I’ve been working on the Archimedes

So for the last week I’ve mostly been concentrating on something else but I have spent some time on the new Archimedes. It’s an Archimedes 440, one of the original family, but it has some upgrades installed. Most noticable is the CPU has been upgraded to an ARM 3 processor, and the OS ROMs had been upgraded to RISC-OS 3 as well.

First steps first, I needed to inspect what I had, and let me tell you, this system has seen some horrors.
Firstly, one of the daughterboards (Lovingly called “Podules”) had a Varta battery on it, which had reached “End of Life” and had dumped it’s gust all over the Podule. It’s a writeoff, but thankfully neither hugely important, nor significant. It also doesn’t appear to have got onto the motherboard, so that’s a plus.
Secondly, the batteries for the main board (which were off to one side) were standard alkalines and had, at one point, rotted and split. The case has sustained extensive damage, but thankfully, so far, it’s been surface damage I’ve been able to treat with vinegar, followed by distilled water, followed by isopropyl alcohol.

There’s also some damage to the main board and this is where I have been concentrating my efforts, so far. There were some tracks on the top that looked “dodgy”, eg attacked by acid. I have cleaned off any discoloured tracks back to copper, cleaned, checked for continuity and covered with “lacquer” (Eg nail polish) to stop further oxidation.

There is also extensive discolouring of the solder on the underside of the RAM are of the board. I tried to reflow it a bit but it’s quite stubborn.

At this point I decided I needed to “bite the bullet” and plug it in and turn it on. I had already stripped it down to clean and desolder, plus I had tested voltages coming out of the PSU.
I was greeted with a red screen. According to the super helpful people over on StarDot, this is actually a good sign! It means the CPU is booting. Next step was to plug the floppy drive in and see what happens. Oddly I got nothing. Further investigation showed a PSU with a low 12v line. This usually means a short somewhere. StarDot to the rescue and I carefully snipped out one of the capacitors (C37) which is apparently well known for going open circuit. The system can run without one smoothing capacitor for quite some time. Long term it’ll need a replacement.

So now I had a system that would start to boot then stop. The Archimedes has a very clever system of self diagnostics where it flashes the floppy light in a series of short and long flashes in sets of four. You can then feed these into a special web page that decodes the error.

As I wasn’t getting a full set of codes, I needed to do more work on the system. I cleaned contacts on the bottom of the RAM and cleaned up connectors on the CPU board and finally got it booting long enough to display a series of flashes.
They were:


Which, plugging into the POST error calculator gave me:
RAM control line failure
CMOS RAM checksum error
Self-test due to power-on
So we’re getting somewhere!

At this point I attempted to build a keyboard adapter that’ll allow me to plug in a PS/2 keyboard and mouse so I can at least get the past the CMOS error. It wasn’t too hard, and was based on a commonly available PIC processor so I had that built in a day.

I reassembled everything and tried it again and…

Nothing. Not even the red screen.

At this point I’m stumped. Something else has failed and it’s not even getting to POST. I’ll tackle the corrosion some more on the bottom of the RAM and see if I can get further but maybe I have killed it somehow?

The hardest thing is I don’t have anyone else in the state that can assist. These are rare systems and without a second system, I don’t even know if my keyboard adapter is working or not…

All in all, a bit of a frustrating outcome, but all part of the hobby I guess.

What a difference a week makes


So on Monday night (after my last post no less) I was feeling enthusiastic, so I stripped down the Mac and gutted out the PSU.
I then removed the switch.
Let me tell you that it’s not as easy as I make it sound with such a simple statement.

IEC at the top, in situ. Below is the switch. See that goop on the switch? That’s not happy-making goop.

Firstly to get to the switch you need to remove the IEC socket, which is soldered into place, and has a weird resistor and capacitor combo soldered across the mains. I had to desolder the wires and loosen a lot of the solder on the capacitor, just to be able to bend everything enough out of the way that I could remove it.

Ugh! This thing!

Next, the switch itself was held in place with this heavy solid glue. I thought it was epoxy, at first, but it melted a tiny bit when I touched the soldering iron to it. I was able to soften it enough to remove by hitting it with some hot air for a few seconds. I was then able to pry off the glue and carefully pop out the switch. It was an extremely tight fit.

Once the switch was out, I could disconnect the connectors for the mains. That freed it up. Going on some advice from the Aussie Apple II users group, I opened up the switch by poking “something” (in this case the tip of some needle nosed pliers) down into the hinge section of the switch case, allowing me to pop open the switch and…

More Eww

“Yep. There’s your problem”.

Thankfully Danny on the forum had already tracked down a replacement switch, which I have ordered, which should be here next week.

Meanwhile Greg (who has kindly donated many Apple things to me) contacted me to offer an authentic Apple IIgs keyboard he’d found in his collection. How could I say no?
I did offer him an Atari ST in return but he declined. He did, however, mention an interest in a Commodore 64, and considering his huge generosity in the past, I tried to return a part of the favour by swapping the keyboard for a refurbished C64c, joystick, multi-game cartridge and one of my scratch built c64 power supplies, as well as a cable. Hopefully this will bring him as much joy as his donations have brought me. Thanks Greg!

Keyboard for the Apple IIgs

But wait, there’s more! On the same day as I picked up the keyboard and dropped off the C64, a package all the way from Poland arrived. Aww yiss! It’s my next upgrade, the DDI5 for my Amstrad CPC 464.

So much potential!

This bad boy gives me a floppy drive emulator, similar to a Gotek, a second Floppy Drive connector and finally an additional 512k of RAM.

I had some initial problems, but, as I was able to quickly determine, this was simply because the connector was filthy. I ended up having to disassemble the system, clean everything with an antistatic brush, then polish the connectors with a sand eraser.

After that it stably booted and I was able to work out almost enough to start playing some classic CPC disk games. One demo I wanted to try wouldn’t actually work, and chatting on the Noel’s Retrolab Discord, Noel himself pointed out I was using the wrong command 🙂

(For anyone following in my footsteps, the correct command to load the demo was

As part of that discussion I got to trial his new CPC Diagnostics software, which demonstrated that, yes, my CPC now had 512KB. Awesome!

512KB is all that anyone should need, amirite?

Finally, yesterday, just before I was due to start my holidays, my newest Unicorn arrived! It was an Acorn Archimedes 440/1

So much potential. If only I can organise a Keyboard / Mouse.

As yet it’s almost completely untested as I simply haven’t had a chance to plug it in. I also need to organise a keyboard and mouse for it, as well as some sort of video solution. 15KHz monitor to the rescue!

More to follow, as I explore deeper.

Long Weekend Updates

So, dear readers, I have been keeping myself amused.

Firstly, Mac SE/30 is still dead. I completely stripped it and thought I had fixed it with some deoxit in the power switch, but it still arcs once it gets warm.
General consensus on the Australian Apple II users board is that the PSU switch is dead. I’ll have to strip the entire thing and remove the switch. Thankfully one of the board members has also had this problem and has found a replacement switch, so as soon as I confirm which switch mine needs, I’ll be ordering that and hopefully finally getting the Mac in a state where I can continue my explorations.
My plans for it are as follows: Get it back reliably. Install System 7.1 and 7.5 on different partitions. Get the CD-Rom that Greg so kindly gave me working, so I can play with CD software as well.
In the meantime, I got a friend (Hi Shane!) to print me a special bracket and now have the SCSI2SD mounted in the slot for the PDS expansion. This means I have access to the SD card and the USB setup port without having to open the system.
It looks so neat as well!

Looking down on the SE/30, before reassembling it. The SCSI2SD is in a printed bracket in the SE/30 external PDS connector spot.
The view from the back. All access, all the time.

Once I had that off the bench, I was able to get to the Atari STᴇ systems. I fired up the first one (Labelled as “4MB”) and the lights on it lit up immediately. An encouraging sign. After a few (long long) seconds I was looking at the GEM desktop. Huzzah!
I then set up the second one (Labelled as “1MB”) and it also fired up first time.
Inspired by my success, I decided to investigate whether the machines had the RAM that their labels claimed. This turned out to be a bit more difficult than I initially expected as I needed to work out how to get software onto a floppy disk for the Atari ST.
Some reading suggested that I could format a floppy disk in Linux, but I wasn’t having much luck.
As a long shot I simply copied the programme onto a preformatted 720k DOS floppy disk, and stuck it into the system and it worked! This also revealed that both systems have slightly gummy floppy disks so I’ll have to clean and lube them somewhere down the track. They did indeed have the RAM printed on the stickers on the front.
Conveniently, STᴇ systems use 30 pin SIMMs, of which I have lots left over from upgrading the SE/30 to 20MB, so I tried upgrading the 1MB system to 4MB and, yes, it now has 4MB which is nifty IMO.

Long term, I’ll need to get two ACSI2STM drives, a couple of Gotek FDD emulators and possibly a single external FDD cable made up so I can plug in an internal FDD as an external drive. This will take time, if only because those 14 pin connectors are as rare as hen’s teeth. Wonder if I could make a little PCB that would plug in with the pins preconfigured? It wouldn’t be as pretty but the 14 pin plugs are unobtanium.

You know how I said I wouldn’t be getting any new Computers this year?

About that…

Well, I’m the “proud owner” of an Acorn Archimedes 440 in a non working state. Looking forward trying to diagnose THAT one. This was an eBay auction. I didn’t expect to win. Imagine my surprise…

I also got two Atari STᴇ computers! Presumed working, but I haven’t had a chance to test them. (My workbench is covered in dead SE/30 at the moment). These were a gift from a friend who knew his parents were trying to give these to a good home. It included a 1MB STᴇ, a 4Mb STᴇ and a monitor. The monitor has a worrying rattle so I’ll need to investigate further before I turn that on.

The haul!
Mac in pieces. Since then I’ve also removed the Analog board and the PSU.

Recapping the VIC 20, plus bonus SVideo.

I had a bit of spare time yesterday and I wanted to do something, anything to distract myself from all sorts of “stuff” going on at the moment.

I figured I was well overdue to recap my VIC 20. The picture from it was pretty awful, so I was hoping it would fix it.

What you can’t see… All the annoying shimmer.

Thankfully the VIC 20 has a grand total of 6 electrolytics, and I’d ordered replacements ages ago and just needed to find them again. Diving into my big bin-o-components, I was able to pull out all 6 needed capacitors.

I fired up my desoldering gun, desoldered the first side of the first capacitor and “brrrz”, the soldering gun jammed. Gah!
It does this quite regularly, so I know exactly how to unclog it, but it’s still frustrating each time.
This is a ZD-915, and they are a little prone to blocking. If anyone else finds theirs jams, here’s my solution

  1. Turn off the unit and let it cool.
  2. Remove the glass solder collector. Probably time to empty it anyway.
  3. Turn on the unit again, and get a soldering iron. Set both to about 360°C and let them heat up
  4. Insert the cleaning rod in as far as it will go. Mine usually blocks at about an inch left to go.
  5. Coming out of the back of the desoldering gun into the glass solder collector is a small metal tube. This is what is blocked. Gently touch the soldering iron to the back of the tube while pushing on the cleaning rod.
  6. After a few seconds, the cleaning rod should push the blockage out and you’ll be able to clean the tube thoroughly.
  7. Turn everything back off and let it all cool.
  8. Reinstall the glass solder collector. Start everything back up and continue desoldering.

So 30 minutes later I had the desoldering station cleaned again, and removed all the other caps. Some are beneath the RF shield, so you’ll need to pop that off too,

“Dead” caps. Probably mostly fine, but after 30 years, the electrolyte aint at it’s finest.

Once they were all gone, I started replacing them, one by one, making sure to observe polarity and make a nice neat job of it. The new capacitors, despite being exactly the same rating as the older ones, were all marginally smaller. Thankfully no “surprise! incorrect polarity!” on the motherboard markings, but I had made a note of polarity before I started, just in case.

And the video was exactly the same.

What you can’t see is all the annoying shimmer, still.

Now I had been planning, for a while, to modify it for SVideo, as the picture improvement was supposedly much better. Unfortunately I needed some bits-n-pieces and I didn’t have the car so I had to wait until today to get to the store and buy the parts.
I needed an additional capacitor, resistor, DIN 5 plug and a mini DIN 4 (SVideo) socket.
For cable I used some leftover USB lead I’d salvaged, as well as some figure 8 cable for the audio. I also had a RCA socket spare.

Firstly I modded the VIC 20 end. That was as simple as desoldering a ferrite bead and a capacitor. (The cap looked suspiciously like a resistor). You then cut two tracks and wire across a fly lead. Finally you replace the ferrite bead with a 100nF cap and the “capacitor” with a 75Ω resistor. This went smoothly, although I replaced my fly with a thicker wire later. I’d originally used kynar again, but, as we’ll see, I had problems, so I’d replaced it as a precaution.

The top side. The mod is just above the chip in the picture.
The underside. the flylead is visible. At both ends is a cut trace.

Onto the video cable. Checking carefully, I soldered everything together, checking at the end pf the process, that everything was on the correct side. Glad I did! I’d managed to get the soldering on the DIN 5 completely backwards. That would have been bad, as one of the pins carries 5v and would have been running that into the audio lead. I neatened everything up, insulated leads with a mix of electrical tape and shrinkwrap, and buttoned everything up. I prefer to use shrinkwrap, but sometimes, there’s not enough space.

The video cable. I use sockets as it means I can have a lead from behind the TV and don’t need to reach around to plug / unplug anything.

Plugging in the cable and turning on the board produced… not good video.
Unplugging, I started inspecting my work. I noticed that the original mod had a component in one of the pictures that had been positioned slightly different to that in the original article. I tried moving mine to the same location as the picture, and… it worked. Boy howdy did it work. It’s crisper than my C128D.

Not present: Any annoying shimmer.

I tried all my games and all produced a lovely picture. About the only issue I see is some are producing slightly off synch and sometimes the picture is not centered on the screen. I suspect this is an artifact of the VIC 20 or the TV, not of the SVideo mod. I’ll try a different TV shortly.

A classic game. Looks great!

[EDIT] Clockmeister has helpfully pointed out that apparently that centering issue is a feature not a bug. Apparently cursor keys will allow me to adjust screen position. I’ll have to try it out.

[Not Retro Related] An idea for a new business concept.

So while almost all of the things I write about here will be related to retrocomputing, occasionally I’ll post something here that doesn’t quite meet that criteria.

So the other day, while thinking about several different topics, including “Pen’n’Paper” RPGs (Specifically Cyberpunk), Music videos I had been watching, and the impact of Covid-19, particularly in the US, where they look unlikely to ever bring it properly under control, I had an idea for a new business concept.

It was the intersection of several ideas, that coalesced into one solid vision.

The technology for this is almost all available “off the shelf”, with only one or two “new” bits that probably can be sourced with minimal effort.

What I am envisioning is virtual, distributed concerts.

The key “new” technology would be a specialised screen and optic imager.

Let me explain.

Firstly, why do we need this? Well, with interstate / international travel curtailed for a large part of the population, but with some areas “safe” due to local immunisation efforts, there will be demand for local “live” entertainment.
Artists will want income, but don’t want to expose themselves to myriad potentially localised versions of Covid.
Finally, there’s a whole bunch of live music venues lying fallow, that need a boost.

So here’s the vision:

Take a venue, rip out booths / stages from the back of the space. Fit in a specialised Display at the back of the venue plugged into a central server (See below for details, as this is one of the clever bits). Add a suitable sound system and fit lights and LED systems that can be controlled from the central server.

Meanwhile, set up “studios” that match the size of the back of the venues. These will have a virtual LED background, similar to those being used in shows like “The Mandalorian”. The front of these stages will have the optic imager I’ll describe in more detail below.

This setup is then connected to 1 or more venues as a virtual concert. The artists perform on the stage and it’s transmitted to the venue, with venue lighting being controlled by the remote Studio, so that the illusion of it being in a single location is enhanced.

Onto the technology of the screens and the imager.

These two technologies are designed to be paired. They are both dependent on each other, but work in a 1 imager to many viewers arrangement.

The imager, which we’ll cover first, consists of a wall sized grid of tiny RGB sensors arranged around lenses to capture light with directionality, so that each “pixel” on the sensor can see straight ahead, a bit to the left and a bit to the right. Embedded between each imager pixel is a small RGB LED, shielded so it doesn’t interfere with the imager.

This is looking down from above on three Imager “Pixels” with 3 sensors each. Each cluster of 3 RGB sensors (Labelled “2”) are behind a lens at an angle to each other. The LED is in a channel (Labelled “1”)

The captured information from these is then transmitted to the Display in the Venue.

The screen is, similarly arranged with directionality, but has RGB LEDs set up to shine out the corresponding direction as the light was received by the imager. This gives a “pseudo 3D” effect that makes the stage look deeper than it really is in real life. From a few paces back, the illusion should be that of standing in the same room as the artists.

This is looking down from above on three Display “Pixels” with 3 RGB LEDs each. Each cluster of 3 RGB LEDs (Labelled “1”) are behind a lens at an angle to each other. The sensor is in a channel (Labelled “2”)

There are also simple RGB sensors in between each display element. These are sent back to the studio and displayed on the embedded LEDs. While not providing the pseudo 3D that the audience gets, it does mean the Artists aren’t performing to a black wall, as they will be able to see the output from a venue, presumably chosen by whoever is controlling the broadcast in the first place.

Some obvious advantages (Or disadvantages depending how you view things):

  1. A performer can perform from anywhere in the world that has a studio.
  2. A performer can simultaneously perform in multiple venues at once.
  3. A recording of the performance can be made, and played back later.
  4. Special effects can be overlaid either behind or in front of the performer with little to no effort.
  5. The producer has a great deal of control over the final appearance, as the venues are somewhat standardised.

Of course this leads to a terrible “One size fits nobody” generic feel to all the concerts, but if the alternative is no concerts at all, it’s a compromise that may have to be made.

Mac Woes, Creativision wins.

Macintosh SE/30

On Friday, another peripheral finally arrived, so I had hoped today’s post was going to be all about my experiences upgrading my Macintosh SE/30 into its final form. While I got a lot of upgrade done, unfortunately I hit a setback that will mean this system is out of action until I get a solid few days to fix a major fault.’

We’ll start at the beginning. On Friday, my much anticipated SCSI2SD adapter arrived. This is a magical device that allows you to use an SD card as a replacement for up to four SCSI hard disks. I sourced mine from the Eastern States, as there’s an authorised assembler / seller over there. (He also has a YouTube channel you might want to check out if you’re into classic Macintosh repairs.)

I waited for Saturday morning before starting as working on the SE/30 is always an adventure. The integrated CRT means there are some excitingly high voltages in there, and the shell is part of the structure.

I started by mounting the SCSI2SD in a 2½” to 3½” bracket, choosing to mount it inverted so the overall profile was much lower. I then grabbed a 2GB MicroSD card and an SD to MicroSD adapter. From there I plugged the SD into the SCSI2SD, and connected the SCSI2SD to my PC via a micro USB cable. I booted up the configuration software and started carving up the MicroSD into chunks. My plan is to run System 6.08, 7.1 and 7.51 from different virtual drives, so I carved out 4 temporary drives as a test.

Mounting hardware
That looks very neat

The software was VERY choppy, which should have been a warning.

Once done, I pulled out the 40MB HDD from the Mac and screwed in the SCSI2SD. While in there, I also gave the board a bit more TLC. This board has been recapped, but I’m not 100% sure I got rid of all the nastiness so I give it an inspect each time and a scrub with isopropanol in any bits that seem dodgy.

I also inserted my BMOW ROM-inator. This has been sitting pretty in it’s original anti-static bag for months as I really didn’t want to have to open the Mac more than absolutely needed. This seemed like a good opportunity.

There’s the ROM-inator, tucked inside. BTW, that battery is less than 6 months old. It’s not going to corrode itself yet.

With both in, I did a partial refit and the system booted up with the ROM-inator “one eyed Mac” logo, and offered for me to boot to ROM if I wanted. I did want, as this is a super fast way to get the Mac up to the point where I could run SCSI tools, before buttoning up everything. Once up I ran SCSI Probe and could see the extra drives. Yay!

(A quick aside here. I made the disks SCSI ID 1,2,4 and 5, leaving SCSI ID 3 free for a CD ROM at a later date.)

I screwed everything back together, plugged in my FloppyEmu and booted to System 6.08. From there I ran the installer and…
“No disk found.”
Okay then! Run the disk initialiser.
“Drive selection failed. Unable to locate a suitable drive connected to the SCSI port.”

Computer says “No”

Yay! Not!

I had a suspicion what the problem was, so a quick trip down to the local supermarket and I had a full sized SD card from a semi-reputable brand in my hot little hands. I unbolted the SCSI2SD and swapped in the new SD card. This time when I ran the configuration software it was blazing fast! Clearly the MicroSD wasn’t up to snuff. It was also amazingly quiet. (I understand why people swap out the fans. It’s the only source of noise. Maybe I should look at a nice Noctua fan or the like). I divided up the card and reinstalled.

I had a little problem reseating the case this time. The Analog board has a nasty habit of jumping out of its rails, at which point the case won’t close properly. I was also noticing a bit of electrical noise on turning on the switch. I put it down to a dirty switch.

This time the disk initialiser worked and formatted the drive…
Do you know how long it takes to initialise a 2GB HDD? I do. It takes about 2 hours.

Firing up the FloppyEmu again, start the installer and several disk swaps and I have a booting System 6 install. Yay! We’re making progress! I was also able to determine that the SCSI2SD will happily run off SCSI temination power, so I didn’t even need the

The FloppyEmu hard at work.

I formatted up the next two disk drives and rebooted the computer and then disaster struck.

Turning on the computer again from cold, I got an angry buzzing crackling from the back left corner. Whoah! I quickly turned it off and unplugged.

A visual inspection showed nothing, so the next step will be to pull out the power supply and inspect that. As that’s a major job (You need to pull the Analog board to get to the power supply, so remember all those scary voltages? You’re right in the middle of them).

I figure this will be a multiple day project so I have put the Macintosh aside until then.

(As an aside, I’d love to see a version of the BMOW FloppyEmu designed to replace the internal Mac Floppy disk. It could have a connector poking through the FDD slot to mount the OLED onto, and a couple of long tactile buttons poking out through the drive slot.)

Creativision / Dick Smith Wizzard

Meanwhile, back with the DSW, I have been busy finishing off the keyboard. I’d already done the schematic in KiCAD, so the next logical step was to fix the mistake with the key footprint and actually lay the keyboard out.

My initial attempt went well, but then I decided I really needed to refactor the whole thing. I’d laid it out as a grid as per the original controller. I decided that was actually a really bad idea, and proceeded to lay it out as a proper staggered row keyboard, similar to a “real” keyboard. I also moved the “arrow” keys around so that the left arrow was in the “backspace” key position and the right arrow was in the “tab” position.
I also added a Shift to both sides, made more room for the Return, and two Space keys at the bottom, so I could join them up and have something much closer to a “real” keyboard size.

I then began the grand routing exercise.

People kept wandering through and asking me what game I was playing. I’d reply by explaining the “rules” of the game, and pretending it was a game made it more fun.

Eventually after about 2 hours, I had it fully routed, without too many long traces and, more importantly, no Vias at all! The trick seems to be to change sides regularly for long routes, which allows other routes to go past. There are plenty of through hole components, so really, there should be no need for Vias. I was just glad I was able to complete it.

The final result is tiny! It’s so cute! It’s not really suitable in this form for “real” typing work, but as this is a prototype, that’s not really the point, is it?

Respect out to the original keyboard designers who did all this without the aid of CAD tools

I’ve uploaded it to my GitHub page.