Monday, October 29, 2018

Simple Logic (Captain)

One of the more useful tools when trying to troubleshoot faulty boards is a small, innocuous thing that is known as a 'logic probe'. So, what does it do? It probes logic, obviously.

What it actually does is give an indication of a logic level. Computers work on ones and zeros. A logic probe simply indicates if the point where you, ahem, stick it is a one or a zero. It's really that simple. And I don't have one. Of course I could buy one, but where's the fun in that? :) So I set out to build a simple logic probe from anything I had lying around.

First things first, lets think about a circuit. Fortunately, I didn't have to think too hard as a helpful youtuber called Aldrin Floyd created a very short video showing about as simple a circuit you could imagine to produce a working logic probe. See his video here.

The schematic he came up with is shown below:



It's operation is simple. When the probe is placed on a low signal, the transistor is completely off, resulting in the full voltage being shunted through the red LED to ground through the probe. Boom. Red LED lights up bright. When the probe is placed on a high signal, the base of the transistor becomes saturated by the probe and allows the flow of current straight down to ground through the green LED. Boom. Green LED lights up.

(When there are no signals, both LEDs light up quite dim, which I assume is because the voltage drop across the resistor and red LED is enough to allow the transistor to partly turn on which, in turn, allows the green LED to turn on slightly. I'm sure an electronics expert would be able to explain it better than I can.)

 A strobing signal also lights both LEDs but is easy to distinguish from the 'just on' state as both LEDs become significantly brighter (with red actually being the brighter of the two).

A list of parts required:

  • A stiff piece of wire or a brass road of about 2mm diameter (for the probe)
  • An NPN transistor
  • A red LED
  • A green LED
  • Two 1k ohm resistors
  • Some wire (two core about two feet)
  • A couple of crocodile clips

First, take your probe and file the end into a point - the pointy end will be the bit you use most
Solder the Base of your NPN transistor to the probe.
Then, solder one of the resistors to the collector of the transistor.
Next, solder the positive lead of the green LED to the other end of that first resistor
Now, on to the red LED. Solder the negative lead of the LED to the probe.
Then solder the second resistor to the positive lead and the end of the resistor to the emitter of the transistor.
Nearly done.
Take one length of wire and solder it to the point emitter and resistor meet.
Take the second length of wire and solder it to the negative lead of the green LED.
Solder a crocodile clip to the other ends of the wires.

Annnnd.... you're done.

Once done, it should look something like this:

Bottom View

Top View

I did this in about twenty minutes and, having coated it in hot-snot to prevent shorts, it's ready to rumble. It's not pretty, I'll give you that.

Hot Glue Gunned

So to test. I took an Amiga A500 motherboard and connected the croc clip from the transistors emitter (the positive) to a positive supply rail on the board, and attached the other croc clip (the negative) to an earth point on the board. At this point, both LEDs light up dimly.

To check it was working I probed the positive lead of another cap on the board. The green LED lit up brightly and the red LED went off. Then I probed another earth point on the board and the red LED lit up bright and the green LED went off. If data lines are probed which are normally 'strobing' then both LEDs light brighter than when in their quiescent state but not quite as bright as a steady high or low signal.

A 'high' signal

A 'strobing' signal

A 'low' signal


And that's it. It's primitive and I'm sure that there are probably better circuits out there but for what I need, it's perfect and almost completely free. To pay lots of money would be illogical. Captain.

Friday, October 12, 2018

Back Home In...A500+ Land

After my foray into the CBM Plus4 (the worlds most fragile 8-bit computer) I have acquired another A500+ motherboard with battery damage. This feels like coming home and, after the confidence bashing of the Plus4, I've picked myself up and I'm ready to DO THIS! C'MON! RARGH! YEAH!!

Ahem.

Oooh - a parcel! What's inside?

It's only an A500+ motherboard with Agnus and ROM!

This time the actual board looks relatively unscathed. The same, however, cannot be said for Gary's socket. The pins are green and, er, mostly gone. Just touching the spring contacts resulted in them disintegrating... But amazingly, the solder mask has not been damaged at all which leads me to suspect that the Amiga was stored standing on end and leaning with its keyboard facing towards the floor.

I spy with my little eye...a borked Gary socket

But no solder mask damage. Hmmmm...


Anyway, this is another 'on the cheap' repair i.e. I have to do this with no additional parts other than what I have available from my previous escapades. Bear this in mind as you read on.

First things first, get Gary's socket off. Now, I don't have a lot of luck removing large chips from boards and I don't know why. It just seems that I haven't got the knack of removing solder completely from the pin holes. I've had some more luck with expensive solder wick but that is, well, expensive. So, I'd decided to bite the bullet and cut Gary's socket off as cleanly as possible with the intention of gluing it back together.

"But what about those disintegrated pins?", I hear you cry. Well, I have somehow ended up with a surplus of 16 pin DIP sockets which have the same style of pins. They are small pieces of spring metal that are pressed into the holes in the plastic socket. It's simple to remove them from the 16 pin sockets and re-insert them into Gary's socket. It is very tedious and I had promised myself I would not do this ever again. But needs must. See here.

Pins, lots of them...

After carefully gluing Gary's socket back together and then re-inserting the pins I soldered it back to the board. Didn't he do well?

As bright as a shiny new pin(s).

With Gary's socket back on the board I checked to see if there were any broken traces. Another chunk of time taken up with continuity checking, but worth it as I identified one missing connection which was simply fixed with a small wire.

Broken Gary trace - strange, but fixed.

Time to dig out the rest of the chips. The board is a straight swap with one of the guys on the Amiga FB group so we did it 'chips out' so to speak. After fishing them all out of the box, and checking I had the right ones - Denis and Agnus are ECS and I have some OCS chips in my spares box - I was ready to rumble. So, the first attempt to boot. Does it boot?

No. Green screen.

Here's a description of the Amiga boot colours, taken from the classic Amiga wiki.

When the Amiga boots, if it encounters a hardware fault the system will display a solid single coloured screen, or a series of these screens if more than one error is detected.
The following colours are valid for all versions of AmigaDOS since Kickstart 1.3. However, there were several more colours in 1.3 compared to anything included after Kickstart 2.0 was released. However all of the colours listed below are the same for all versions of Amiga kickstart.
  • Red - An error in the Kickstart rom as detected.
  • Green - An error in the Chip Ram was detected.
  • Blue - An error in the custom chip set was detected.
  • Yellow - The CPU encountered an error before the system's error-trapping code (the code the calls up the Guru) was in place.
  • Black - No CPU detected.
  • Grey - CPU Passed the test.
  • White - CPU failure.
In addition to colour screen error codes, Guru Meditation error screens are also a common error screen encountered when an Amiga encounters a problem

This is helpful but doesn't give much detail. All I can glean with a green screen is that my chip RAM may be borked. But Agnus may be faulty, or it could be a damaged trace or it could be something else. There is more information available on the actual sequence of startup:

Taken from Lemon Amiga at this link. I've reproduced it here as this type of info has a habit of disappearing over time...

Amiga Boot Sequence 

1. Clear all of the chips of old data. 
2. Disable DMA and interrupts during the test. 
3. Clear the screen. 
4. Check the Hardware to make sure the 68000 is working. 
5. Change the screen to Black, screen stays Black (or with red Stripes) if CPU fault. 
6. Do a checksum test on all the ROMS. 
7. Change screen to Red if ROMs are faulty, or change to Dark Gray if ROMs are OK. 
8. Begin the system startup. 
9. Check the Ram at $C0000, and move SYS_BASE there. 
10. Test all of the chip RAM. 
11. Change screen colour to Green if RAM is faulty, or change to Light Grey if RAM is OK.
12. Check to see if the ROM software is coming in OK and being executed. 
13. Change colour to White if the software test is OK, stays Light Grey if faulty, or Blueif Custom Chip error. 
14. Set up the chip ram to receive data. 
15. Link the ROM libraries so the machine can identify connected devices and peripherals. 
16. Check for additional memory in these devices and link it to the computer. 16b. Chance screen colour to Yellow if there is an error with external cards/devices. 
17. Turn the DMA and Interrupts back on. 
18. Start a default task. 
19. Check to see if there are additional CPUs on the system, accelerators or a maths coprocessor. 
20. Check to see if there CPUs are OK. 
21a. If there is an Exception (processor error), or issue reported, the system will reboot. 
21b. The Kickstart Image is Displayed showing that the rom has been installed to memory, is running, and everything so far has reported OK. 

After this, the ROM GURU trap handler will display errors as codes on the screen. 

My A500+ board fails at step 11 i.e. the chip RAM check failed. No great advance there. After poking around at the logic chips (U10, U11, U12 and U13) it was obvious that something was not right with U12. The voltages at the output pins were all very low i.e. non-existent so I decided to start here. This is good as the chips at U12 are cheap - and I already have a couple of spares - and can easily be socketed.

Removing U12 was not easy. The problem is that the leaky battery liquid death reacts with the solder, resulting in a crystalline lump that does not want to melt. After failing miserably to get more than a few pins released I resorted to an articulated pin disconnection tool (a pair of wire cutters). Cutting through the pins on one side allowed the chip to be bent up and, after a few wiggles, the chip came off. It was then a simple matter of flooding the pin area with new solder and using a solder sucker to remove the mixed old crappy stuff and new stuff, pin remains included. Repeat for U10. Once done a simple clean with some IPA was carried out.

Before soldering a socket in, I decided to check continuity again at U12 since this was the chip that took the brunt of the Varta death. The results were not good:

Pin 1 to pin 19 - broken
Pin 2 to U13 pin 2 - broken
Pin 3 to U13 pin18 - broken
Pin 7 to U13 pin 14 - broken
Pin 8 to U13 pin 9 - broken
Pin 11 to U13 pin 12 - broken
Pin 18 to U13 pin 3 - broken

It looks like the battery had eaten a good chunk of the tracks where they are not protected by the solder mask, just at the point where they meet the solder pads (holes) for the pins. Cleaning up with flux, solder and then IPA was probably enough the break the last vestiges of track at these points. I have had this before - see here.

Pin 1 looks unhappy

Generally unhappy here too.

Urgh..

Flippin 'eck. 

OK. At least I now have something to aim at. I'll need link wires on the underside of the board to rectify these broken traces. I also need to break out my box of sockets so I can put in new chips but have the ability to remove them easily.

Next problem. No 20 pin sockets. I do, as I mentioned earlier, have a big bag of 16 pin sockets though. And a pair of wire cutters. And a tube of superglue. Hmmm. I have a cunning plan. What could possibly go wrong? ;)

The sockets are quite brittle so I had to resort to cutting the socket a little bigger and then filing down the rest to avoid cracks. A dab of superglue was perfect to attach these extensions to an unsullied 16 pin socket. The results, while not perfect, are more than acceptable.


Socket to 'em! (See what I did there?)

One final point, which I have mentioned before elsewhere in my blogs. When soldering sockets on to the board, always do opposite corner pins first. That way, you can press the socket down onto the board and ensure it's flat before soldering the rest of the pins, otherwise there is a risk it won't be flat and will look like it's making a bid for freedom.

Talking of which, have a look at this chip here - U15. Would you believe that this was from the factory?? It looks like the chip was supposed to have its legs folded under the board, like all the others, but for some reason they only managed to do this on a few of the legs at one end. This allowed the chip to dangle and start to fall out of the board. I'm not sure of the process they used to solder components to these boards but, more by luck than judgement, this chip managed to get soldered and stay soldered..

U15 - Have you got somewhere else to be?

Anyway, sockets now sorted.

Now, link wires. I thought I had a stock of kynar type wire which was green and is nice and fine. I had about two inches left. Fortunately, I also have a winding of copper wire which is coated with laquer. This will do nicely instead. All I need to do is scrape off the laquer at each end and solder as normal.

Here's the list:

Shopping list of new link wires...

New link wires installed. Nice.

So now, with new sockets, links and the addition of a couple of new 74LS244N chips, lets start this Amiga up again. Does it work?

It's Alive!
Of course. :)



Tuesday, July 31, 2018

Plus 4 Adventures - Part 2

If you read the Plus4 Facebook group you'll know that I had all but given up on my Plus4. It displayed garbage on the screen at startup. Pressing reset occasionally got back to the boot screen but then garbage would return almost instantly. After a few minutes things would get progressively worse until, after turning it off then back on, only a black screen was displayed. The list of things I did is below for posterity. It was heading for the bin...

Then, in stepped Ian from Mutant Caterpillar...

Ian (not to scale)

Long did I wander, searching for the mystical stranger, known only as 'Ian'. Through valleys and forests, across mountains, down the A44 (turn right at Rhayader), until the distant horizon stretched out before me into the deep, foreboding ocean. Wind whipped around me as my hands clasped the precious package, strange smells emanated from open doors, mixing and polluting the air (the local KFC - well worth a visit). I knew I must continue - after I'd finished my boneless bucket - for the end of my journey offered hope of redemption and a new life. Finally, I came to his door and threw myself on his mercy.

Me.

And, Lo! he did take my Plus4 board and it was re-made in the forges of Aberystwyth! And strong was the tape loading, and long was the soak test! And through it all it became known as 'working again'!

My board with 8551 removed
(Pic from Ian on FB)

Loading from tape!
(Pic from Ian on FB)


Yes, Ian at Mutant Caterpillar fixed my Plus4 in less than a day. He traced the fault to the 8551 I/O chip spewing random crap on to the data bus. There was also a dodgy RAM chip (not a surprise). I had thought about that particular chip - the 8551 - at some point early on in the process of all this but I'd not done anything as it didn't really connect to anything that I was going to use..

So, what have I learned in all this?

1) If you have a computer that needs repair you seriously can't do better than Mutant Caterpillar
2) No, really, re-read point 1.
3) Plus4's are delicate beasts but, ultimately, worth the effort as they're getting rare (in a working state)
4) Don't discount anything unless you've tested it!
5) Did I mention how great Mutant Caterpillar are?

Fin.




My escapades retained for info below:

My Plus4 is faulty and I'm having a hard time finding the fault. Note that it runs a 6510 rather than an 8501. The 8501 was identified as dead early on. I also have a replacement kernal which includes modifications to accept the 6510 (in case you were wondering).

Here's a video of the fault in action:


Although I said it dies, it actually lives... A while after this video was shot I tried turning it on again and it worked for about 5 seconds before I got a repeat performance.

So, things I have done, not necessarily in chronological order:

1) Checked the power supply - there is a steady 5.23v coming from the supply which reads as 5.13v at each of the chips. There is just over 10vac on the other part of the supply which shows up OK on the board. Power supply seems OK.

2) Just in case I ran it with just a straight 5v 3A supply. Same symptom as in the video.

3) Replaced TED socket. The original was a single wipe and I couldn't see the condition of the contacts so I just replaced it with dual wipe socket. No change. Same symptoms as above.

4) Replaced TED. At least, a known more or less working TED. The 'new' one I have does not like joysticks but otherwise operates as expected in a working machine. No change. Same symptoms.

5) Replace CPU socket. Although the 6510 adaptor board has nice turned pins, the original socket was another single wipe affair. I replaced it with a shiny dual wipe socket. No change.

6) Replaced TED socket. Same again really. The original is a single wipe and I replaced it with a dual wipe. No change.

7) Removed 3+1 ROMs. These prompt a white screen when run anyway so I just removed them (but keeping them safe!). No change.

8) Replaced PLA socket. Now out of sockets. No change.

9) Obtained a Diag264 cartridge. This seems to start but then stops after a few seconds presumably because of the fault.

10) The Diag264 cartridge managed to run almost a complete set of tests before the fault appeared. Here's the screenshot:



11) Re-capped it. More or less. This is controversial but at least if with re-capping it then I will not be wondering if an out of spec cap is causing all my trouble.

12) Start looking at the RAM (more unobtanium). The strong suggestion is a RAM fault, so I may do this first. Strangely, the MT4264 chips in my Plus4 are actually 4164 1*64 DRAM which I think I have few of somewhere... I didn't but I managed to get some from Retroleum.co.uk.

Friday, July 20, 2018

Plus 4 Adventures (Part 1)

In my batch of retro computers that I bought from eBay there was a Commodore Plus 4. This is an interesting machine and there are plenty of videos on YouTube describing it's history in more detail than I ever could. I highly recommend watching any of the videos from Bill Herd who actually worked on it (he wasn't the designer though! ;) ). Try this one, or this one from The 8-bit Guy who has also recently covered the Plus 4 as part of his Commodore history series (in which a certain Bill Herd also makes an appearance).

Stock Picture - Mine isn't this clean...
To start, the Plus 4 has a similar power supply to the Commodore 64 but uses a four pin square plug, similar to the C128 or Amiga but with fewer pins. These connectors are impossible to find these days. Fortunately, I already have a power supply that I have checked is OK (for supervised testing) and so I'm ready to rumble.

Unobtanium
My purchase included the two manuals, dataset and joystick as well as the computer itself. The only video cable supplied was a standard RF which my monitor won't support so I had to resort to a slightly bodged up arrangement as I don't have any 8-pin DIN plugs..

Wires soldered to strategic points for composite video..

Anyway, I connected everything up, turned on my monitor and hit that 'on' switch. On a positive note, my power supply did not burst into flames / destroy the world / kill any kittens. Sadly though, the only display I could get was a black screen with lines slowly moving down it. Arse.

At this point, we should probably talk about how fragile these things actually are. Let's start at the top.

First of all, the CPU is not a 6502 or even a 6510 as used in the C64. It's a special version of the 6502 called the 8501. Earlier machines had the 7501. Either way, these things die and die quickly. I have read that Commodore didn't care much as long as they lasted over the year warranty period. In some cases they lasted weeks. There is apparently a fault in the design of the chip which focuses heat in one spot, makes it massively overheat and then, well, die. Once that happens, it's all over.

Then there's the TED chip or TExt Display chip. This is the second 'heart' of the machine and drives most of the functions outside the CPUs control such as the video and sound generation. The TED chips also suffer from the problem of heat. They get hot. They die.

Then there's the PLA. This acts like a traffic cop and directs signals and data around the main board. It runs hot. It dies. (Can you see a theme building here?)

In other words, unless you lived somewhere cold and kept your heating off, these things were always going to die and are now mostly dead. I have it on good authority that about 80% of Plus 4's with faults have dead 8501s.

So, what ails my most fragile of machines? Having left it on for about ten minutes I used a sophisticated thermal detection unit (my finger) to determine the relative temperature of each chip. All were getting nice and toasty, except one. One chip remained stone cold. Given the reputation these have I think I can safely say I knew which chip was dead. Can you guess? You have an 80% of being right.

Yep. Dead 8501. Just to verify this I sent my chips off to Mutant Caterpillar who very kindly offered to check them for me for the cost of postage. Sure enough, all chips OK, including TED, ROMs etc, except for the 8501 which is officially dead. It is an ex-cpu. It has shuffled of its mortal coil and joined the choir invisible etc. Double arse.

I need to consider that there may still be other faults on this Plus 4 so what do I do? There are several options here:
  1. Give up, stick it back on eBay as faulty for parts
  2. Scour eBay looking for a donor C16/Plus 4 or used 8501
  3. Fork out for an FPGA replacement 8501
  4. Fork out for a 6510 adaptor board

Option 1 is out. I want to get this working if I can. Option 2 is unlikely to yield any results as working C16 or Plus 4s are way more expensive than Option 3, and non-working machines are likely to have dead 8501s anyway. Option 3 is a fully compatible, modern FPGA replacement for the 8501 that does not suffer from the heat issues of the 8501. It's also £40 plus postage from Poland. Option 4 seems the best compromise between cost and functionality since the 6510 requires a revised Kernal and has some minor incompatibilities which I'll highlight later.

As it turns out, sellmyretro revealed that there were no FPGA boards left anyway so I decided to go for the 6510 board. And it has already arrived. Yay!

There are two steps to fitting this board (possibly three). First, remove the dead 8501 and fit the 6510 board. This has been designed to push the chip backwards thereby allowing it to fit in the existing socket without fouling the keyboard. Earlier versions required the socket to be removed and the board soldered straight to the main board. Next, remove the existing kernal chip and replace it with the supplied EEPROM. This contains a modified kernal that allows the serial port to be driven by the 6510.

New 6510 and modified kernal installed

There are a couple of things with this board. First, the motor in the datasette runs all the time. This is due to the 8501 having one more i/o than the 6510 so, by default, the motor runs all the time. It's fine to run like this or, alternatively, by soldering a diode to the board at a specific location, it turns the motor off unless the buttons are pressed. It does mean the Plus4 can't control the tape deck itself, though. In my case this shouldn't be a problem.

Anyway. It's now installed. Does it work?

Yes. 

And no.

The plus 4  boots. That's a massive improvement. 


It's alive!
Because of this I went on the hunt for some heatsink material that I could install on the precious TED and PLA chips. Having dismantled a couple of old Freeview boxes I found two ideal heatsinks contained within. A dab of thermal paste and the two chips are now protected and should last a bit longer...

Feelin' hot, hot, hot
Feelin' less hot, hot, hot
Moving on. Let's run the internal software and see if it's really that bad. I won't be able to do much as the software only allows files to be saved to disk and not cassette. It doesn't matter too much but it would be interesting to see what it was like. Except I can't. I can press F1 which brings up the command to run the internal software. But as soon as I hit return , the screen goes blank. The colour seems to be white or a very pale grey. Oh, and I have to reset to get back to a working screen...
Bugger.

Next, I tried the datasette. With the 6510 the motor always runs (as mentioned) which should not be a problem but I wanted to double check. I don't have any Plus4 tapes so I typed in a basic program and tried to save it tape. I typed in 'SAVE "CIRCLES" but as soon as I hit return I got a flash of a message "PRESS PLAY AND RECORD" and then a blank screen which could be pale grey or white....
Out of interest I entered the command 'LOAD' and pressed return. The same thing happened. I had a quick view of the message "SEARCHING" and then it locked up and gave me a white/grey screen.

F.

Figuring stuff out...
So now what? Fortunately, I had a conversation with Ian from Mutant Caterpillar on FB. In my ignorance of the Plus4, coupled with the limitations of the 6510 mod, I hadn't realised that the white screen when trying to 'SAVE' or 'LOAD' was normal. You see, when the command 'SAVE' is entered, the computer instructs you to press play/record on the tape deck. But with the 6510 the tape motor is running all the time so it thinks that you pressed the buttons on the tape deck immediately. The Plus4 is the same as the C64 and blanks the screen when performing tape operations. 

To put this to the test, I entered the command 'SAVE "FOO" and pressed return and waited. About a minute later, the screen came back and I had a working Plus4. I am, officially, a dumbass. Oh, and it appears that the 3+1 ROMs are actually faulty but they can be removed with no impact on the Plus4's operation.

So I have a working Plus4 now right? Hang on one second there. You see, when I was entering 'SAVE' I actually noticed that the tape motor was NOT running even though the datasette was plugged in. I turned it off and unplugged the datasette, plugged in back in and tried again. Garbage appeared on the screen for a few seconds, then a black screen. It just died. 

Later on, I came back to it and tried again. It actually seemed to come back to life so I grabbed the camera to record some footage for this blog. In actual fact, I recorded the Plus4 dying, on camera, this time permanently. It is dead. Completely. On switching on, there is only a black screen. So something else has gone wrong with this Plus4.



It's dead.

I told you they were fragile.

Tuesday, July 10, 2018

All Hands on (Tape) Deck! - CPCs Part 2

I am constantly amazed by how quickly time marches by. In the blink of an eye we went from arguing whether the Spectrum or the CBM64 was better (the Speccy obviously - I mean, c'mon) to arguing about whether iPhone or Android is better. The more things change, the more they stay the same...

In the same way, how we handle data storage has changed dramatically. As my brain sees it, this has played out over the last couple of years or so. As reality sees it, this has played out over the last three and a half decades. 

When we got our first computer - a shiny ZX Spectrum Plus - it required programs and games to be loaded via tape. This was fairly standard for most home micros of the time, at least in Europe. Our US cousins tended to use discs instead, the reasons for which are debatable.

Anyway. Tapes. Tapes were generally reliable, they were definitely cheap and absolutely sloooowwww. With a typical data transfer speed of 1535 bits per second, it took about five minutes to fill 48 Kilobytes*

*A note about units. I've said it before, I'll say it again. A kilobyte is 1024 bytes. It is not 1000 bytes. A 'kibibyte' is a steaming pile of horseshit dreamt up by someone who clearly couldn't understand powers of 2. So there.

But tapes were cheap, relatively easy to mass produce and, good news for all us schoolkids, monstrously easy to copy. Not that I ever did that...nope, no siree, not me.  To my 12 year old son it seems quite quaint to have a bulky mechanical extension attached to the side of keyboard. And, oh, the horror of having to wait up to five minutes to play a game (I suspect he has never measured how long it takes to get the average PS4 game up and running).

I digress. The CPC464s that I have, have been re-assembled and had some limited testing. They appear to work based on my 'heath-robinson' video cable (yes it's a couple of bent bits of wire pushed into the correct pins with a phono plug on the other end). On the banged up one (which I shall refer to as 'Bert') a tape will play as soon as 'play' is pressed - sometimes. On the one that is in slightly better condition (which I shall refer to as 'Ernie'), the load sequence has to be initiated before the tape will start to play. Curious.

Nicking a composite output...
Also, Bert's deck sometimes won't rewind unless the button is given an extra stab, and the tape counter reset button on Bert is smashed to pieces. I think we can safely say which end hit the floor first when Bert was clearly dropped.

Ernie's deck seems to be OK in that all functions work without issue. The tape counter is fine on Ernie. I suspect Ernie just needs a new belt.

Time to put my repair pants on.

Getting the tape deck out requires the removal of a lot of screws. And the power switch. And the volume control. And the keyboard... Actually, you can get away with not removing the keyboard if you de-solder the power LED from he cassette control board. So, once you've done that the whole thing lifts out of the case.

Complete unit. LED removed from case (not desoldered)

Both decks needed a darn good clean. Isopropyl alcohol and cotton buds. Enough said on that. To cure the slightly hesitant rewind on Bert I got a small piece of 1000 grit sandpaper and very gently rubbed it between the flexible switches on the unit. This worked surprisingly well and rewind now works first time, every time. It also cured the accidentally intermittent 'play'.

Flexi-switch - now works properly.
There is another definite fault on Bert's tape deck. A previous owner has soldered two wires together from where one the flexible switches makes contact. I tried putting them back the way they should be but then the tape deck didn't play at all, so I put it back the way it was originally. I'll just have to live with it. Strangely, the relay in the deck still clicks when the CPC is trying to start or stop the tape but, with the wiring modified the way it is, it just keeps on playin'.

Blue & Purple wires are linked. Hmmm.
On Ernie, I noticed that the pinch roller had a dent in it. I realised that someone must have pressed the play button and left it in that state for some time. This dent resulted in a distinct wobble to the sound every second or so. It didn't affect the loading of the game I managed to record to tape (Harrier  Attack - purely for testing purposes ;) ) but was very, very noticeable. This was a pain as I intended to sell Ernie on to help re-coup the initial outlay on my bundle. I began to wonder if I could swap the pinch roller over.

After imagining being neck deep in mechanism bits and springs and washers I decided, what the heck. If I could remove it easily enough from Bert then I should be able to do the same with Ernie. Fortunately, I was correct to have a go at this. The pinch roller was held onto a plastic shaft by a simple plastic pin that pushed into the top of the shaft. Gently prising it up allowed it to pop off without any drama. I repeated the same on Ernie and immediately installed the non-dented pinch roller in about ten seconds. Easy. Certainly easier than I originally expected!

While the dented pinch roller was out of the deck I attempted to squish it back into shape before re-installed in Bert. Bert will be staying with me so I have no problem with a bit of a 'wobble' from physical tapes. I managed to make it a little better but then decided to go all out and sand it down a bit. (*Imagines the gasps of horror from CPC aficionados...*). It actually worked quite well. The rubber is starting to perish and so I had to be careful not to take too much off or knock chunks out of it.

Now, the main event. Both decks need new belts and, as luck would have it, the chaps at Retro Computer Shack had a twin pack of belts for only a few pounds. Installing them is relatively straightforward, if a little bit fiddly. On the underside of the deck, simply undo one retaining screw, lift the arm and pull out the old belt. Fitting a new one is the same, in reverse. This took about five minutes on each deck and the new belts are nice and firm and grippy. Nice.

Beltin'

Repair action shot.

The buttons also required some attention as they were really dirty. To get them off is also fairly simple. Remove the circlip at the end of the shaft that runs through the buttons. Carefully remove the shaft and the buttons should drop. Then carefully remove each button from the frame. There a tiny springs, three in total, on the bottom part of the buttons that stops them rattling around when they are 'on'. Try not to lose these! They have a habit of pinging off when you remove the buttons so be careful!

Pingy springs - easy to lose, impossible to find again...

The buttons were soaked in soapy water and cleaned with an old toothbrush. Isn't everything? ;)

While I was at it I decide to have a look at the broken tape counter reset in Bert. The button had snapped off the end of the wedge that pushes against the counter mechanism. I tried to fix this with superglue and a small piece of wire across the previous break to give it some strength. It sort of worked but the mechanism is quite stiff and I suspect that the repair has partially failed. It's OK though, as it's more cosmetic than anything and I intend to use an input mod to load stuff up so I won't actually need the counter to work properly.

Before (partially) successful repair.
At this point Ernie is complete. Captain obvious alert. Re-assembly of the CPC is the reversal of disassembly. No great shakes, just a lot of screws... but Ernie is ready to be sold to his next owner.

Bert is getting a little extra something. The thing is, I have no tapes. Or at least, I only have one tape that I've managed to get a working copy of Harrier Attack on to. To allow me to load tapes when I don't have tapes I shall add an input socket. This is easy enough to do by soldering a couple of wires to the tape head and feeding those wires to a 3.5mm socket. By using a 3.5mm plug to 3.5mm plug cable I can now feed sound directly into the CPC from my PC. See?

Wires connected to the head.

Epoxy'ed into place. 

Exterior view - looks almost factory.


Wide angle view.
Using a simple cable and a piece of software called 'CDT2WAV' I can now load any game. It's not perfect as I have to still press play on the tape deck and, of course, the games still load at uber-slow speed. But at least I can play them!

I've really enjoyed playing with the CPCs. They're surprisingly well built and are refreshingly simple in their construction. I'd highly recommend you get one of these if you're looking to start a retro collection as they are easy to work on and, in the main, the only bits that fail are the tape drive belts.

The Speccy's still best though.


Saturday, June 23, 2018

A Brace of CPCs..

That wonderful site, eBay, can be a strange place. Sometimes it's possible to find a bargain but most times it's basically someone either:

a) trying to rip you off by demanding ten times the actual value of an item - "it's old so it must be worth a fortune, right?"

b) trying to scam you - "please post to my brother-in-laws cousin in North Africa and I will send you twice what you asked by cheque"

c) selling broken crap - "sold as untested as I don't have the TV cable that I could get for £1 from Wilko 'cos I know my dog peed in it and it's never gonna work again.".

This time, I struck lucky. For a modest sum I bought a job lot of computers which included:

1 x Commodore Plus 4 with cassette drive, joystick and manuals
2 x Amstrad CPC 464

eBay Score!

There were no power supplies with any of them but, as luck would have it, I managed to acquire a Plus 4 power supply which could have been problematic - but the Plus 4 is another post to come. For the CPCs I just needed a 5V, centre positive supply which I happened to have already, and a video cable.

So, the units themselves. They look like they've been in a very dusty, dirty environment for a very long time. They may have been in a barn given the amount of dirt (not dust - actual dirt) that caked every inch of both of them.

A Bit Dirty
Yep - Dirty.

By peering into the user port (and then I took them apart) I can see that they are both slightly later versions of the CPC 464 as the motherboards only occupy half of the enormous case. The joystick port and headphone socket are at the end of the case too, rather than on the back. The motherboards are very slightly different to each other though. One has a 40007 ULA, the other a 40010. The 40007 is heatsinked and is located in the 'other' space for a chip on the board. In the photo below, there is a big gap with the outline of a chip just above the 40010 ULA - this is where the 40007 is on the other board.

Does this look big on me?

One of them has been clearly dropped at some point too as there is damage to rear of the case at the cassette recorder end (also visible in the photo above).

A very brief CPC464 history. Alan Sugar saw the computer market and wanted in. Amstrad released the CPC464 on 12th April 1984 following a bit of a false start. But with typical AMS business sense he made sure that there was software available and the machines just worked out of the box (unlike the Sinclair QL). It's worth reading the entire article linked to above. It's a surprisingly interesting read.

Clean up time. I decide to start with the non-damaged unit (comes with the 40007 ULA). Dismantling these things is straightforward. Six screws in the bottom, lift up the top carefully and remove the cable from the motherboard and the keyboard connections. Early 464s had a cable to connect the keyboard but later ones switched to using membranes, and it's this type that I have.

There are another six screws holding the keyboard to the top case and six more holding the motherboard to the bottom case. Amstrad must have over-ordered on screws...

Once the keyboard is free of the case, it's tempting to just start pulling keycaps off. Don't. On the back of the assembly are small black clips. Press on these gently to release them from the metal back and, once they're all released, the back plate and membranes can be removed leaving just the actual key assembly itself. Turn it over and it becomes obvious how to safely release each keycaps with little chance of breaking them. If you just start pulling key caps you WILL break them. So don't. 😀 Using a finger to hold one clip, use a screwdriver to gently push the other. The keycap should just pop out. 

Membrane and Backplate

Keyboard Unit (and glasses)

Be careful, there's a spring on the other side! Don't lose any of them! This style of key return is similar to that of Amiga keyboards. The main difference here though is that each key has a smaller spring embedded in it which is what makes contact with the keyboard membrane. These smaller springs are fairly well secured in the keycaps though so you shouldn't lose any of them. 😀

Keys and Springs - Don't Lose Any...
Yuck.

With all the keys removed, I washed the black plastic section with mild soap and water. It was, to put it mildly, a bit grubby.


That's Better.
Then, each key was washed and dried by hand. And there's a lot of them. It took about an hour to do all of them but the results are great.


And....relax...

It's worth noting that there are several keys that have metal bars attached to them, notably the 'shift' keys and the space bar. These need care but are relatively simple to remove. Press the key from the key side and on the back, gently push the bar out of its clips. If done right it shouldn't need too much pressure. Re-fitting is just the reverse procedure. Drop the clip into roughly the right position, press the key and gently ease the bar under its clips. Simples. 

Next, the case sections need washing. To do this I removed the motherboard (duh!) from the bottom section, and the tape drive, volume control, power switch, power LED (might be glued in - be careful!) and speaker from the top. There's another load of screws to deal with here and, if you have one of these computers and decide to do something similar I would recommend you take a picture before you start so that you can see the routing of all the cables when you have to put all those pieces back together...

Bath Time!
Once the case had been thoroughly washed down with a mild detergent and a bit of help from a nailbrush, they were toweled down and left to air dry.

End of part 1. In the next part, I put it all back together to see if it works...