of old age, or whether a fault was on the Torch board that killed it. And, if there was, has that fault damaged the Torch mainboard? I know it was drawing a lot of current but the ATX supply could provide a crazy amount i.e. 25 amps at 5v and 16 amps at 12v. I could try and drive the board from my bench supply but then I can only provide the 5v. I may have to bite the bullet and buy another ATX power supply. I did have a spare but it went into the Amiga A2000. 

Unlimited Power! (Well up to 280W..)

To keep myself amused and useful I decided I should look at the schematics of the Torch and try to re-create them in KiCAD. 

I have used several CAD packages in the past but KiCAD is specifically aimed at preparing electronic PCB layouts and schematics. The learning curve is not too hard if you're only copying a schematic - and when I say copying I mean "make it look like the picture, I don't care if I've had to modify a 68000 block diagram to make the MMU..." so there's no chance of using my output to generate a new gerber file for the motherboard or anything daft like that.

The main reason I'm actually quite keen to get the schematics done is that the originals that I have are not really originals. They're photocopies of hand-drawn (possibly) originals and so the quality in some places has really gone down and there are several areas, notably at the very top or bottom of the drawings, where some information has been lost. This isn't as drastic as it sounds as it can pretty much be derived by looking at the context of the lines that seem to stop suddenly at the top (or bottom) of the page.

Anyway, here's an example which shows the 'Bus Masters' and is probably one of the most complicated of the drawings to generate.

Bus Masters. This is one of the most complicated ones.

While I work my way through the schematics and try not to lose my sanity, I will start searching eBay for an ATX power supply that also has -5V. 

“If it were easy, everybody would do it. Hard is what makes it great.”

With Great Power... - Torch Triple X Part 2

Quick re-cap. The Torch Triple X is uber rare, and has a slightly whacky power supply which allowed for 'soft start' and 'soft shutdown'. Unfortunately, to achieve this, amongst other things, the power supply actually had a battery in it.

Curse you generic battery supplier!

The leaky battery has very seriously damaged the PCB for the supply. In some spots the track has just gone completely and several resistors just disintegrated during cleaning of the board. So I replaced the resistors and the RIFA cap. I also bridged the broken tracks and reflowed as many joints as I could. Battery leakage is a real pain as, apart from the destruction of tracks it also combines with solder to give a grey, hard crystalline lump that is almost impossible to solder. This can be corrosive and loses conductivity so it is worth trying to sort out.

After several hours of work, I was ready for the first switch on. Following advice from Twitter (now 'X') I took an incandescent bulb - actually an old spotlight with screw cap - and soldered a couple of wires to it so that I could add it in series with the live wire. The idea is that if there is a fault, then excess current flows through the bulb making it glow bright. A dim bulb means that there isn't excess current - but no guarantee that there's no fault...

Good job I still had those bulbs...

So, what should happen is that when I switch the power on, there should be nothing from the supply until a pair of pogo pins have a small capacitance - provided by a finger - applied across them. This signals the board to start up and provide 5V to the main board, whereupon the main board will then send a 5V signal back into the supply to signal it should stay on.

For my test, I'm just intending to poke the pogo pins, check the supply starts up and then wait until it shuts itself down after two seconds (it does this automatically if the 5V from the main board isn't seen after two seconds). Here's a quick video of the first switch on:

Yikes.

So, unsurprisingly, I decide against poking the pogo pins with my finger.. I tried switching on again a few times with similar results. It doesn't get any better in slow motion..

Something is very wrong with this supply and, despite my best efforts, it may be beyond my skill to repair. The bulb going bright and then dim is fine as this is likely the capacitors charging. However, the loud cracks and arcing from the relay and the extra arcing from somewhere else on the board are not fine.

Ohhh - extra unexpected arcing. Spicy!

But fear not. After some careful thinking I came to the conclusion that the 'soft touch' feature could probably be bypassed if I used a straight PC ATX power supply and fed it straight into the motherboard. So I removed the connectors from the original supply and soldered some wires to the back of them allowing me to use my Heath-Robinson bench ATX supply. 

What could go wrong?

Going for broke and putting it all together..

So, it came to the first switch on. And I got nothing. And I mean, NOTHING. Hmmm.

After poking around the main board for a while I came to the conclusion that there must be a short to ground on the 5v rail. My reasoning was that the 5v was showing 0.3v. The resistance between ground and 5v was about 12 ohms (foreshadowing).

And now, dear reader, I made an error. One that may haunt me but I hope not. I decided to inject some voltage into the board using my bench power supply to check if there was indeed a short. I started at 4.8v with a current of 100mA. The voltage immediately sagged to virtually nothing and the current limit kicked in. So I increased the current to 500mA. Same thing - and I also realised that nothing on the board was getting hot or even warm. Increase to 1A. Same. 2A and I got the same. At this point I stopped as the current was going somewhere but I could not work out where.

Back to basics, I had another look at the connectors. But this time I took the time to take out all of the conductors from the plugs (relatively simple to do - poke the retaining hook with a sharp pointy thing and gently pull them out). And they were not good.

So I spent about an hour carefully dipping the conductors in white vinegar. The immediate 'fizz' made me realise that they were still heavily contaminated with battery guts.

Removing conductors - sharp pointy thing
in hand.

Then I took some very fine sandpaper and gently brushed the inner part of the female conductors and the outer part of the male conductors for extra cleaning. Putting them all back together I tried again, but this time with just the board and the 10inch monitor connected.

And this was the result:

It's alive! Somewhat..

This is a great start as it shows that there is still some life in there. As a separate exercise I also tried the clean cables on the floppy drive and hard disk 'ring'. Both sprang into life immediately where previously they had appeared to be dead. Most importantly, the hard disk started with no real complaints and made all the right noises for an old but very much working hard disk.

So the blue screen confirms that the video output circuitry is working. But how do we get any further?

After reading the technical manual, which is crap by the way, I realised that the old power supply would hold the NMI signal on the 6303 processor low for two seconds to start up. Grounding the correct pin on the main board connector would achieve the same thing. If the computer is switched on and at the blue screen then holding the NMI to ground for a couple of seconds will generally switch it off i.e. the display goes blank. 

This is also positive as it shows that the 6303 processor is doing its job and sending the right signals out to shutdown the system. 

After a couple of hours of probing and fiddling I noticed that after a certain amount of time the display would change. It would show a definite rectangle of garbage at the top followed by lines running down the screen. Sometimes these would start out black and then change to a dark yellow. I tried to take a photo and quickly realised how difficult it is to take a decent picture of a CRT! The 'best' pics are below:

Garbage.

More garbage.

From this display I am going to take a punt and say that there could be a RAM problem here. What the system should do when it starts up is copy the 'Caretaker' program to RAM and run it. This is what checks to make sure it knows you have the 'key' disk and then starts loading up from the hard disk (if present). I'm speculating that there is an issue with the RAM that is preventing this from happening or there is an issue with the video RAM that means I can't see what's happening. The video RAM on this is an odd chip I've not seen before and that is not supported by my RAM tester either - HM48416 which are 16384 word x 4 bit chips i.e. they have 4 data bits per chip. 

Another quick meander through the documentation showed that the system should beep to indicate startup and also provide patterns of beeps to identify issues which would be really helpful at this point. But this machine has been silent. I think that this is because the power supply does not have -5v. I had assumed that this was for the serial ports only but it looks like it may also be used in driving the sound signals that get piped through the monitor. Or should I say that there are + and - connections labelled 'speaker' in the video connector so that may be an assumption on my part.

To get -5v I could modify the -12v output from this power supply. I first tried a straight voltage divider but this didn't seem to work and I ended up with +0.3v on the -5v rail.. I may have to look for an LM337 or LM137 negative regulator.

But my money is on it being RAM in any case. More soon.