A5000 woes

[RobC]

My A5000 has died

It was previously working fine but I tried to power it on today and, apart from the power light coming on, it seems to be pretty dead. No POST fault code (or other signs of life) from the floppy and the keyboard lights are usually off but sometimes all come on power up (and then stay permanently on). There's no output on the video either.

There was a tiny amount of battery damage when I got the machine but I neutralised it and repaired a single damaged track. I've checked the CMOS circuit and it all buzzes out, the battery is charged and there doesn't seem to be any other track damage (no signs of bare copper etc.).

I've only had time for a quick look on the scope and there doesn't seem to be much happening but the PSU output voltages are okay.

Does anyone have experience of this happening and any pointers of what to look for?

Thanks,

Rob

[vectorlight]

I had a similar dead A5000 recently with the usual battery damage. The fault with mine was the ACT74 (IC61?) chip that feeds the 25mhz signal was dead. I suppose if you have a scope, you could check the clock pin on the ARM3?

[SteveBagley]

Have you tried pressing RESET? Mine, oddly, will sometimes only boot if RESET is pushed…

Steve

[RobC]

Thanks both.

I have tried the reset button and I believe the 25Mhz clock was present at the ARM3 but I'll check again.

[lcww1]

Well, I too am wandering in an A5000 related vale of tears - I've got two A5000s (both with socketed MEMCs) showing similar behaviour to your machine. There are signs of life - both produce a video signal, and the keyboard leds light up on pressing reset, but neither machine will boot. I've reached an impasse with both machines. Recently, I checked all the caps using an ESR meter, but found that all the caps appeared to be ok. Most recently, I attached my logic analyser to the test connector (see: http://www.stardot.org.uk/forums/viewto ... 92#p197306) - I was hoping that I might find something related to the floppy POST error message, and whilst I did get an interesting looking signal on D(0), I can't interpret what I'm seeing. I don't have a scope at present, but even if I did, I'd need some guidance as to using it to diagnose my A5000s' ailments. It's frustrating, as I'm sure both of these machines could be got working - their PCBs look ok, and I've not been able to find any problems with broken or corroded traces, so far.

[danielj]

This might be helpful? http://pi-star.co.uk/riscos/A5000TRM.pdf

Also: http://chrisacorns.computinghistory.org ... 5000SM.pdf


d.

[steve3000]

Sorry to hear about your A5000. I've had exactly the same too, and posted about it a while back: viewtopic.php?f=16&t=9378

Exacty like SteveB has said above, my A5000 would intermittently boot up if you pressed reset repeatedly after switching on, but it got progressively worse and less responsive. I tried various things using my trusty logic probe and multimeter, but without a scope I was struggling to get any recognisable signs of life. So I retired my A5000 until I find new ideas and/or a scope... and it currently resides in the spare bedroom, under the spare bed.

I didn't try IC61 or replacing the 25MHz clock, but that could be a really good place to start, since anything along that front edge of the PCB is easily hit by the dreaded battery damage.

Last time I though about this properly I was considering it could be a dead CMOS RAM chip, as a really dead or missing CMOS chip can hang RISC OS before it gets to POST. I must have a go a replacing my CMOS chip... You mentioned battery damage - did you replace the CMOS chip at the time?

[RobC]

Thanks Daniel and Steve.

From the looks of things, I didn't replace the CMOS chip or crystal as it worked when I got it. From memory, it just needed a new battery and a charge. I repaired a damaged track but I don't even think it was ever completely broken.

The battery damage really doesn't look that bad but the CMOS chip would have been next in line after the damaged track. I'll replace it and report back.

One other thing I thought about was the hanging issue that affects A4s. This eventually resulted in my A4 being dead until I reflowed the IOC. However, I've tried reflowing all the main chips and that hasn't brought the A5000 back to life

[danielj]

I've certainly seen RPCs that have to all intents and purposes been dead to the world until the CMOS IC was replaced...

d.

[steve3000]

One other thing I thought about was the hanging issue that affects A4s. This eventually resulted in my A4 being dead until I reflowed the IOC. However, I've tried reflowing all the main chips and that hasn't brought the A5000 back to life

Out of interest, did your A4 give any POST error messages when it died? Mine is giving IOC register failed... I've assumed that it needs a new IOC, but wonder if it could be solved with reflowing.

[RobC]

Out of interest, did your A4 give any POST error messages when it died?

No - it started freezing more and more frequently and then eventually stopped booting with no POST error (or signs of life).

Definitely try reflowing as it fixed on fibble's too. I just applied some no-clean flux around it and then touched each pin with the iron.

If you do end up needing a new IOC, PM me as I may have a spare...

[steve3000]

Definitely try reflowing as it fixed on fibble's too. I just applied some no-clean flux around it and then touched each pin with the iron.

Thanks Rob, I'll definitely give that a try

[Andrewcee]

Probably not related but maybe worth checking the 'bleeding obvious'?

Fixed a couple of 5000s recently. Battery in one was a little fury but hadn't leaked, was working fine although now replaced. (Glad I kept it, now a temporary replacement in Risc PC) Swapped out cmos as post reported cmos & rom errors. With hindsight, believe cmos to be okay (kept original in case ever needed) as after short charge of new battery post reported only rom fault. 2nd 5000 arrives,... That's when I realised my mistake. Having removed pcb & throughly cleaned - white fur around links, etc - from memory links 10/11 (?), I'd fitted wrong way round, think they have to be north/south, have to lift lid to check. Anyway, swapped around & correctly fitted, no post errors and both machines working a treat.

Should say, found link error after several hours with scope, dvm, circuit diagrams, etc. Checked clocks, power-up delay, even confirming cmos being read on startup. Ditto all address lines from roms, tracing data lines I learnt each rom carries 1/4 of the 32bits, think it was, ie. All 4 roms needed (obvious statement). All this, and I'd fitted a couple of links incorrectly!

Good luck fault finding, both entertaining & frustrating. I'm working on a RISC PC trying to get IDEFS working.

[Andrewcee]

Forgot - cmos reads into two lines on I/O chip, difficult to find on circuit diagram. They toggle for maybe 1/2 sec on power-up. Although unable to 'read' the data coming from cmos using scope, I confirmed CMOS being read okay on power-up.

Hope this helps.

[lcww1]

but maybe worth checking the 'bleeding obvious'?

Always worth bearing in mind! It took me a little while to notice that the socketed 66MHz oscillator was missing from the non-booting A5000 alpha board that I’m trying to revive....... Unfortunately, merely adding an oscillator has not fixed the machine, so the quest continues...............I will eventually fix this machine, though I can see that it may turn out to be a lifelong project

[DutchAcorn]

A corrupted CMOS chip can prevent the computer from starting. Instead of replacing the cmos chip, you could just snip leg 8 of the chip (+5V) and see if it starts. In my experience the A5000 will boot to supervisor if the CMOS chip is removed.

The configuration you end up with a dead CMOS chip typically has the video set to "auto". Auto detection does not play well with modern flat TFT monitors (although it may work). No point in trying to "blindly" correct the configuration, it will not be retained. A CRT VGA monitor will eliminate that issue.

It can take up to 40 seconds before the computer shows the supervisor prompt. Don't switch off too soon.

If you want to check the two data lanes (pins 5 and 6 on the CMOS chip) you can check the connectivity to the IO chip but it may be easier to use pins 13 and 14 on the expansion connector (SK9, where the backplane plugs in). You can put a header in there and buzz out the connections. The tracks run mostly on the under side the PCB and have through-hole connections to the other side of the PCB. Relatively easy to find and to fix.

[RobC]

I've replaced the CMOS chip but, unfortunately, it's made no difference

I've also dug out my non-working A5000 Alpha so if the kids don't object too much, I'll try to get the scope out tomorrow and have a big repair session.

[steve3000]

I've replaced the CMOS chip but, unfortunately, it's made no difference

I've also dug out my non-working A5000 Alpha so if the kids don't object too much, I'll try to get the scope out tomorrow and have a big repair session.

Shame to hear the CMOS chip didn't solve it. Did you get any further with your A5000 repair?

[RobC]

Did you get any further with your A5000 repair?

Not yet - I've been trying to sort out a few other things for the show on Saturday but I'm hoping to take a good look at it soon.

[myelin]

Just received an A5000 in the mail that fails to POST... looked pretty clean inside until I took off the drive caddy and saw that pretty much everything to the right of the RAM, between the backplane connector and the front of the machine, is a familiar deep green colour :/

First disassembly issue: it looks like the IDE and floppy cable have corroded in place. Pulling on them doesn't do much. Any tricks for getting them out?

[daveejhitchins]

As you're going to have a good cleaning/patching job I'd try some WD40. Spray liberally and leave to soak-in.

Dave H

[myelin]

Looks like I may have overdone the vinegar washes... here’s the “before” photo:

And after:

Managed to get rid of quite a bit of verdigris, at least. Looks like quite a few address lines are disconnected, although the ARM3 is running — I see A3 and LA3 toggling; next step is to try reflowing the pins on the ARM3 to see if that brings any more back to life.

[daveejhitchins]

Managed to get rid of quite a bit of verdigris, at least. Looks like quite a few address lines are disconnected, although the ARM3 is running — I see A3 and LA3 toggling; next step is to try reflowing the pins on the ARM3 to see if that brings any more back to life.

Lots and lots of flux and a hot air reflow gun should do a good job.

Dave H

[myelin]

Lots and lots of flux and a hot air reflow gun should do a good job.

Hot air reflow gun ordered!

In the meantime I've tried reflowing with a soldering iron and flux pen; no luck bringing any of the connections back to life, although it bubbled nicely and I did see a small solder ball forming on my iron after 20 or so pins, so it's not in as bad shape as one of my A3000s

Buzzing out the A2-9 the connections to IC36, it looks like the vias near the ARM3 *are* connected, but the pads aren't connected to the vias, which gives me some hope that once I reflow it properly, they'll reconnect. Fingers crossed!

[daveejhitchins]

One thing to remember, with flux . . . it needs 'activating'! - I find, with stubborn joints: apply the flux and heat, allowing it do its job. Then apply soldering iron. This also helps to reduce thermal shock to the part being worked on.

Dave H

[myelin]

One thing to remember, with flux . . . it needs 'activating'! - I find, with stubborn joints: apply the flux and heat, allowing it do its job. Then apply soldering iron. This also helps to reduce thermal shock to the part being worked on.

I didn't know that! So I should apply flux, heat it a bit with the hot air gun, then use the soldering iron to reflow the individual legs of the chip?

With something like this problematic VY86C020, do you ever find yourself needing to actually desolder the chip, touch up the pads, then solder it down again, or is it pretty much always sufficient to just apply flux and heat without actually removing the chip?

[daveejhitchins]

With something like this problematic VY86C020, do you ever find yourself needing to actually desolder the chip, touch up the pads, then solder it down again, or is it pretty much always sufficient to just apply flux and heat without actually removing the chip?

Solder, over time, changes its structure! As part of a learning curve project, I had to complete while working in production engineering at Rank Hi Fi, I read a post office report on the long term reliability of soldered joints. It became very obvious why they went with wire wrapping! The former degrading with time and the latter improving with time.

The best way to proceed 'would' be to remove all parts and refit with fresh solder. But highly impractical . . . Although not applicable, in our case, it's interesting to watch how Louis Rossmann replacing parts in Mac Books (YouTube). He removes the parts then either wicks the remaining solder or just 'scoops up' with a soldering iron tip - then replaces with lead free. But if he suspects that the part is OK but may have a few 'bad' joints, he'll just add lots of flux and reflow.

So I believe that just adding flux and reflowing will be sufficient. Suppose I'm lucky, so far, as I've not had to deal with large surface mounted parts! But I've lots of Archimedes machines just waiting for me - I hear them chuckling away in my cupboards

Dave H

[myelin]

This has been sitting in the garage for most of the last year, and I had a burst of energy a couple of nights ago, so I got in and fixed some more of the bad connections. Buzzing out all the interesting pins on the ARM3 (everything except the lines that just go to the empty copro socket), I found that D30, ALE, DBE, and 8 of the address lines were disconnected. Most are now okay after very carefully soldering in a bunch of short pieces of enameled magnet wire. Luckily most of the failed pins just connected to nearby vias, which were themselves fine (as the bottom of the board is almost pristine, unlikely the heavily corroded top), but a few of the lines required wiring all the way up to the MEMC.

Aside from the bad pins on the ARM3, I can see that some of the pins on the buffers that generate LA2-LA21 are disconnected, so I'll probably need to wire some of them to the MEMC, rewire some of them to their terminating resistors, and rewire some of the terminating resistors to their vias that connect to the upper half of the board. There's all sorts of damage around the IDE and FDD sockets, but with any luck the machine will boot without any of that working. It looks like many of those pins connect directly up to the fine pitch IO chip, so repairing them will be about as much fun as repairing the ARM3 address lines, but if I can get the machine to start up at all, I think I'll be pretty satisfied. Crossing my fingers and hoping that there aren't any broken tracks around the memory chips!

[flaxcottage]

That's what I call a labour of love. Good luck.

[myelin]

That's what I call a labour of love. Good luck.

Yes indeed... I'm determined to resurrect this machine

Spent some more time on it last night, adding tons of little jumper wires (wiring PHI1 from MEMC to the three latches, reconnecting their enable pins, fixing 25 bad tracks on the latch outputs, and wiring A18-21 from MEMC to IC46), and now LA2-21 all appear to be working, probed from the 5th column ROM socket. On powerup they all count up at 6MHz for about 0.7 s, which I assume is the power-on reset period, then start to do something different. I'm not seeing any video output; I'm going to probe the PCF8583 and Vidrq* (VIDC) lines next to see if it's getting far enough to talk to either chip.