by Pure Amiga recently gained a new machine to the collective collection, in the shape of an Amiga 4000. It formed part of a bulk collection made by Phil, and although I’m lucky enough to own one it was still a gap in his retro portfolio so space was made, it all worked first time and that’s the end of this post.
If only, eh?
The real story is that the original owner stopped using it one day in the late 1990s, presumably when the price and capabilities of a PC became too tempting and he, well I suppose he would say he upgraded but I’m sure you can make up your own word. Let’s just say this machine is probably more desirable than the one that replaced it, 25 years later! Whatever happened, the 4000 was relegated fully working to the loft where it sat for over two decades. Phil bought it, tested it and found it not to be working, and over it came to the Pure Amiga workshop for a bit of fettling.
Opening it up we find that the original owner spend a bit of cash on this, and although it was originally a 68030 model it now has a Apollo 4040 card. This is a nice find, as it means we also have SCSI capabilities which the desktop 4000 lacked from the factory. Elsewhere we see a CD-ROM has been added, and despite the sticker on the rear of the machine claiming an 80MB hard drive is fitted (presumably at the dealer), there’s actually one 540MB IDE drive and an Apple-branded SCSI drive too. Still, none of that is useful if it doesn’t work, and it also means this machine has a distinct lack of space inside with ribbon cables crammed in everywhere, and almost no airflow to either the power supply or CPU board. Something to think about with the notoriously warm running 68040 – but again, we’re getting ahead of ourselves. All this came out and was safely stored so we can start to look at what faults we have, and why.
The faults we’re experiencing are a yellow screen on boot, and a power LED that remains dim. These aren’t particularly useful symptoms – a yellow screen simply means “an error occurred before error-handling was in place” and really means there’s something fundamentally wrong with the basic hardware – memory, clocks etc – that caused an issue so early on the Guru routines weren’t yet running to display that familiar black and red screen. The other fault with the LED not going to the higher intensity a second after power-on generally means the CPU isn’t responding, but at least on the 4000 you can sometimes ignore that. More on why, later! We can’t really do much more in the way of testing but the Apollo 4040 card was swapped for a known working A3630 card – the CPU board it would have originally had from new – and the extra fast RAM on the board was removed to check these weren’t causing issues. Off came the daughterboard too, the A4000 will work quite happily without that and it makes things really easy to get to. None of this changed the yellow screen fault, however.
The first thing to notice is this 4000 has fairly severe battery damage. There’s plenty of information on why this happens all over the internet, suffice to say it needs cleaning up – and is more than likely responsible in some way for the fault. Certainly there’s no point diagnosing further without cleanup and making that area good – if it fixes things, great but if not certainly we know we have a good base to work with. Let’s look more closely at what we’re dealing with….
As you can see, there are darkened areas on many of the copper tracks, or “furry” looking areas. Both of these mean the alkaline material that’s leaked out the battery has crept down the copper tracks and lifted the solder mask (the green coating), and started to corrode the copper underneath. It also means that a wash over with IPA doesn’t help much, the solder mask forms a “roof” over the damage so even with a good scrub the corrosive material is still there, and will still be doing damage. What we need is something more drastic, which is a fibreglass pencil that has just the right level of abrasiveness to take away solder mask, but it won’t take away the copper too quickly. Here you can see the effect of going over those darkened areas, with bright shiny copper revealed. In other areas however taking away the solder mask showed a distinct lack of copper, it had been totally corroded away. We still needed to find this and fix it, but sadly it means more work for us.
The other pain in the bum is that those tracks have a pesky habit of going under other components, and where the tracks go the damage follows. This means to do a good job we need to start lifting components off and inspecting underneath, and we always want to do a good job so out came Mr Hotair Rework, and Mrs Desoldering Pump to play. The trick here is to lift anything that might have damage underneath, and a quick look shows that includes all three surface mount chips on that front edge, the clock chip (the larger one next to the yellow trimmer) and even the front SIMM socket. Soon it was looking fairly drastic, but it’s more important to nip this in the bud so we don’t have to go down this same route in a few years time.
Don’t worry! It’s all fine. But it’s also the end of this instalment, so tune in next time to follow the fault fixing process.
