My Original CD32 and the Repairs I Put Off
Pictures to come soon.
The final machine in this little repair pile is my original Amiga CD32.
This one suffered from fairly bad capacitor leakage. The audio was affected, and the board became notoriously unreliable. Back in December 2013, I bought a replacement CD32 motherboard, complete with the metal shielding and the bottom half of the case, for what now sounds like the absurdly low price of £16.
That replacement board fixed the reliability problems at the time. Oddly, when I later tested the original CD32 motherboard again, it worked. That is very on-brand for old hardware: fail dramatically, get replaced, then behave just well enough to make you question everything.
In January 2023, I upgraded the CD32 with a Terrible Fire TF330 accelerator, a CompactFlash hard disk, and an ESP8266 connected to the TF330 serial port for Wi-Fi network access. It became a much more interesting machine, although not without its quirks.
The Recap That Did Not Go to Plan
I later decided to recap the CD32, but my soldering skills were not great at the time.
I think I damaged the board by overheating it while using a hot air gun to remove the capacitors. After the recap, the image on screen was fuzzy and distorted, with waves moving through the picture. That was not the satisfying "fixed it" moment I had been hoping for.
Frustratingly, when I recapped the second CD32 motherboard, exactly the same thing happened again.
After more than a month of trying to work out what was wrong, I bought a new-old-stock CD32 board from Analogic. I do not seem to have a record of the price, which is probably for the best. Unfortunately, that board arrived faulty and only booted to a black screen.
To their credit, Analogic were very good about it and replaced the board as soon as they received the faulty one back. The replacement worked perfectly, although I do not know the state of its capacitors. They are probably still the original parts, which means that little job is sitting quietly in the future waiting for me.
The Surprise Video Fix That Was Not Really a Fix
Recently, I decided to go back and fix the damage caused by my earlier, less experienced repair attempts.
I tested both older CD32 motherboards using the original CD32 power supply and the TF330 riser card to provide RGB output. I did not connect the TF330 itself at first, just in case one of the motherboards decided to do something silly and damage it.
To my surprise, both motherboards worked and produced a perfectly stable video signal.
That was a very strange moment. Previously, I had used a CD32 power connector to Molex adapter with a PC power supply, so it is possible the strange distortion I saw back then was caused by that power setup rather than the motherboards themselves. I am still not completely sure, but I am very glad I tested them again before assuming they were still faulty.
Once I was confident the boards were safe, I connected the TF330 and booted into the system.
Saving the Worms Before More Prodding
Before getting too carried away with repairs, I dumped the CD32 NVRAM containing my old Worms save data, including the customised worm names created by me and my friends.
That probably sounds like a tiny thing, but it was one of those little nostalgic saves that I really did not want to lose. I used a newer utility by Saimo from the English Amiga Board, as the TF330 appears to interfere with the normal library used to access the CD32 NVRAM. That matches the problem I had with the standard tools.
With that safely backed up, I felt much happier about prodding around the boards again.
The Easier CD32 Board
I started with the board that had the fewest bodge wires fitted. This was the replacement board I bought in 2013.
That board turned out to be a relatively easy repair. It mainly needed a good clean and several loose capacitors replaced. Some capacitor pads had been lost, including one on C334 where a trace ran directly through the pad from both sides. That made the repair a little more fiddly, as I had to use copper wire to restore the connection properly.
The other capacitors with missing pads could be soldered directly to adjacent traces and then secured with UV solder mask. I am sure someone with more experience could make it look prettier, but it works, and at this stage I am counting that as a result.
That board now works well, although it could still do with a soapy clean and a thorough dry to remove the remaining solder flux residue. Earlier on, I had tried to repair it using very gloopy solder flux, which did not clean off easily even with IPA. It is amazing how one bad choice of flux can make a repair look ten times worse than it really is.
The Original CD32 Board Is the Interesting One
The other board is my original CD32 motherboard, and it is proving more interesting.
From the missing audio chips, U15A and U15B, it looks as though I previously scavenged them for the other board. I now have replacements for those parts in stock, so that should at least be fixable.
I also noticed that the pads for Q342 were completely missing, along with the component itself. Q342 is a SOT-23 PNP transistor in the CD32 audio and headphone output section. Q365 also looks suspect. That one is a SOT-23 N-channel MOSFET used in the audio and headphone switching circuitry. The upper pin appears to be missing its pad, and a bodge wire has been fitted to what looks like the right audio phono output pin.
I have ordered a replacement for Q365 as well, because it looks as though part of the component leg may be missing. At this point, I would rather replace the suspicious parts and know what I am dealing with than spend hours chasing a fault caused by a half-attached leg.
Just from looking at this board, it is clear that it has suffered quite badly, either from capacitor leakage, earlier repair attempts, or both.
I am not as hard on myself for making this board worse. At the time, I felt awful because I thought I had destroyed two fully working CD32 motherboards, and I was honestly ashamed of it for a year or two. Now it feels like the right time to go back and fix them properly, with better tools, better materials, more experience, and slightly less blind panic.
Ready for Pad Repairs, Mostly
I have removed all of the capacitors from the original CD32 board because several looked as though they had been heat damaged by the hot air gun, and others simply looked suspect. Some smelt fishy and showed signs of leaking underneath again, which is never a smell you want from a motherboard.
The exposed pads have now been cleaned, and I have removed the large amount of sticky flux left behind from the earlier repair attempt. The board is ready for pad repairs using a mixture of new copper pads made from copper tape, new traces made with copper repair wire, and UV solder mask to hold everything in place.
I am not yet sure how good the audio will be once this board is repaired. I also cannot remember whether I removed the audio chips while trying to fix an existing sound fault, or whether I removed them while trying to repair damage I had caused myself. Either way, it should be an interesting repair, especially with the Amiga Test Kit and my oscilloscope.
Fixing the Damage, Not Rewriting the Past
So that is where my CD32 repairs currently stand. One board is already looking much healthier, and the original board is finally ready for the sort of careful pad and trace repairs I probably should have done years ago.
This is not really about building the most upgraded CD32 possible. It is more about undoing some old damage, saving the original console where I can, and learning from the repairs that did not go so well the first time around. The TF330, CompactFlash setup and network access make the machine far more useful than it ever was originally, but the board underneath still deserves to be electrically sound rather than held together by old flux and regret.
None of this is meant as a neat repair guide. The CD32 audio section in particular has already had parts removed, pads lifted, and bodge wires added, so my repair path is very specific to these boards. I will need to check the schematics, test carefully, and resist the urge to assume anything is fixed just because it looks better under the magnifier.
Still, after thinking I had ruined two CD32 motherboards, it feels pretty good to have both of them showing signs of life again. If my original console can end up clean, stable, and sounding right again, I will be very happy with that.