The TFW8b SX-20 - Portable VIC20

On April the first this year, there were many dodgy AI generated April fools posts.

Two posts that were produced without the use of intelligence of any kind, artificial or otherwise, were as follows:

And

That one was a fake, created by me in photoshop (by which I mean GIMP, but it seems to be accepted as a generic term, like "I googled it on Yahoo" or "I hoovered with the Dyson")

This took an entirely unjustifiable amount of time for what it was, but never mind, at least I didn't get cancelled.

The other was also obviously a fake, a Commodore SX-64, the portable C64, with a fake VIC20 screen showing on it.

Wasn't it?

No, it wasn't a fake.

It really is an thing.

Something TFW8b has been working for many years (with occasional input from me).

He even cleaned the screen for the video.

(well, sort of)

And covered over the place where future secret upgrades are to follow.

It has been a spare time project since before the pyramids were built. (well, since about 2018 which feels almost as long ago, when the world was a very different place)

Of course it has taken rather a long time as neither of us has much in the way of spare time to idle away on a frivolous pet project (or usually PET project in my case).

TFW8b is currently documenting the last 8 years work as this project might actually be drawing to a close.

• https://www.tfw8b.com/please-be-standing-for-the-sx20/

Keep checking that post as it will continue to be updated until it is complete.

I will be writing a few posts on areas where there is more to explain, like the S-Video mod I covered last week, applied in a slightly modified form here.

• http://blog.tynemouthsoftware.co.uk/2026/04/vic20-s-video-conversion.html

In this post, I will expand on the power input side of things.

Powering the SX-20

It is important to say that no (good) SX-64's were harmed in this process.

This was not a mint, boxed, low serial number SX-64, signed by the entire cast of the A-Team.

For reasons of taste and decency, I will not include any more pictures of the donor SX-64. Let's just say it was in bad shape.

Someone had been through it and bodged and broke everything. Lots of wires were cut and some rejoined, connectors bodged, plastics cracked etc., and bits of the monitor board were rattling around inside.

It was bad.

It was also 120V, although I think the monitor was PAL. This really was a collection of broken parts someone made into the shape of an SX-64 and listed on ebay.

Which makes it the ideal candidate to become the SX-20.

Rather than having to use a stepdown transformer, or trying to modify the power supply (which looked to be fried anyway), it has been converted to 12V in.

With the power supply removed, there was a bit of space at the back where the power supply was.

Big enough for, oh I don't know, three large ex-UPS 12V lead acid batteries. 3.6Ah in total.

This is 0.0432 KWh, and just for comparison, an electric London bus battery pack is 400KWh.

The SX-20 is a step up from the SX-64 in that it is actually fully portable. No need to snake a mains lead out to beside the pool when you want to use your SX-20.

To make that work, it needs a new power board.

Version 1

This was the original version.

It worked, it did the job, but was a little sketchy.

Beautiful composed photo of load testing. It was Christmas break, so of course his test load was on theme.

(this was many years before the testers he recently built)

• http://blog.tynemouthsoftware.co.uk/2026/03/tfw8bcom-low-voltage-dc-power-supply-tester.html

Ready to install.

I won't say anything about the wire colours, I will leave it to your own judgement to rightly conclude they are wrong.

That fitted in a space inside the SX-20 case.

But wait, there is loads of space behind that board.

Ah, perfect, just enough space for a Penultimate Cartridge.

To help explain what is going on, TFW8b helpfully drew a schematic.

I think that explains everything you need to know.

Version 2

That's more like it. The "less flammable version" from January 2024.

This does the same job, but with a revised design and implementation.

This uses three DC-DC converters with built-in displays to neaten things up.

Here showing the input voltages.

The schematic for this one is a little closer to the implementation.

I will redraw it to explain what is going on.

The SX-20 works in two modes, on battery and with 12V DC in.

To simplify this (a little), I will first draw those out separately.

SX-20 Battery mode

When running on batteries, the SX-20 needs power for three sections:

1. The main VIC20 board (and accessories such as the Penultimate Cartridge)
2. The monitor
3. The 1541 disk drive

These are powered as follows:

The battery isolator switch and the main power switch both need to be on to use the SX-20. Congratulations, you have invented an AND gate.

The main 5V rail for the VIC20 and accessories is generated from the first DC-DC step down converter.

The VIC20 is almost entirely 5V DC, the only time it uses the 9V AC input is for the userport, and to generate the 6V DC for the datasette motor. 

Neither of those are an issue here as they didn't fit when the photo of the VIC20 board was cropped into the photo of the empty case. (oh what a giveaway)

The 12V to the monitor is switched, to allow the monitor to be turned off to save power.

The disk drive power is also switched, with a separate 5V DC-DC step down for the drive's 5V rail.

The third DC-DC is not used.

I have drawn those as simple switches, but they are actually posh illuminated ones.

SX-20 12V DC in

When 12V DC is supplied externally, things are wired as follows:

Here the power switch controls the main 5V, the monitor 12V and the disk drive 5V and 12V, as before.

The third DC-DC converter is a step up to 14.4V to trickle charge the 12V batteries via the charge resistor(s). The battery isolator switch can be used to disable charging.

SX-20 Automatic Switching

The actual circuit used includes a relay which switches between those two modes.

The relay is powered by the 12V DC input. When that is not connected, it routes the battery to the power switch and disconnects the charger.

When 12V is applied, the relay connects the 12V input to the power switch and connects the battery to the charger.

Photos of the unit installed to follow, once TFW8b reaches 2024 in the photo archive.

More to follow I am sure, including things like the custom ROM I made for it, and I expect an upgrade to a PU+3 DCR.

So keep a check on the TFW8b post for more updates as he works his way through 8 years of pictures and maybe even some videos.

• https://www.tfw8b.com/please-be-standing-for-the-sx20/

---

Adverts

For reasons currently unknown, the Tindie website is down, so I will instead tell you to go to TFW8b.com and buy something there.

• TFW8b.com

(SX-20s are currently out of stock)

---

Patreon

You can still support me on Patreon, assuming their website doesn't go down before you read this post.

There I am currently working on the new Mini PET, with lots of development posts if you want to follow the progress.

• https://www.patreon.com/tynemouthsoftware

Commodore SX-64 Repair and 1541 Diagnostics

This is an old post, preserved for reference.

The products and services mentioned within are no longer available.

It seems recently I've had a run of repairs which have turned out to be fairly straightforward, some like last week's C116 repairs are worth writing up as they are interesting machines, the rest are the same old usual suspects on the same old machines, so I've not written any of the up. This one however, qualifies as it's another interesting machine.

The Commodore SX-64, a Commodore 64, 1541 and a monitor, all in a box with a handle on. What Commodore described as an 'executive computer'. Something the busy executive has someone lug around for them so they can do their work by the pool.

This one isn't going to help our executive file his expense claims, it's showing nothing but a grey screen. This is in fact a black screen, with the brightness turned up.

To double check this, there is the same 8 pin DIN monitor connector as the Commodore 64 on the back, so plugging in an external monitor shows the same, a black screen. The disk drive is not spinning up either, so there is something more than just a monitor fault. A quick check with the Dead Test cartridge also showed a black screen, so time to look inside.

Looking on the back, it looks like quite an early one, late 1983, but I'm not sure how long these were in production for.

Inside the SX64, you are presented with the inside of the monitor as well as the computer and disk drive side, so as ever, use a little common sense and don't test voltages on the monitor with your wet fingers.

Leaving the monitor side well alone, it's a little difficult to access the rest, but my first test was to measure the supply rails, and the 5V rail wasn't there. Something was pulling it down and probably getting very warm. I didn't leave it on for too long, to check for hot chips, instead I dismantled it a bit further to gain access to the boards.

The first board, the one that sits along the whole right hand side of the unit, is the CPU board. This contains a large part of the Commodore 64 side of the machine. There is a nicely heatsinked VIC II chip, ceramic by the looks of it, and it's associated circuitry.

There is 64K of DRAM and it's mux chips. These are a common cause of that sort of thing, and you would often find one chip baking hot, shorting out one of the supply rails. I hope it's not that, as it will be difficult to test without making up some extension cables.

In the centre are the main chips, the CPU, the PLA, the SID and three ROM chips. The standard C64 BASIC and character ROM, and a modified SX-64 KERNAL in the centre.

There is a board plugged into this at right angles, this contains the IO side of things, the two 6526 CIAs, and various other miscellaneous parts including the 7406 IEC buffer. The round silver can top left is a 60Hz oscillator that is used to generate the time of day signal which is normally derived from the AC mains. That is not a cheap part, and they could have used mains reference on the SX-64, perhaps there was some plan to run the SX-64 from batteries?

To rule a few things out, I thought it would be easier to try these chips one by one, in a known working C64 board. I skipping the 2564 EPROM for the moment, as that is a non-standard pinout.

The ROMs, the PLA and the CPU, and all were found to be working. The machine had arrived for repair accompanied by a PLA chip on some foam. This was apparently the original chip, and has been removed by a previous owner in an attempt to fix the problem. I also verified that it wasn't working.

Whilst I had the C64 board on the test bench, I thought I would test the SID, make sure that was working, then I could leave it out of the system until I had finished testing. But when I tested the SID on the C64, I got a black screen. I measured the 5V rail and it was low, I touched the chip and it was very hot. Oh dear, it seems the SID was dead, and was causing the black screen.

I reassembled things without the SID and it booted straight away. I ran through the Dead Test again, and all passed. I think it is possible that the previous faulty PLA might have caused the SID to burn out, or it could just have overheated itself.

I'll switch back to the external screen captures at this point, as they are easier to see. I'm running a modified version of the Diagnostic ROM at this point, which knows about the SX-64. The Fails on the IO chips are due to the missing SID and the lack of loopback plugs.

So that's all working nicely. Back to BASIC, and interesting to note seeing these side by side that the SX-64 KERNAL uses the same colour scheme as the Dead Test and Diagnostics cartridges, rather than the less readable blue on blue colour scheme that the standard C64 uses

The keyword was initially a bit unresponsive, but after a bit of use, most of the keys have come back into operation. Could probably do with a full cleanup if any of those don't come back.

Time to test the disk drive. The light comes on, a good start.

It seeks, but doesn't read anything. I wasn't sure that this point if the drive was at fault, or the IEC circuitry, so I disconnected the IEC bus from the internal drive (the two six pin connectors are the IEC bus, and are wired in parallel).

There is a user port and IEC port in the same style as the Commodore 64, so I plugged in a standard SD2IEC drive and tried that out. (Note there is no cassette port on the SX-64, so you need to specify the Userport connector when ordering).

That worked fine, so the problem was the drive.

The drive inside the SX-64 is pretty much a standard 1541 drive, rearranged to fit in the SX-64 case. It's not very easy to access though, as the drive mechanism is blocked by the 'glove box' that fits above it.

I tried a few tests using the 1541 Test/Diag cartridge (by World of Jani, available from TFW8b.com). This has a good selection of utilities to diagnose issues with Commodore disk drives, which all run from the cartridge.

I started out with a few tests, but it was clear it was failing everything.

It does look like you have to remove everything to access the drive, but a handy tip is to unclip the two studs on the front sides of the 'glovebox', this allows it to be flipped upwards.

With that raised, you can access the head and the rails of the disk drive mechanism.

The head is on the bottom side of the disk, and there is a sprung clip above you can lift to access the head. I've propped it up here to get a better view.

Oooh, nasty.

That might explain why it couldn't read anything.

After a good clean, it's looking a lot better.

I also used some Molykote lubricant on the drive rails to allow them to run free.

Back to the 1541 Test/Diag cartridge to repeat the tests. I started with the performance test, that starts with a format and then various read and write exercises.

That all passed, so this is looking good.

I also tried the alignment test, that showed 100% on track and not too bad on the half track reads.

You can also do all sorts of fancy stuff such as the sector viewer.

The last test I did was the speed test, which was also within acceptable tolerance.

I noticed when putting the machine back together, that the strobe marks on the bottom of the drive flywheel are visible through the slot for the keyboard connector, so you could do it that way if you prefer.

The drive appeared to be working, but I wanted to do some further testing, so I again unplugged the IEC bus from the internal 1541 drive and plugged in an SD2IEC. Testing that using the Epyx Fastload to save typing, it responds to commands to the default ID 8.

I wanted to use the two together, but both start off as ID 8, so I will change the SD2IEC drive to ID 9. To change the ID that, use the following command: 
OPEN1,current address,15,"U0>"+CHR$(new address):CLOSE1

This will last until the power is removed. To make it persist, use this command: OPEN1,new address,15,"XW":close1

Note the drive is already drive ID 9, so you need to send this second command to the new address.
With the SD2IEC now as drive 9, I reconnected the original drive, plugged in the Epyx Fastload cartridge and booted up CBM Command to give the drive a work out.

This identified the two drives correctly, and was able to copy files back and forth.

I also tried writing a D64 image from the SD2IEC to a real floppy disk in the internal drive. That takes a while, but completed with no problems.

This was obviously for testing purposes only....

Oh dear, I can't possibly play this without a SID. I had sourced a replacement SID chip, so time to put everything back together.

To prevent further issues, I fitted some heatsinks to the SID and PLA chips.

Time to try that again. Much better, sounds great.

With everything all back together, I thought I'd finish the testing with the new Ms. Rodman cartridge.

That's running well, a nice portable games machine. Only slightly easier to carry around than a full size arcade cabinet.

Just clearing up the test bench, and I noticed, for a machine with an internal disk drive, I seemed to use rather a lot of cartridges in this repair.