A1200 rare fault

[jltursan]

I'm trying to diagnose and fix a rare fault in an Amiga 1200 rev.1D1 and it's driving me nuts.

 

Testing the machine with Amiga Test Kit v1.20, all starts with the music module not being played at all (it happens also to a bunch of games, others play music fine) and the CIAB timer tests faults pointed out by ATK, it shows CIABTA, CIABTB and CIAB TOD IRQ as faulty.
The test displays exactly this screen:

 

Loss of int6 at even cia

 

So, a naive approach suggests that INT6 signal is missing and the CIA is bad as appeared in the net in pages like this (being CIA-A/INT2 this time):

 

A1200 CIA Timer Failure

 

As I've already socketed U8 (Even CIA), I've replaced it with at least 6 new CIAs, some of them sourced from working motherboards (I mean, at least the CIA subsystem...) and none fixed the errors.

The socket seems fine, I've checked the following:

 

• Socket is in good shape, no cracks.
• Every CIA-B pin seems correctly soldered and shows continuity with the rest of the components as shown by the A1200 schematics and Amiga PCB Explorer.
• No bridges between neighbouring pins.

 

AFAIK, INT6 signal is generated by CIA-B itself and then injected to Paula (U3) and Gayle (U5). There's only a passive component (R954A) in the signal path and seems fine, measuring 1K. If I run the ATK music test and check pin 23 (INT) with an oscilloscope, I can't read anything at all, no signal, 0V.

So, I'm wondering the following:

• Could it be that all my CIAs, even the ones working as CIA-A are faulty and can't generate INT signal?. I know that CIA-A INT signal must be a working IC as ATK tests passed OK 100%.
• Are any other signals involved in generation of INT6 by the CIA-B and some are missing or out of spec?
• Could a faulty Gayle or Paula pull down the signal and making it disappear?. I can try isolating the CIA pin; but I hate cutting tracks...

Any help will be appreciated...

Edited August 13, 2023 by jltursan

[Daedalus2097]

The interrupts are all active low, so if you're finding that the line is always low, there's a definite issue here. Since you've socketed the chip, what happens with that line when you power it up with the CIA removed? It won't boot obviously, but the INT line should still be high. The interrupts are open-collector, so the line should sit high and only be pulled low when something requests an interrupt. R954A is a pull-up, so even if it's measuring fine you should check it relative to the 5V rail. E.g., with the power off, check the resistance between pin 23 of the CIA and 5V at the floppy power connector. Also, if you've socketed both CIAs, you could swap it over and if INT2 is working with the suspect CIA in position U7 then you know the CIA itself is working fine and the issue lies elsewhere. Pin 23 on U7 can also be used as a reference for how the signal or pull-up should look as the circuitry in both cases is practically identical.

 

Yep, the even CIA (U8) is the main source for INT6 on the motherboard, but that interrupt line is also available to other hardware on the trapdoor expansion port and on the internal clockport, and can also be driven by Gayle for PCMCIA hardware (IIRC the IDE port uses INT2). Are you testing with no other hardware attached at all, just the bare motherboard? Any signs of bent pins or damaged contacts on those expansion connectors?

 

If the line is still pulled low with the CIA removed, and the pull-up is working as it should and has similar resistance to ground as INT2 at U7, it could indeed be a fault with Gayle or Paula. It's not a fault I've seen but anything's possible, and difficult to isolate without removing chips or cutting traces. Paula is the easier of the two to remove, and is the one actually responsible for handling interrupts, so if it came down to it, that's where I'd look.

[jltursan]

Very good points, thanks!...in fact, I've missed that indeed INT6 is active low!...

 

It's a bare A1200 motherboard, with PCMCIA, ClockPort or expansion connector in good shape.

 

I've checked the resistance between pin 23 and 5V and indeed, it shows 1K while powered off. Same value against ground and also both values are mostly identical as the ones measured in U7.

 

EDIT: So, cutting a track, I've isolated Paula from the INT line and now, the machine boots erratically, some times it sees the floppy others just hangs; but the couple of times it has boot the ATK, the CIA timers tests are mostly correct. Now I have CIAB TOD IRQ test passing OK and CIABTA/CIABTB failing with wacky values, far from the 21/22 expected. Also during the music test can be heard some very faint noises, something is trying to going on.

Oh, well, that's something.

 

Paula smells...I'm going to try socketing and replacing it.