Dead A3010 help... PSU maybe?

Arc/RPCs, peripherals, RISCOS operating system & ARM kit eg GP2x, BeagleBoard
steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Dead A3010 help... PSU maybe?

Postby steve3000 » Sat Jun 03, 2017 12:43 pm

After some thoughts/pointers on this.

I have recently been repairing a seriously battery damaged mezzanine A3010 (probably the most badly damaged computer I've tackled in terms of battery damage), and finally last night it was nearly 100% working (everything apart from some tracks on the keyboard connector) - even the CMOS ram and battery backup were back working 100%.

So I left the computer soak-testing overnight and 8 hours later it was still fine this morning :).

Rather than switch off, I thought it'd be useful to give it another 2 more hours of testing, while I went out to the shops... However, when I got back home, the computer was now completely dead :(

Nothing happens on switch-on, other than the PSU's normal humm, no keyboard lights, disc drive lights, and no sound.

So I tested the voltage on the board, and +5v is reading 0.6v, +12 is reading similar, -12v is reading -1.2v...! So I suspect something on the PSU circuit has fried. I had though it could be a short on the main board, but would this only affect +5v?

Any thoughts where to start?

User avatar
flaxcottage
Posts: 2790
Joined: Thu Dec 13, 2012 8:46 pm
Location: Derbyshire

Re: Dead A3010 help... PSU maybe?

Postby flaxcottage » Sat Jun 03, 2017 12:58 pm

If there is a tantalum capacitor across the 5v line or even a normal electrolytic then yes. Use a multimeter to check for a short on the motherboard.
- John

Currently running Level 4 Econet with BBC B, BBC B+ 128K, Master 128K, 4Mb A3000, 4Mb A3020, 4Mb A4000, 4Mb A5000 dual FDD; UK101; HP41CX setup; Psion 3a, 3mx and 5mx; Z88; TI-58c, TI-59 and printer, HP-16C programmer's calculator

Zarchos
Posts: 2355
Joined: Sun May 19, 2013 8:19 am
Location: FRANCE

Re: Dead A3010 help... PSU maybe?

Postby Zarchos » Sat Jun 03, 2017 1:00 pm

Close all windows in order not to be tempted to throw it in the yard ... :cry:

And remember it's always possible to connect an external power supply :wink:

User avatar
vectorlight
Posts: 177
Joined: Sun Apr 10, 2016 12:23 pm
Location: UK
Contact:

Re: Dead A3010 help... PSU maybe?

Postby vectorlight » Sat Jun 03, 2017 1:13 pm

I've had 2 issues with the A3010 PSU dishing out the wrong voltage and in both instances was fixed by replacing the 2 big electrolytic caps (C10 & C17 I think) :D
3D Printing and Acorn A3010 Owner!

User avatar
1024MAK
Posts: 6782
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...

Re: Dead A3010 help... PSU maybe?

Postby 1024MAK » Sat Jun 03, 2017 1:16 pm

Not having looked at the schematics for this PSU, this is generic advice: If the power supply detects a short circuit of the main output (read as normally the highest current rail), it will reduce the power output, hence all outputs will have reduced voltages.

Check for tantalum capacitors, these are a known cause of shorts. Electrolytic types mostly go open circuit, or fall in value. But can go short circuit. Ceramic types, especially disc or rectangle through-hole types can go short circuit (mostly due to physical damage).

If a high-ish current is flowing in one of the outputs due to a short, if you have a good meter that can measure mV DC, you may find a downwards voltage gradient between the PSU output and the short circuit. Test the + and 0V as near to one another as you can, rather than leaving the negative meter lead on one place.

It can also be a failure of a component in tha PSU. Quickest way to test, is to use another PSU.

Most switch mode power supply unit faults are degraded electrolytic capacitors, or dry solder joints.

Mark
For a "Complete BBC Games Archive" visit www.bbcmicro.co.uk NOW!
BeebWiki‬ - for answers to many questions...

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Sat Jun 03, 2017 1:19 pm

flaxcottage wrote:If there is a tantalum capacitor across the 5v line or even a normal electrolytic then yes. Use a multimeter to check for a short on the motherboard.

Thanks.

Humm, 2.1 ohms from 5v to 0v..! Sounds like a short?

Need to isolate the problem now, which is slightly tricky with the psu being part of the main pcb... Back to the circuit diagram...

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Sat Jun 03, 2017 1:20 pm

Zarchos wrote:Close all windows in order not to be tempted to throw it in the yard ... :cry:

Windows closed! :)

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Sat Jun 03, 2017 1:21 pm

vectorlight wrote:I've had 2 issues with the A3010 PSU dishing out the wrong voltage and in both instances was fixed by replacing the 2 big electrolytic caps (C10 & C17 I think) :D

Ok, great. I'll take a look at these next.

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Sat Jun 03, 2017 1:25 pm

1024MAK wrote:Not having looked at the schematics for this PSU, this is generic advice: If the power supply detects a short circuit of the main output (read as normally the highest current rail), it will reduce the power output, hence all outputs will have reduced voltages.

Check for tantalum capacitors, these are a known cause of shorts. Electrolytic types mostly go open circuit, or fall in value. But can go short circuit. Ceramic types, especially disc or rectangle through-hole types can go short circuit (mostly due to physical damage).

If a high-ish current is flowing in one of the outputs due to a short, if you have a good meter that can measure mV DC, you may find a downwards voltage gradient between the PSU output and the short circuit. Test the + and 0V as near to one another as you can, rather than leaving the negative meter lead on one place.

It can also be a failure of a component in tha PSU. Quickest way to test, is to use another PSU.

Most switch mode power supply unit faults are degraded electrolytic capacitors, or dry solder joints.

Mark

Thanks Mark, the PSU is part of the main pcb, but there must be a way to isolate it.

Will look for this, as it'll help determine if the fault is in the PSU area or on the computer side.

Given that the computer was running fine for 8 hours, and the only noticeable heat was from the psu, that's where I think the issue will be - and as noted above, there are two large electrolytics next to the transformer...

User avatar
1024MAK
Posts: 6782
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...

Re: Dead A3010 help... PSU maybe?

Postby 1024MAK » Sat Jun 03, 2017 2:05 pm

If the PSU output circuits are each 'transformer winding', followed by a rectifier diode, and then smoothing capacitors. You can temporarily hook-up the outputs of another suitable PSU in parallel.

As the rectifier diodes will prevent the current from flowing through the transformer windings (that assumes the cause is not a diode failure, but this is easy to test for with a meter). Best to visually check the electrolytic (and if used, tantalum) smoothing capacitors first though.

Mark
For a "Complete BBC Games Archive" visit www.bbcmicro.co.uk NOW!
BeebWiki‬ - for answers to many questions...

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Sat Jun 03, 2017 10:03 pm

Right, had a poke around and looking at the A3010 circuit diagram, LK1 (wire link) bridges the PSU +5v line to the computer side +5v line. LK2 (push-on link) does the same for +12v line.

So I removed LK1 and LK2 and checked resistance between +5v and 0v on PSU and computer side.

Computer side was 2.2 ohms from +5v to 0v; PSU side was open circuit.

To confirm the PSU was OK, I wired up a simple LED test circuit and powered it from the A3010's PSU, while testing the voltage.

DSC_3876.JPG


5.1V :) so at least the PSU is OK...

Next step I guess is pulling the 7 tant caps on the computer side and testing each...?

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Sun Jun 04, 2017 12:01 am

steve3000 wrote:Next step I guess is pulling the 7 tant caps on the computer side and testing each...?

All tant caps removed and tested OK.

Next... electrolytics! There are a lot of these, any good place to start?

User avatar
1024MAK
Posts: 6782
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...

Re: Dead A3010 help... PSU maybe?

Postby 1024MAK » Sun Jun 04, 2017 10:10 am

steve3000 wrote:Next... electrolytics! There are a lot of these, any good place to start?
No, not when looking for a short circuit :(

I presume you have checked the resistance between +5V and 0V after removing each tantalum?

Have you refitted them? Have you tried powering up the computer section of the board?

Mark
For a "Complete BBC Games Archive" visit www.bbcmicro.co.uk NOW!
BeebWiki‬ - for answers to many questions...

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Sun Jun 04, 2017 11:19 am

1024MAK wrote:I presume you have checked the resistance between +5V and 0V after removing each tantalum?

Hi Mark, yes tested this after removing each one and with all removed. And tested resistance of each cap individually.
1024MAK wrote:Have you refitted them? Have you tried powering up the computer section of the board?

Tants refitted now.

I've not tried powering the computer side separately because it seems clear it's shorted?

Looking at the circuit diagram, there are actually only 4 electrolytics on the 5v 'bulk decoupling' line (between 0v and 5v) all the others have resistors or diodes at some point between the cap and supply or ground (so presumably could not be the cause of the 2.2 ohms between rails...).

Will tackle these today...

User avatar
1024MAK
Posts: 6782
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...

Re: Dead A3010 help... PSU maybe?

Postby 1024MAK » Sun Jun 04, 2017 12:41 pm

I mention re-trying the power, as sometimes a heat cycle (desoldering/soldering) has an affect on the capacitor concerned.

Mark
For a "Complete BBC Games Archive" visit www.bbcmicro.co.uk NOW!
BeebWiki‬ - for answers to many questions...

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Thu Jun 08, 2017 10:16 pm

Time for an update. I made some progress today, but first here's update part 1, which I should have posted two days ago.

Before tackling the electrolytics, I noticed that the mezzanine board itself had two surface mounted tants on top. I don't have any surface mount removal kit, and was just using my unregulated £5 soldering iron, but these tants came off easily, so I tested them - and both were ok. See picture below, the yellow surface mount tants are top right and bottom right of the mezzanine board.
Mezz1.png

Next I looked under the mezzanine board, just in case there were any hidden tant caps...and to my horror, there were! Two surface mounted tants are neatly stuck to the left underside of the mezzanine board...

Feeling rather un-confident, I then proceeded to tackle these tants by propping the PCB up neatly, and manoeuvring my large clunky soldering iron through the only available gap I could find. Somehow - I managed to get to one side of each of the underneath tants, to desolder that side from the mezzanine board. This allowed me to twist them out of contact, and test their resistance. See below:
Mezz3.png

Mezz2.png

Of course both were fine (grrr...!).

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Thu Jun 08, 2017 10:38 pm

At that point, I'd had enough for the time. But after a couple of days, I came back to the electrolytics - starting with the 4 electrolytics marked as "bulk decoupling" on the circuit diagram. I removed and tested all 4, but all were OK, and no change on the PCB +5v to 0v 2.2 ohm resistance. I was then going to start with all the remaining electrolytics which measured the same 2.2 ohms from + to -... but I ran out of time and energy, so slept on it overnight, which brings us to today.

Today, after l looking again, I decided I was not going to tackle the electrolytics yet, but would try switching the computer on. Just in case any of the heat cycles on those tants or electrolytics did anything... So I reconnected LK1 and LK2, and switched on, but this gave no response - no lights on the keyboard, and the voltage from +5v to 0v was still down to ~0.6v, exactly as before. I also noticed a faint ticking from the speaker. :(

Wondering if it could still be the PSU at fault, next I tried firing up with LK2 (12v connection from PSU) removed, to see if this made a difference. And it did! There was a loud pop from the speaker, and then repeated pop pop pop poping from the speaker... At the same time, my multimeter registered the +5v line bouncing with each pop, from 0v to 1-2v. Then I smelt smoke! So I switched off, and reconnected the 12v line via LK2.

After a few minutes, I tried switching on again - and this time no response for a few seconds, with +5v reading 0.6v, then pop pop from the speaker... then nothing. And then more smokey smells... I continued with this a few times, while testing the +5v line at different locations, and looking for any sign of a source of the smoke/smell.

I never found a specific source of the smoke, other than it probably came from somewhere near/around the CMOS battery area. However, after a few switches on/off, the popping stopped and the +5v line then settled back at +5.1v, and is now seemingly completely stable. There was still no life from the computer though, even when holding delete on switch on...so I tried a trick I'd discovered while resurrecting this A3010 previously, I removed the link from pin 5-6 of the genlock expansion connector LK6 on the PCB.*

And now... the floppy drive light flashes when I switch the computer on! So she is alive...ish! :D

However, I suspect the CMOS RAM chip has fried itself. :(


*Removing link from pin 5-6 of LK6 on the A3010 disconnects the video clock from the VIDC chip and so causes the RISC OS selftest to fail immediately, before RISC OS checks the CMOS RAM. This is really useful because a dead CMOS RAM chip can cause RISC OS to hang during the POST check, without reporting an error, however a missing VIDC clock will always trigger the flashing floppy drive light. So it's a great way to test if the ROMs are being read and the ARM is running, when you have a suspect CMOS RAM chip.

Zarchos
Posts: 2355
Joined: Sun May 19, 2013 8:19 am
Location: FRANCE

Re: Dead A3010 help... PSU maybe?

Postby Zarchos » Thu Jun 08, 2017 11:11 pm

Thanks for sharing all this.
Last bit really good to know and remember !

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Thu Jun 08, 2017 11:14 pm

steve3000 wrote:However, I suspect the CMOS RAM chip has fried itself. :(

Definitely looking like this now.

Computer is flashing out the code for "CMOS RAM checksum failed" on start up, even with delete-power on.

I'll need to find a replacement for this IC to get further I suspect, however I'm still wondering how and why there appeared to be a short on the PCB, and this has now gone...? (+5v to 0v resistance is now measuring >100 ohms).

User avatar
1024MAK
Posts: 6782
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...

Re: Dead A3010 help... PSU maybe?

Postby 1024MAK » Fri Jun 09, 2017 10:42 am

Whatever was causing the short has now damaged itself further due to the power from the PSU causing localised heating in the component. The bouncing voltage on the 5V line indicates that the "insulation" in the defective component kept "breaking down". It's now likely that the localised heating has either fried the conduction path so that a short circuit cannot flow, or the 5V feed or 0V GND path elsewhere (hopefully in the failed component) has gone open circuit.

Mark
For a "Complete BBC Games Archive" visit www.bbcmicro.co.uk NOW!
BeebWiki‬ - for answers to many questions...

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Fri Jun 09, 2017 12:17 pm

1024MAK wrote:Whatever was causing the short has now damaged itself further due to the power from the PSU causing localised heating in the component. The bouncing voltage on the 5V line indicates that the "insulation" in the defective component kept "breaking down". It's now likely that the localised heating has either fried the conduction path so that a short circuit cannot flow, or the 5V feed or 0V GND path elsewhere (hopefully in the failed component) has gone open circuit.

Good explanation - and yes I'm hoping the defective component was only the CMOS RAM IC. It is located right next to the battery (which had leaked and seriously corroded the PCB in that area) - so not surprising if the IC itself had suffered some internal corrosion.

Next steps are to buzz out all the circuit around the CMOS RAM IC to make sure nothing else damaged, then replace the chip :)

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Mon Jun 19, 2017 9:20 pm

She lives!

DSC_4041.JPG

DSC_4045.JPG

DSC_4046.JPG

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Mon Jun 19, 2017 9:52 pm

So what was the cause of the short circuit and how did I fix it?

I think I've at least narrowed it down to one of two things:

1) failed IC5 (CMOS RAM/clock IC PCF8583T) causing a short
or
2) failed C74 electrolytic cap causing a short

The burning/smokey smell I think I've narrowed down in part to the +5v supply line going from R61 to IC5 pin 8 and C74, because this track has burnt out! The track is now showing high resistance >100 ohms and looking slightly darker than other tracks nearby. I know from testing previously (when I buzzed out all the tracks around IC5 during initially cleaning the battery damaged area) this track was fine (0 ohms) before the short circuit.

The failed track is the one just above the red overlay on the picture below:
CMOSchip.png


This failure would suggest either C74 or IC5 being the cause of the problem, however right now both the IC5 and cap are showing high resistance, so I can't pin down the problem to either one.

For the moment, I have bridged this track (as shown in the photo) and swapped the original IC5 for a new PCF8583T - and by doing this, the computer is back fully functioning :)

...however I will test the old IC separately to determine if it's completely dead (and if not, then suspicion will move to C74).

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Mon Jun 19, 2017 10:08 pm

Some useful learnings:

1) RISC OS 3 won't boot with the CMOS RAM IC (PCF8583T) missing - it will just flash out a POST message from the floppy drive, with "CMOS RAM Unreadable" as the error.

2) Failed internal keyboard circuitry is a pain to deal with on the A3010, mainly because you cannot reset CMOS RAM to default settings - so if the battery has failed, the CMOS RAM will be scrambled. This should cause POST error "CMOS RAM checksum failed" but in my experience on this A3010 it doesn't - and the computer just freezes on power-on, with no apparent cause.

3) Switching LK14 from 1-2 to 2-3 disables the internal keyboard, allowing attachment of an external Archimedes keyboard to LK3, this is really useful to reset CMOS RAM, to sort out issue #2 above :)

4) As I mentioned a few posts up, removing the jumper from pins 5-6 on LK6 on the A3010, disconnects the VIDC clock from the VIDC chip, which can be helpful in triggering a VIDC IRQ error during POST and usually* (but not always) leads to start up error code flashing from the floppy drive (even works with issue #2 above). [*I say "usually" because I discovered after leaving the computer off for a couple of days, this trick didn't work first time but needed a few attempts switching the computer on/off, before the error code was flashed out]

Zarchos
Posts: 2355
Joined: Sun May 19, 2013 8:19 am
Location: FRANCE

Re: Dead A3010 help... PSU maybe?

Postby Zarchos » Tue Jun 20, 2017 6:53 am

Valuable infos !

Please show us 3)

steve3000
Posts: 1711
Joined: Sun Nov 25, 2012 12:43 am

Re: Dead A3010 help... PSU maybe?

Postby steve3000 » Tue Jun 20, 2017 10:00 pm

Zarchos wrote:Please show us 3)

OK: viewtopic.php?f=16&t=13291#p173332

:D

User avatar
1024MAK
Posts: 6782
Joined: Mon Apr 18, 2011 4:46 pm
Location: Looking forward to summer in Somerset, UK...

Re: Dead A3010 help... PSU maybe?

Postby 1024MAK » Wed Jun 21, 2017 10:12 pm

Well done :D

As small electrolytic capacitors don't cost much, I would recommend you renew C74 anyway.

Mark
For a "Complete BBC Games Archive" visit www.bbcmicro.co.uk NOW!
BeebWiki‬ - for answers to many questions...


Return to “hardware”

Who is online

Users browsing this forum: No registered users and 6 guests