Thanks to uBee and Brad, I now have all the parts to finish off my 16k MicroBee.
It’s a beauty! Yes, it’s yellowed. Yes, there’s engravings in the top cover. Yes, the keys look like the teeth of a chain smoker, but that is exactly the type of character I like in my systems.
Some quick notes:
The screws holding the top case to the lower case turn out to be an exact match for a bunch I purchased in bulk from Jaycar a few years back. I think they were the part number HP0414, but I’ll check and post here later.
The “case badge” was designed up in CorelDraw and printed onto silver paper. I covered it with clear contact, cut it with a roller cutter for straight edges and then applied some workable fixative spray glue. This kind of pipeline (CorelDraw, Laser printer, Roller Cutter, Workable Fixative) has become my “go to” to make surprisingly realistic badges and labels for a whole heap of systems. They’ll never pass anything more than a passing inspection, but they do look better than nothing at all.
In my last big order from a ubiquitous Chinese e-commerce site, I ordered FIVE Gotek drives, earmarked for a variety of systems, including my Amiga 1200 (when it gets built), my Amiga 600 and the Atari STᴱ.
I had a cunning plan, and one of the reasons I’ve been so quiet recently was that I’ve been working on the plan.
You see, one of the things I don’t like with the Gotek is, by default, it has only got a 3×7 segment display, and a pair of buttons to run things. I wanted upgrades in the form of a nice big OLED display and a rotary dial. But where to put them? When I’d been working on the Archimedes, I’d had the clever idea to mount the screen on a cable, and that seemed to work well. I wanted to take that to the next level, so I’d been thinking about the design for a while.
The long linear connector of the design for the Archimedes took up space and was not very elegant. A better design was needed. I knew I only needed 7 wires.
“Seven you say? Surely the rotary encoder needs 5 and the OLED needs 4. Isn’t that nine?” Aha! you’ve been fooled! The GND and +3.3V pins are shared between the Rotary encoder and the OLED, and the Rotary encoder actually only needs 4 pins not 5.
I decided on using a DuPont connector for the computers and Pin Header for the OLED Display. I settled on a pin being removed to make sure the connector couldn’t be plugged in backwards and from there a 2×4 row of pins was the logical size.
I sat down with the Wiki and carefully plotted out my pins. Then I looked at them again and rearranged the pins completely to be more logical 😀
Grabbing out the crimper I made up the socket in about 40 minutes. Note to self, next time use multicoloured “rainbow” wires. They’re a lot easier to keep track of what pin goes where.
At this point I made the other half. This consisted, on the connector side, of some experimenter’s board, some pin header and a zip tie for strain relief. I soldered on some nice flexible 8 core cable and soldered on the OLED and rotary encoder to the other end.
At this point I tested everything and it worked as planned.
I’d already flashed the Goteks with FlashFloppy, but I needed to get some more pin header, with 2 rows and a right angle, as if the pins stick straight up, you can’t close the case. A quick trip to Altronics fixed that. Now the connectors could lie flat inside the case.
Next I 3D printed two different styles of bracket (one for the Amiga 600 and one for the STᴱ to hold the DuPont connectors in place. A dab of super glue and the whole things were done on the computer side.
Next up I printed up a nice case for the OLED and the Rotary dial. It’s almost good enough for release. I need to tweak some dimensions but it still looks good.
Finally, having accidentally shorted two wires and severed a third, I resoldered the connector, covered it in hot glue, then a piece of cloth that I’d impregnated with hot glue, then finally gently heated the whole thing so everything hot glue bonded together. THis gives me an incredibly strong connector I can easily remove.
The Archimedes continues to confound me. I finally got a replacement 1772, plugged it in and… No change! At this point I have swapped out all the chips that feed the floppy disk. If you look at the schematics, I have socketed and tested IC29, IC30, IC38, IC 46, IC47 and RP1. No change. It’s going to be continuity testing from here on out, which won’t be fun.
Some of the upgrades I recently ordered started flowing in at last as well.
I got an ElkSD128, which brings the Acorn Electron up to an insane 140k of RAM and also adds both a Joystick adapter AND a SD card interface. This makes the Electron a little powerhouse!.
I got a TFW8B SD2PET Future which finally adds easy to use storage to my Commodore PET. Files at last! (I promptly loaded a few games to muddle around with. I do like how well documented this was.)
I also got a set of five Gotek FDD emulators which I have converted to FlashFloppy firmware. So far I’ve only installed one in my Amiga 600, with a nifty external rotary encoder and OLED. (I also have some 3D bits and pieces to make it up into a nice little package. I’ll be covering all the Gotek stuff in a separate post in the next few weeks.)
Finally, I got a DIVMMC Enjoy Pro. Unfortunately this isn’t working. The DIVMMC isn’t displaying its proper BIOS screen. Next step was to clean the edge connector, but that didn’t help. (I also reflowed the power connector because I was seeing some unreliability.) As this still didn’t help, and the next step involved getting a new Z80, I decided to completely resolve the edge connector as being a potential issue. The Spectrum doesn’t have fancy gold plated edge connectors. It has a simple solder plated edge connector. This was looking a bit “well loved” so I took the time to flow and clean each and every edge connector “pin” on the edge connector. My methodology was pretty simple. I dropped some fresh flux across the pin, then ran fresh solder onto the pin, using the hot soldering iron tip to “scrub” the underlying corrosion patches to remove them, finally following up with solder braid to bring them back to a smooth “factory finish”.
Alas the problems persisted. I was a bit depressed by this and was concerned I’d need to buy a new Z80 just to test. Then I remembered I actually had a spare Z80 in my collection. Out came the desoldering gun, and with a little effort I was able to desolder the z80, drop in a new socket and test.
Unfortunately the system still failed to load the DIVMMC BIOS. I’ve reached out to ByteDelight for more diagnostics hints. Hopefully it won’t need to go back to them. It’d be almost be cheaper for them to send me a new one.
I caught up with Brad from the MSPP forums for a coffee (Hi Brad!) and we chatted for a good solid hour and twenty minutes. Brad had very kindly donated me the missing key for the 16K Microbee. It’s keyboard is complete! I have a potential top and keyboard cover coming, which would nicely finish this system.
I’ve also started work on an evil project to build an adapter that will allow me to use a 27c2001 ROM in the place of one of the ROMs in the PC85 Microbee as a switchable 16 ROMs. Silly but fun.
I also got some PCBs. The biggest one was a full set of boards for the Omega MSX2 which should keep me amused for a few weeks :-). I also got an amplifier I designed as well as twoboards designed by Necroware.
The third Microbee was now nominally “working” insomuch as it was booting reliably to the command prompt. Only thing is, I couldn’t test anything further without a keyboard.
The keyboard was partially working, and having removed the keycaps I could see a mix of 3 different keyswitches. White stemmed with a horizontal bar, Black stemmed with a horizontal bar and Black stemmed with a vertical bar. Without exception every single vertical barred keyswitch had failed.
Thankfully I wasn’t the first to have to fix these. Microbees have a certain reputation for keyboard failure. There is a great article on exactly how to fix them.
Out came the desoldering station, and without too much effort, I was able to get 39 of the keys out without difficulty. A small handful resisted so I left them for later. Conveniently, one of my sets of pliers was exactly the right width to grip on to the top of each keycap and remove them once desoldered.
My procedure was pretty simple. First, mark which keyswitches need desoldering with a marker. Desolder with the desoldering station, check the pins were loose, and if both pins were loose, gently lift from the frame with the pliers. Some needed an additional “suck” with the desoldering gun but most only needed one pass.
Later, I would remove the remaining keyswitches by reflowing, desoldering and in one or two extreme cases, simply heating the solder joints while gently tugging with the pliers. Usually only one pin was still stuck so that got the recalcitrant switches.
Once all the switches were out in a batch taking them apart was a combination of terror, brute force and careful levering.
The procedure I adopted (Based on the article above) was to jam a thin flat bladed screwdriver down one side, next to the pins and lever the plastic away from the body, then jam it down besides the other pin. Finally wiggle the screwdriver sideways and down so it forced the middle area between the two already abused sections apart. Once done, you could force the screwdriver gently in further, then lever the back off the switch. At this point if everything is going right, the switch would come apart into it’s component pieces. If not, then either the plastic cracks or the back sproings off and the spring flies off, never to be seen again. Trust me, I speak from experience.
Thankfully I only lightly cracked two switches and I found the spring eventually.
Each switch consists of a body, a base of two gold plated contacts, a spring, a keystalk and a small carbon embedded contact on a strip of rubber.
Once apart, I cleaned the two contacts with a soft ink eraser followed by the ever popular isopropyl alcohol. I then dragged the contact gently across a sheet of paper once or twice., followed with a dab of isopropyl. After that came reassembly followed by testing. Using a multimeter and a pair of alligator clips I tested each switch after reassembly. One or two needed a second pass, and maybe adjustment of the contact positions by gently heating them with a soldering iron. I could then straighten the pins in the softened plastic and reassemble.
Before cleaning, most had a resistance between 2KΩ and 10KΩ. After most were around 200Ω with arange from 400Ω to 80Ω. Basically a tenfold improvement in conductivity.
Once all the first wave were done (All 39) I reassembled a single key and soldered it in. Testing it worked perfectly! OK, the other 38 went back in. One turned out to be a stabiliser for the shift key so I pinched the keystalk from it for one that was broken, and used the broken one behind the shift with a jury rigged spring to provide counterpressure. (I’m not entirely happy with this. I may swap out the spring for a different type in the future,)
At this point I removed the remaining eight switches (as detailed above) and cleaned and reseated them too.
There was a little excitement when the bodged in 33k resistor array being used as a keyboard pulldown snapped a leg, (It mounts across two of the screws in a very annoying fashion) but was able to rejoin it.
Finally, I was able to test all the keys, plus loading software. Everything now seems to work correctly, within the limits of the system. This is a 16k computer with an old version of the ROMs, so compatibility is somewhat limited.
Finally, I cleaned up the speaker and remounted everything in the lower shell.
In related news, some kind people are helping me source the one missing keycap for the CTRL key, and possibly an upper shell for it! Awesome!
For anyone trying to follow along at home, last we saw the Microbee computers, I’d managed to make the cable for the Microbees, and proven the base 32k “Communicator” model was working fine. I’d resurrected the PC85 and upgraded it to an 85b model complete with a bunch of embedded games. Now back to the last of the Microbees, the 16k early model.
Straight up this was missing the “A” ROM, so I needed to arrange to get one. This unit used the rarer 2532 ROMs. I played around with the idea of getting some 2732s and making adapters, but in the end I found an eBay seller selling them so I ordered and waited… and waited and waited. Approximately 2 months after my order, they finally arrived. They looked a bit scabby but that was OK as long as they worked. Unfortunately I had no way of burning these models, but I had a friend who could help. I dropped them off to him and waited. Turns out both ROMs were duds. Neither would take an image. D’oh! Thankfully the seller refunded without argument (which makes me suspicious). The friend was able to find a 2532 in their parts bin and was able to burn me the ROM.
Now, at this point I have one good confirmed working ROM but that’s it. I plugged it in and turned it on and… Nothing. Not even video synch.
Hmm. Not the end of the world. I was kind of expecting this. Next step is to clean all the ROMs and the legs from the top “Core” board to the lower “Main” board. While doing this I noticed that the last row of pins on the “C” ROM were completely missing. I can only assume at some point in the past, someone had inserted the ROM incorrectly and accidently sheared off the two pins, and then hadn’t noticed when they put it in correctly. This was easily fixed by simply soldering on some replacement legs made from offcuts from a resistor.
Aaand this time we get a loud “BEEP!” from the speaker. Alright! Still no picture though.
I checked up with the various Microbee forums and someone confirmed that a beep means the CPU has started and is running code. Excellent! Now to look at the picture.
At about this point I bumped the board and suddenly got an inverted picture with a barely visible cursor. That’s promising! I’d occasionally get a screen full of garbage. At this point I knew what to do next. I replaced all the ROM sockets, the socket for the 6545 video chip and the connectors between the base board and the top board. I also cleaned the legs of every single ROM and the 6545.
I try swapping the 6545 for a known good one and still no picture. Hmm.
Oh wait! I haven’t plugged in the video cable.
Now I’m getting a picture! A little grey but a picture nonetheless! Keyboard is rubbish but not unexpected. Let’s put it in the case and… No picture… What?
OK so something was obviously loose or failing here. I knew it wasn’t the chips as I’d swapped them over. I knew it wasn’t the sockets as they had been replaced. I pulled out the schematic and had a look at likely candidates. Next in the chain was a resistor (R20, an 82Ω resistor) or the socket the video plugged into. Both were a bit “crusty” so out with the desoldering gun and out they come. Replacements went in and… no improvement. Oho! It’s going to be one of those kind of jobs. By this point I was having to go to Jaycar or Altronics more than once a week, just to keep the forward momentum going.
I actually broke out my BitScope at this point and started looking for video signals, starting at the composite connector and working backward. At the composite connector there was absolutely no signal. At the connector side of R20, there was no signal. At the other side of R20, nearest the transistor (TR2, not labelled properly here) I could see a video signal.
It took me a while to realise the signal was at a way too low voltage of about 0.3v before it went into R20. It was supposed to be closer to 1v peak to peak on the other side of R20 and it’s not going to go up at all. If I jumpered around R20 I’d actually get a picture. Could it be the amplifier transistor at TR2?
OK so on the forums everyone thinks that TR2 is the likely suspect so back to Altronics I go. And that makes no difference at all. Gah! I’d also swapped R18 out as it was out of spec. I’d also reflowed every single solder joint in the area and gone over the board with a magnifying glass, just in case there was a short.
Could it be the Capacitor? Hmm.
At this point, someone on the forums mentions that I must have a very old Microbee as it sounds like the video fix has never been applied. Their Microbee didn’t even have a capacitor at C20.
OK it turns out that MicroWorld had issued various fixes, and one of the BIG fixes was updating the video circuit with a bunch of changes. Diving into the “Microbee Hardware Notebook with Updates”, on page 63 was a complete change to the video circuit. What the hey? I have all the components on hand. Let’s give it a go.
Guess what? It worked!
Nice stable picture, no issues.
Huzzah! Interestingly, once I got C20 (the Capacitor I was suspicious of) out of the circuit, I tested it and it was showing only a few nanoFarads. I’m curious if the circuit would have worked “as is” with that replaced, but I’d much prefer a “fixed” circuit. The picture is wonderfully crisp.
The next step will be to work on the keyboard. It’s an interesting mix of three different styles of keyswitch. Desoldering has commenced.
My long term plan is to take the nicest looking of the three and see if someone is interested in trading for a “premium” disk based model, with me paying the difference. I’d really like a disk based model. Stick a Gotek in and I’d be right to load all the software I could want, without having to rely on tapes!
A huge thanks to everyone on both MSPP and the MicrobeeTechnology forum for all their help and support. You folks rock!
EDIT: See the new final paragraph for some updates.
When I got the Amiga 2000, the initial fault reported was “Dead Power Supply”, but as it had been in storage for a long time, I suspected a secondary fault of “Battery damage” may have joined the original fault. The Amiga 2000 has an onboard rechargeable battery that, after 20+ years will ooze a green alkaline solution across whatever it’s near and proceed to eat it.
As such I opened the Amiga with some trepidation. As I went, I made an interesting discovery, based on a throwaway comment on the Perth Amiga Users Group. It turns out what I have is a much earlier Amiga 2000. One of the original “German” systems. These are notably different insomuch as rather than being a “big box” version of the Amiga 500 desktop system, they’re the “big box” version of the original Amiga, aka the Amiga 1000.
This leads to some limitations in design that impact what can be done with these systems. Firstly, the 1MB of RAM is 512K on the mainboard and 512K on a riser in the slot that later models use as a CPU upgrade slot. Secondly the chipset is very much stuck in OCS territory, and it uses the older DIP format, the same as the Amiga 1000. Thirdly the Video slot is much truncated compared to later versions. It really only has the same signals as the rear 23 pin connector. Fourthly the Zorro slots are unbuffered, which means what upgrades will work is severely limited.
Onto the cleaning. I removed all the cards in the card slots and the memory riser. This is where I found something interesting! The power connector had been connected incorrectly. This may be the source of the reports of power supply failure. EDIT See final paragraph.
I removed the drive/ power supply sled and the extent of the battery damage was revealed. It did not look pretty in there. The CPU had green legs and there was a circle of about 5cm of corroded components.
(For those watching from home, this is where I discovered all the photos I’d carefully taken were all blurry! WTF camera?)
Removing the battery was… interesting. The corrosion has a particularly bad reaction with solder to render it into something closer to a ceramic. A mixture of metal oxides and other crud.
Initially I whacked on some of my good flux, tried adding some fresh solder, and using the desoldering station with the widest nozzle. The first conductor came clean straight away. The second conductor needed a quick refresh with some additional solder to clear the rest of the way. The final, most corroded pin that was attached to the groundplane? It sat there and sniggered. I added more solder, which pooled and blobbed on my soldering iron. I added more flux, which rapidly turned into fancy smoke, making no change to the joint. I scraped away the top layer of brown gunge with a small jewelers screwdriver, added a more aggressive flux, added more solder until it had a bead on it, and then hit it with the desoldering gun and… it finally cleared!
A little pressure from the other side and the battery popped clear.
Right let’s inspect…
OK so I have some work in front of me clearly.
Initially I hit the affected areas with white vinegar. This has the effect of stopping the corrosion from continuing by neutralising any remaining alkaline. I gave everything a scrub with a toothbrush at the same time.
I then rinsed the whole board in tap water to wash away as much of the vinegar as I could.
Out came all the socketed chips for inspection, a quick spray with isopropyl and, if needed, a further cleaning. Most were fine but some needed attention with an ink eraser. I’m still not convinced the 68000 will be OK. if the alkaline has crawled up the legs into the innards, it may be an ex-processor.
Next I gave the board another scrub down, this time with distilled water. I then liberally applied isopropyl and put it out in the sun to dry, turning to make sure both sides were getting sun. I redosed with more isopropyl and repeated.
Amazingly, the board didn’t look as bad as I had initially expected. I took to the traces around where the battery had been with a fibreglass pen (I hate these things. They drop glass fibers everywhere which are a skin irritant. My solution is to put a piece of paper under what I’m working on and throw the whole sheet in the bin at the end.) and cleaned up everything including surrounding vias.
Everything seems to still be intact. It’ll need a new solder mask in a lot of places, but all things considered, it’s not to bad at all. There are two capacitors that literally only exist as leads now, but I’m not sure if that was caused by the shorted rail or by the corrosion or a combination of both.
For now, it’s going in storage until I have significant time to look at it. The corrosion should be mostly neutralised, or at least dramatically slowed. In the new year, I’ll drag this out again and begin the slow job of seeing what needs further work.
EDIT: Thanks to Stewart Greenhill for this Amiga. (I never post names without permission first) He has mentioned that the offset rail on the PSU was post the PSU blowing up so was actually not the root cause. It’ll make this repair more interesting. I suspect my easiest solution will be to replace it with a modern PSU, plus a “Tick” circuit. (Amiga 2000 has a clock signal synched to mains frequency, the “tick”, that it uses for video timings.)
Yes, you read that right. I just got another computer. Yes I know what I said about peripherals. The TF1260 counts as a peripheral and I hope to be ordering more in the next few days. Meanwhile, I got offered this so how could I say no? Thanks to Stewart Greenhill for his prized Amiga!
This is an Amiga 2000, the first of the “big box” Amiga range. It’s not working and I’ll need to organise a keyboard replacement as this one is lacking that crucial peripheral. The kind gentleman who gave it to me also gave me the 1081 monitor it would have been paired with initially, as well as some key cables, a mouse and some disks.
This unit also includes a Janus PC Bridgeboard, that allows the system to run PC compatible software, albeit, very slowly. It also has an MFM HDD, I’m guessing of around 20MB of capacity. I won’t hold my hopes out for that after 35+ years.
Possibly most worrying is that the original onboard battery is still onboard and the motherboard shows signs of corrosion. I’ll need to strip it, and give it the old 1.2.3. of Vinegar, followed by Distilled water followed by Isopropanol.
The monitor has a bung switch, but that is typical of these monitors, and I already have a replacement switch somewhere I can use to swap it out.
All in all, a nice pickup, although I suspect it’ll be quite the project to get from here to the “insert workbench” screen.
My last question though is “Where the heck am I going to put it?”
To give it some context to the PC centric heathens out there in the audience, this is like going from a 286 with 2MB of RAM to a Pentium with 128MB of RAM.
It’s quite some upgrade. The TerribleFire cards are the result of work by an extremely gifted electronic engineer by the name of Stephen Leary. They’re open source and I sourced mine from RetroKit.
I currently have it working but there was some effort to get it there. Have I mentioned my Amiga 1200 is not in a case? I do have plans to make a custom cut case for it. (Would anyone be interested in an article about that?)
First up, when I initially plugged it in, it was a decellerator, as the Amiga refused to boot. This was fixed by applying some deoxit to the edge connector followed by an ink eraser.
Next up, while the machine would happily boot into a plain workbench disk, it wouldn’t boot from the HDD. It would throw up Guru Meditation errors. After some great advice from Matt and David on the PAUG forums, I installed the MMULib and had it happily booting to an Amiga 3.1 Workbench.
There was a little more mucking around until Matt pointed out I could bypass the warning message on SysInfo to actually run it with an ‘060 in there.
Why SysInfo, you may ask? Because it’s the gold standard for irrelevant benchmarks that prove nothing and are of dubious value. And in that spirit here are the before and after shots.
That’s some upgrade! Hopefully by the time I get around to building the case, I’ll be able to use the onboard IDE port to connect to an IDE CD-ROM I have earmarked for the system. At that point I will truly have the Amiga of my dreams 😀
So it’s been a frustrating week of stuff not getting fixed. The kind of week where the hobby feels more like a “job” than “something fun”.
Firstly the Archimedes is still having problems with the floppy disk, despite me swapping out the 74HC574. I will need some time to put it on the bench and properly investigate the root cause. In the meantime, the Archie will have to sit in the naughty corner, or, in this case, the plastic tub of shame.
After that failure, I decided to recap the Amiga 600. This went surprisingly well. I recently got some more tips for my TS100. These were a KU tip and a JL-02 tip. The former is like a knife blade, and is great for drag soldering and anywhere where lots of heat is needed. The latter, on the other hand, may be one of the most useful soldering tips I have ever used. It’s a long, delicate tip with a curved end. It’s not great for heavy heat transfer, but for delicate work, it’s amazing!
There were four capacitors wedged hard up against plastic fixtures, that I had been unable to desolder. With a little foil and polyimide tape, and the long thin tip, I was able to successfully solder all four caps without any grief. In fact, those were probably the two capacitors I was most happy with.
The rest proceeded uneventfully, except for one cap being in backwards. (Always check your polarity before you turn on. I did, and spotted the offender before I turned on, thus avoiding blowing it up. I quickly removed it using the two iron “Chopstick” method and resoldered it.)
Once it was fully recapped, I powered it up and success! I had a system patiently waiting for a disk.
I had ordered a IDE to CF Adapter so I thought “Let’s see if I can get this installed?”.
This is where things started to go wrong.
I initially set up the CF on an USB to IDE adapter, connected to an Amiga emulator, and after some false starts, I was able to get this booting happily on the emulator. As soon as I tried to get it to boot on the real hardware it would simply be ignored.
In desperation, I cannibalised the Gotek out of my BBC Master system, and tried installing directly from Workbench images directly from the A600 itself. While I could recognise the CF card and partition it to my heart’s content, I could never get it to actually show the drive on my desktop so I could format / boot it.
I tried two different CF Cards. No luck. Eventually Matt on the PAUG forums pointed out there was a bug in the version of the Kickstart ROM I’m using that makes it incompatible with hard disks bigger than 40GB. Huh. The smallest CF I have is 64GB, and that one isn’t seen as a Hard Disk. I guess I’ll need a new Kickstart ROM.
All in all, a frustrating week, despite the success with the recapping.