The Amiga 1200 That Fought Back: The Faults I Missed the First Time

I recently repaired an Amiga 1200 with a difficult to find fault. Unfortunately, it came back to me with more unusual faults that I didn’t discover the first time. So, here are the things I missed.

Story so far

This A1200 basically had an invisible break in one of the traces to Paula, which caused odd behaviour such as IRQ failures. In addition, it had been recapped, but most of the polarised SMD capacitors were backwards. This is the end result after I fixed and cleaned it.

The faults

When the owner took it back, he listed the following issues:

• The video output was pink.
• RGB to VGA adapters were not functioning.
• Audio with certain demos was badly clipped.
• Using a PS/2 mouse adapter doesn’t work.

All weird, since this board passed my usual tests when it left here. I was not able to reproduce the pink output in my workshop, but I have some ideas what happened there. I’ll get into that later.

RGB to VGA adapter issue

This one was weird, but had a couple of good clues. I usually test with an RGB SCART cable, which worked fine. But when using a VGA adapter, there was no sync. The exception was when this adapter was used:

I had a theory as to why this one worked, and looking at the schematics, my theory was confirmed. But first, I need to talk about video sync signals.

There are three sync signals coming out of an Amiga:

• VSYNC – Vertical sync, 50Hz here in the UK.
• HSYNC – Horizontal sync, about 15kHz on an Amiga.
• CSYNC – Combined horizontal and vertical sync.

SCART uses the CSYNC, whereas VGA uses separate VSYNC and HSYNC. So, my theory was that either the VSYNC or HSYNC signals were broken. This was confirmed when I looked at the schematics of this adapter. That large switch on the left of the image, that switches between the HSYNC output to the VGA being generated by HSYNC or CSYNC. It is set to the CSYNC position.

This means that the HSYNC to the RGB port was likely broken. Sure enough, with a few seconds of probing, I found this trace in purple to be bad.

That is the HSYNC signal, and putting a tiny patch wire there fixed the issue.

PS/2 mouse issue

I usually test with regular Amiga mice or trackballs, and these were working fine, but when a PS/2 adapter was used, nothing worked. Not even a left mouse click. This is important because the left mouse click is routed directly to one of the CIAs, separate from the rest of the signals.

At first I thought that this was maybe a power issue, there was a power issue, and I will get to that, but it was not related. The adapter also worked fine on port 2 of the Amiga.

What I did next was to probe each pin of port 1, testing for the pull-up value, to see if a pull-up was causing issues with the microcontroller. Sure enough, I found that on pin 1 there was only about 1 KΩ of pull-up, this should be around 4.7 KΩ.

Pin 1 goes to E355R, which is a 68 Ω resistor and E355C which is a 6800 pF capacitor. It then goes to R370H for the 4.7 KΩ pull-up.

E355C is connected to ground on the other side, and there is a small resistance between GND and +5v. Desoldering this capacitor showed that it had failed near short-circuit. So, this was replaced, and the PS/2 mouse started to function again. My guess was the pull was so hard that it was crashing the microcontroller.

Audio problems

I usually use the Amiga Test Kit’s audio test to check audio functionality. This sounded normal to me. But the owner tried a high quality demo with a lot of base, and it sounded terrible.

Whilst I have done a infrared test on this board, I figured I’d do another one. With this one I spotted things.

You can see that some areas that aren’t chips were slightly warmer than the surrounding board. This is an indication that ceramic capacitors have started to fail short circuit. I then tested the resistance between 5v and GND. It was 4 Ω! I’d expect this to be in the 30-60 Ω range. In fact the resistance was so low that my LCR tester thought there was a dead short between 5v and GND.

I ended up replacing four ceramic capacitors using hotspot and resistance testing, all were near total failure. With this in mind, I checked the resistance around the AV power circuit. This provides power to the Paula audio output using a voltage divider. The results were not as expected. Both C3 and C305 are bad.

We are now at total of 7 bad ceramic capacitors, but, everything stated to measure correctly and the audio sounded great in demos. I even used my PiStorm32-Lite to play an MP3 to test it. It sounded spot-on.

Conclusion

First of all, I think the pink video output would either have been related to the capacitors causing a power issue, or a video cable / connector issue. It is likely the green signal was missing. I tried a couple of different screens and was not able to reproduce it.

Here’s the learning experience for me: I have to change up my testing. I normally do a power resistance test to spot for failing ceramics and I didn’t on this board. I also didn’t spot capacitors getting slightly warm instead of hot. In addition, I need to improve the audio tests I do to include audio that will definitely clip badly in a similar scenario.

I’ll be using new additional testing techniques for future machines that come in to make sure such things are spotted early.