Battling Entropy

One of the things I have noticed is that Retro Computers can spontaneously stop working. These systems are often 30+ years old. Expecting them to still work beyond a presumable lifespan of 5 years is a bit much. Here are some system repairs I conducted recently.

Apple IIgs Keyboard

This one was interesting. They keyboard itself worked fine, but whenever I plugged a mouse into it, the mouse would “judder” all over the screen. I suspected a cold joint on the mini 4 pin DIN initially, but reflowing didn’t make a difference. A post on FB had the Group Moderator, Jeremy Barr-Hyde suggest replacing the two tiny electrolytic capacitors. I tried this too, with no luck. I was starting to look around for a controller chip for the keyboard, and was doing another inspection when I noticed an incredibly fine hairline crack through one end of the PCB.

I polished back the solder mask and it became more visible. It ran through several traces on the board. I used my fiberglass pen to expose more PCB and ran some solder across the cracks. For the largest area, I also reinforced the break with a spare leg from a large capacitor for extra strength.

I painted all the exposed copper and solder with nail polish, put it back together, and much to my joy, it worked again. This keyboard has the same styling as my Apple IIgs, so I was very glad to get it working again.

VZ200

A gift from a fellow collector, my VZ200 had been gifted to me as working, and had progressively got worse. I’d ordered an expansion cartridge for it (The excellent Ben Venn SD Loader) but it wasn’t reliable enough to even boot.

I’d procrastinated on this system for a LONG time due to the annoying nature of its construction. There was a permanently joined keyboard, a huge soldered down RF shield, and two daughterboards connected by flex cable. Flex cable is nasty stuff that has a habit of snapping down against the PCB where the wires enter. This is because after 30 years, the plastic has gone all hard and inflexible so the point where there’s still flexibility is where the wires are exposed.

Nasty stuff.

My big issue here was getting to the key chips to check what’s going on would involve removing all those daughterboards and the RF shield.

I began the project by desoldering the RF shield. This was a combination of desoldering gun and braid. One trick I have discovered is my desoldering gun works a lot better if I touch the tip of my soldering iron to it. It just adds a bit more thermal mass and components tend to drop out. I really should look at getting a new desoldering gun.

After that I desoldered the two daughter boards and replaced them with pin header and pin sockets. Finally I built up some ground leads to replace the RF shield. I figure you can never have too much ground plane.

At this point I quickly tested it and… nothing. The PSU made an angry buzzing noise so I turned it straight back off. That noise usually means a short and, yes, across the +5 and GND was a dead short.

After much tracing and removing all the bonus ground leads, I finally found the short near the voltage regulator. Back on went all the ground leads and tested and… It worked!

Not just “You have the garbage screens again” but full on “Your system is booting and you can type a “Hello World” BASIC program in.”

LOLWUT?

It’s not perfect. Every now and again it reboots on power up, but I swapped out the electrolytics near the PSU and that has certainly stabilised things.

No idea what the fault was, but she’s good now.

(OR IS SHE??? I can’t get the Ben Venn cartridge to recognise games. I’ll raise it in the VZ Facebook group and see what people have to say)

Dick Smith System 80

The System 80 has had a long term problem loading games from tape. Unlike the Amstrad where I was successfully able to load games from a “CD-to-Tape” adapter, I was having significant problems with doing this from the System 80, despite everything being stacked in my favour. (the System 80 has a volume level meter built on. I could never get volume to show above about a quarter.

Consulting with the fine folks at the Microbee Technologies website (They have a section for Non Microbee Systems that often covers System 80s) it was suggested I mod the System 80 to swap the onboard and internal tape decks, which was apparently a very common mod “back in the day”.

How common? It’s documented in the technical manual for the System 80.

What did I have to lose?
(Hint: Quite a lot of time)

Effectively the mod is to put a simple DPDT switch in the back to switch the two signal lines for the internal tape to the external tape and vice versa. Easy mod from the manual, but several things went wrong…

Firstly, while trying to desolder a component I lifted a trace. My own silly fault. I was trying to lift the component in an awkward spot as I didn’t want to remove the board. (Foreshadowing)

Now I HAD to remove the board and put a bodge wire on the underside. After some work, I was able to use an unused cutout in the back of the case to drop a neat little switch into place and everything was good…

…except the keyboard was playing up.

Ruh Roh.

See, the System 80 uses these truly terrible ribbon cables. They’re made of single core of wire between plastic sheets. There’s one between the keyboard and the main board, and there’s a second one between the main board and the video board. The plastic becomes stiff with age and the single strand of wire is not as flexible as stranded wire would be. Yes these are pretty much the same ones as in the VZ200, except longer, wider and more brittle.

I did some sounding out and the keyboard cable seemed to be OK. I could see the signals needed making it most of the way across the board. I swapped one suspect resistor, and eventually I swapped out a logic chip near the keyboard that the keyboard passed through.

It’s at this point the whole system started bugging out. Garbage on screen.

Ruh Roh.

I broke out the oscilloscope and started tracing around. The Z80 seemed to be dead in the water and some probing revealed something interesting: No Clock!

Referring to the schematics (The technical guide is excellent here), I followed the clock back, first to a buffer, and after a false start, through to the ribbon cable between the main board and the video board. (The clock is generated on the video board. No I don’t know why either. There’s lots of “interesting” choices in this design)

I now checked on the video board and was able to trace the clock as present all the way to the ribbon cable. *Sigh*.

Checking continuity on the cable and sure enough, it’s gone non-conductive on the clock wire. The clock is the very first wire, too, so the most likely to break.

I kludged in a replacement for just that wire as I wasn’t quite ready to replace all 32 wires in the ribbon cable yet. (More foreshadowing)

This gave me a clock at the Z80, and when I plugged in the video, I got a recognisable, if corrupted screen. W00t!

By this time I’d been battling the system all day, and needed a break. I did work through the logic to see that there was something wrong with Data bit 3.

I stopped for the night (And for a quick drink) and left it for the morning.

Next morning I posted my progress on Facebook and started further investigation. Initially I was confused as my first theory was that bit 3 was failing on the ribbon cable, but probing around with the multimeter seemed to show a good connection. Later I realised I’d miscounted which pin the bit 3 was on.

A good suggestion from FB was to check if it was one of the two 2114 video SRAM chips was the source of the problem, but swapping them didn’t change anything. This was the point where I realised I’d tested the wrong wire on the ribbon. Ugh! Sure enough, it was also broken. This one, however, was in the middle of the cable, so there was no replacing that one easily like the clock. I had to make a decision. It was time to replace the whole ribbon. But with what?

I thought about this for quite a while, and eventually came up with a “cunning” plan. I’d put pin sockets on the board and I could connect the two halves with some veroboard and pin headers. Best of all the veroboard would make it incredibly easy to get everything neatly lined up.

Out came the desoldering station, and I began the long slow process of desoldering all 64 wires. 32 on each end of the ribbon. As I did so, several broke off hard against the board, more proof that it wasn’t a matter of “if they would fail” as “when they would fail”. I also decided to tackle the keyboard ribbon at the same time for the same reason.

It was at this point I made a horrible discovery. While the ribbon cable spacing for the 32 way between boards was at a standard 2.54mm pitch, the ribbon cable to the keyboard was much closer to 2.5mm. What fresh hell is this?

I decided to put that off for “later” and concentrate on the inter board ribbon. It went on reasonably easily and I was quite happy with the final result. It’s semi rigid and easy enough to remove for servicing the rest of the system.

The system now booted to the READY? prompt without difficulty or obvious glitches.

Time to tackle the keyboard ribbon.

The keyboard itself still had a pin connector on it that would accept the ribbon directly into it, so I decided to keep that side, and concentrate on getting the signal to it. I have a stock of silicone wrapped wire. It’s lovely and flexible, so it was an obvious target for using in this task. I cut 10 strands each of two different colours and stripped the ends. One end I soldered the wires directly into the main PCB.

I then took the old keyboard ribbon and cut out a section in the middle of the ribbon where it tended not to get folded very much. I stripped this too and soldered the silicone wires to it to make a connector to connect to the keyboard. I shrink-wrapped each wire as well for strength and reinforced the old ribbon with a piece of offcut PCB glued on. At this point I tested the keyboard and it worked. Now I was happy the design was valid, I also ran hot glue around the wires where they joined the PCB to make sure any twisting happened in the wire, not in the more brittle solder joint.

At this point, all that was left was to button her up, and make an external tape cable. Again, the Microbee forums were invaluable here and I was quickly set up, ready to test, using an iPad as the audio source…

It worked first go!

OMG!

Apple Mac Se/30

The mac had started glitching out with an annoying “sort of raster” pattern.

I assumed it was bad capacitors on the Analog board as they do have a bad reputation. I posted the photo on Facebook and two posters (The amazing Bruce Rayne and group admin Jeremy Barr-Hyde) suggested that I check for dry joints.

When I finally got around to investigating the Mac, I figured I’d do a visual inspection of the Analog board for two reasons. One: to check for leaky or unusual looking capacitors and Two: to check for dry joints.

While all the capacitors looked absolutely fine, I did notice some very suspicious looking solder points.

I chose to reflow all the larger joints on the board, especially around the connectors for the various power leads.

Plugging it all back in again, just in case, I was surprised when it immediately sprang to life. Huzzah!

IBM “Battleship” M122 Keyboard

This was another kind donation from Greg, who has always been an absolute legend of a supporter of both me and this blog.

He’d found it in storage and asked if I was interested.Boy howdy, yes I was interested! Buckling springs!

Greg and I met for coffee, and I collected the keyboard from him.

What a monster!

Now in the dying hours of my holiday, I figured I’d have a look at how hard it would be to connect this up to my PC and use it with Windows 11.

Some research quickly brought me to a neat mod called the “Soarer’s Converter” which looked to do what I needed. Best of all, it used a Teensy 2.0++ as its brains and I had one of those to hand! Looking through my parts bins, it appeared I had everything to hand so I thought “Why not?” and proceeded to quickly run one together. Looking at the documentation, it appeared to be as simple as connect connect a few pins on the Teensy to an RJ45 socket, connect a few LEDs to some other pins, flash new firmware to the board and connect it to my PC.

It came together quickly, and before long I had the whole thing assembled in a jiffy box I made from two different jiffy boxes I’d salvaged from older projects.

Not quite believing how easy it had been, I cautiously plugged it in and…

The lights came on. I tapped the capslock key and the capslock LED on my adapter also came on. OK, this was looking good. I put my cursor into a test window and tapped a few keys. It was fully working!

In fact, I’ve typed the majority of this post with it :-)

I still need to give it a clean, and have a look at the space bar stabiliser as it’s a bit flakey, but overall I’m very impressed.


Posted

in

by

Comments

2 responses to “Battling Entropy”

  1. Greg Howell Avatar

    Great to hear the keyboard works. And it’ll double as a load bearing beam in you house if you ever need one….
    A couple of very curious faults and repairs – nice work!

    1. Cheshy Chesh Avatar
      Cheshy Chesh

      Also suitable for sheltering under in case of nuclear attack :-D

Leave a Reply

Your email address will not be published. Required fields are marked *