Very dead A3000 (now very much alive)

discuss the archimedes & risc pc, peripherals and risc os/risc os on pi
Post Reply
Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Very dead A3000 (now very much alive)

Post by Boydie » Mon Oct 21, 2019 6:48 pm

I’m trying to resurrect an exceedingly dead A3000, previously suffered battery damage.

So far, I’ve cleaned it all up with vinegar etc. The zero-ohm links were all shot, so replaced. Same with the keyboard connectors. There are no obviously bad tracks/components left.
When I fire it up, nothing. The supply is steady at 4.8V. The keyboard (I’ve tried 3 remains unlit). It’s only RO2, so no floppy codes.

I decided to work my way through the Service Manual, and I think it fails at the first hurdle. With my logic probe, IC13 pin 29 remains stubbornly low immediately from power-on. Despite this, RST on IC44 pin 44 and IC37 pin 9 start high, then stay low (IC37 has a brief flurry of activity first).
For reference, the reset switch tests fine.

Is the activity on IC13 abnormal. If so, where does it point?
I’ve not got the scope out yet, but my meter reports LK28 as being 24MHz (waveform unknown).
2FBACB03-B0FA-441E-BD7F-812834FB410A.jpeg
Last edited by Boydie on Fri Nov 01, 2019 10:07 pm, edited 1 time in total.

Mark_G
Posts: 1
Joined: Sun Jan 27, 2013 11:38 am
Contact:

Re: Very dead A3000

Post by Mark_G » Mon Oct 21, 2019 7:06 pm

Hello,

After removing the leaked battery on my computer, I got the same problem. The machine didn't give a sign of live.
My A3000 is connected to a VGA monitor, by means of a 9-pin to 15-pin VGA Adapter.

I could start the machine this way :

It works in 2 steps :
First step : Hold the "delete" key, and power on. You hear a 'beep', but no screen.
If I'm right, this resets the cmos.

Second step :
you power the machine off.
Then you press the "3" key and power it on.
This select VGA as screen.

Between the first and second step, you have some time; You don't need to hurry.

This works for me.

Regards,

Mark

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Mon Oct 21, 2019 7:19 pm

Thanks. That was my hope originally. I forgot to say it failed.

I also forgot to mention that all the keyboards look dead - not even the power LED, let alone the Caps Lock etc ones.

Edit: Just tried the scope. The waveforms on LK28 and IC47 look okay (similar to those in arkleyjoe’s post). The waveforms on IC15 however are not so clean...
5411D7B7-BD6E-49FF-B55E-054165EDAA3D.jpeg
C27
675B5830-CD3F-46AE-AD8C-80179E03C5C2.jpeg
C28
E22C75AF-0072-48ED-8224-F1D26A388B7F.jpeg
29C
Edit 2: clock inputs to vidc pin 19 and memc pin 67 have the same waveform as LK28

User avatar
Tills
Posts: 44
Joined: Fri Aug 30, 2019 2:08 pm
Location: Newcastle, UK
Contact:

Re: Very dead A3000

Post by Tills » Mon Oct 21, 2019 8:38 pm

Perhaps it's the photo, but is that a damaged track underneath ROM no. 3?

Otherwise, the board looks in pretty good shape around the chips and keyboard connector, considering the battery leakage.

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Mon Oct 21, 2019 10:04 pm

Well spotted. Even close-up it looked intact, but was in fact bad.
Pin 29 on IOC now starts low, then changes to high after a second or two, so that’s an improvement.
The keyboard is still unresponsive and unlit.
The display shows nothing, nor does it beep, but that may well be because the CMOS needs resetting. Which, given the lack of keyboard, may be a bit of a challenge...

I know failure of IC4 and/or IC5 causes the keyboard to misbehave, but can it stop even the power LED from doing anything?

Edit: Interestingly, SK6 has 4.8V coming from a lot of its connectors. SK7 has 0.6V at most, including for the power on LED, and Pin16, which should be the 5V supply.

User avatar
Tills
Posts: 44
Joined: Fri Aug 30, 2019 2:08 pm
Location: Newcastle, UK
Contact:

Re: Very dead A3000

Post by Tills » Mon Oct 21, 2019 11:14 pm

As there's no battery, I believe the CMOS is reset each time the Arc is when switched off (allowing a little time for the capacitor to lose its charge). Are there any upgrades to the A3000, like an ARM3 or a VIDC enhancer? I found that my RAM upgrade card needed reseating to eliminate my start-up failure (though I see there isn't one fitted to yours).

As for the keyboard power light, you could try testing the socket by measuring volts from the 5v pin to the power-LED pin. Looking at your upside-down photo, it would be top row, first pin left and fifth pin left - or more correctly, SK7 pins 16 (5v) and 20 (Power on LED). You could even try plugging in an LED directly into the socket if you have a spare, which would give you a nice visual indication of when the Arc is switched on. From what I can glean from the schematics, this does not depend on the keyboard controller chip (IC2); the LED is simply connected from the positive to negative rails through a resistor. Of all things, this LED should work regardless.
Last edited by Tills on Tue Oct 22, 2019 12:32 pm, edited 3 times in total.

SteveBagley
Posts: 256
Joined: Sun Mar 15, 2015 8:44 pm
Contact:

Re: Very dead A3000

Post by SteveBagley » Tue Oct 22, 2019 1:38 am

The A3000 motherboard lets you connect up an A310 style motherboard if you cut a couple links and solder in the correct socket — I used a PS2 extension cable cut in half to get the socket to attach.

Might be worth trying that to get a working keyboard if you have one to hand?

Steve

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Tue Oct 22, 2019 12:48 pm

Tills wrote:
Mon Oct 21, 2019 11:14 pm
As there's no battery, I believe the CMOS is reset each time the Arc is when switched off (allowing a little time for the capacitor to lose its charge). Are there any upgrades to the A3000, like an ARM3 or a VIDC enhancer? I found that my RAM upgrade card needed reseating to eliminate my start-up failure (though I see there isn't one fitted to yours).

As for the keyboard power light, you could try testing the socket by measuring volts from the 5v pin to the power-LED pin. Looking at your upside-down photo, it would be top row, first pin left and fifth pin left - or more correctly, SK7 pins 16 (5v) and 20 (Power on LED). You could even try plugging in an LED directly into the socket if you have a spare, which would give you a nice visual indication of when the Arc is switched on. From what I can glean from the schematics, this does not depend on the keyboard controller chip (IC2); the LED is simply connected from the positive to negative rails through a resistor. Of all things, this LED should work regardless.
It’s already downgraded to a virgin, unexpanded A3000, RO2. So no expansions to muddy the picture.

Both the +5V and LED lines on SK7 read 0.3-0.6V relative to ground, and always the same as each other. So do many other pins on SK7. I’ve not checked, but I’m willing to bet that it’s 0V between pins 15 and 20.

It looks like a massive short, somewhere, affecting the entire length of SK7. I suspect when I find it, the whole thing will work...

User avatar
Tills
Posts: 44
Joined: Fri Aug 30, 2019 2:08 pm
Location: Newcastle, UK
Contact:

Re: Very dead A3000

Post by Tills » Tue Oct 22, 2019 1:10 pm

You should be reading 5 volts on pin 16 (I had erroneously wrote pin 15 but swiftly corrected myself). This supplies the power for all the LEDs, returning on pins 1 and 17–20. If you're not reading 5v then there may be a bigger problem elsewhere.
Attachments
Acorn_A3000TRM_MainPCBCircuitDiagram2of2.jpg

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Tue Oct 22, 2019 3:48 pm

Yeah, I meant 16 as well. :roll:

It definitely only reads 0.3-0.6V, and it seems to be an issue local to SK7. Other areas that should be 5V read the 4.8V that the psu’s supplying.
Since the other pins that should read 0V read the same 0.3-0.6, presumably something’s shorting that 5V supply yo ground...

User avatar
IanS
Posts: 1256
Joined: Mon Aug 31, 2009 7:02 pm
Contact:

Re: Very dead A3000

Post by IanS » Tue Oct 22, 2019 3:55 pm

Just as a double-check, are you counting from the right end. Pin 1 is at opposite ends for the two keyboard connectors.

User avatar
1024MAK
Posts: 9969
Joined: Mon Apr 18, 2011 5:46 pm
Location: Looking forward to summer in Somerset, UK...
Contact:

Re: Very dead A3000

Post by 1024MAK » Tue Oct 22, 2019 4:31 pm

More likely that something is open circuit rather than short circuit. A short circuit would likely cause either a track to have a hole blown in it, drag down the +5V rail elsewhere, or cause the PSU to go into protection mode (shut down).

But before then, are you using a known good and reliable 0V/GND point?

Mark

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Tue Oct 22, 2019 7:48 pm

Okay, small progress.
As IanS surmised, I was counting from the wrong end of SK7. :oops:
When I looked at the real pin 16, it was reading 0V and no continuity to the 5V pin on the motherboard.

One replaced socket, and the keyboard now lights up and the various lock keys glow brighter when reset is pressed.

Other than that, it’s still dead - no response to power-on CMOS reset.

So, are those waveforms on SK15 (previously labelled by me as IC15) indicative of anything?

I’m using either the 0V PSU spade connector or disc drive 0V line as GND/0V.

Edit: Unless anyone has any better suggestions, I’ll replace the two 74Ls145s next. They’re in the line of fire for battery damage, and that damage would prevent the keyboard giving power-on CMOS reset.

User avatar
Tills
Posts: 44
Joined: Fri Aug 30, 2019 2:08 pm
Location: Newcastle, UK
Contact:

Re: Very dead A3000

Post by Tills » Tue Oct 22, 2019 10:26 pm

Boydie wrote:
Tue Oct 22, 2019 7:48 pm
Edit: Unless anyone has any better suggestions, I’ll replace the two 74Ls145s next. They’re in the line of fire for battery damage, and that damage would prevent the keyboard giving power-on CMOS reset.
Also in the line of fire is IC10, which provides the 'life-line' clock signal to the keyboard controller chip, IC2. Makes sense to check continuity from pin 5 6 of IC10 to pin 18 of IC2 (via LK14), as I think this trace runs on top of the board to the right of the battery. I'd check this out first. There's also pin 6 5 to LK15, but if I'm reading the schematic correctly this is not actually connected to IC2.
Boydie wrote:
Tue Oct 22, 2019 12:48 pm
It’s already downgraded to a virgin, unexpanded A3000, RO2. So no expansions to muddy the picture.
Was there a VIDC Enhancer fitted? Because if this has been removed, certain links need to be returned to default positions.
Last edited by Tills on Wed Oct 23, 2019 8:29 pm, edited 1 time in total.

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Wed Oct 23, 2019 7:44 pm

Pin 5 of IC10 is connected to Pin 18 of IC2. Pin 6 of IC10 and Pin 19 of IC2 are connected to LK15, which is not made (which is what my circuit diagram suggests should be the case).
For reference, this is the signal on IC2 Pin 18:
6314D61A-25C0-4329-BAE8-187DFE3038B0.jpeg
Edit: I’ve gone a bit further in the Service Manual testing. ICs 29 and 30 both have flurries of logic probe activity on all the relevant pins on power-up. Pin 15 of IC31 is stuck low - bad IC or bad input?

Edit 2: Looking at the circuit diagram, it would seem that Pin 15 of IC31 is connected to LK20, which in RO2 is made to 5V rather than IC31. So probably a red herring

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Fri Nov 01, 2019 2:44 pm

Well, my hopes weren't high, and I wasn't disappointed!

Changing the 74LS145Ns hasn't helped. I've also swapped out the 32.768MHz crystal because that looked a bit manky.
I've also continuity tested all the pins for the ROMs as far as ICs 29-31 for the address lines, MEMC p60 for ROMCS, and the corresponding resistor for D0-D31.
I've tried swapping RO3 in to see if there are any floppy error codes. There aren't.

Where's best to try next?
Are the odd-looking clock signals on SK15 any form of a hint?

Kazzie
Posts: 1552
Joined: Sun Oct 15, 2017 8:10 pm
Location: North Wales
Contact:

Re: Very dead A3000

Post by Kazzie » Fri Nov 01, 2019 3:28 pm

I seem to have lost track of where you got to on this repair project.

How far through the "System failure" tests in the service manual did you get? (Reset signal on IC13 pin 29, RST signal on IC 44 pin 44 and IC 37 pin 9, etc.)
BBC Model B 32k issue 7, Sidewise ROM board with 16K RAM
Archimedes 420/1 upgraded to 4MB RAM, ZIDEFS with 512MB CF card
Acorn System 1 home-made replica

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Fri Nov 01, 2019 9:28 pm

Kazzie wrote:
Fri Nov 01, 2019 3:28 pm
I seem to have lost track of where you got to on this repair project.
Driven halfway up the wall and at the end of my tether. :)

Working my way through the Service Manual...

1) IC13 pin 29 starts low on power-up and changes to high after a second or two. Originally it didn't, but fixing that broken trace sorted that.
2) Both IC44 pin 44 and IC 37 pin 9 are high whilst Reset is pressed, then go low
3) System clock is present on LK28 and looks clean
4) SKT15 has waveforms on 27C-29C, but these look abnormal, as if a second, lower amplitude, higher frequency sine wave was superimposed (see photos above)
5) Clocks are present on IC44 pin 67 and IC41 pin 9, and look remarkably similar to the signal on LK28
6) As Arkleyjoe did, here's a (less pretty representation) of what's happening on ICs 29-31:

IC 29 IC30 IC31

H111122222 H111111112 HLLLL11112
L11112222L L11111111L LLLLL1111L

Where H=permanently high, L=permanently low, 1=H/L lit and pulsing, 2=L lit and pulsing. I don't know if 1 and 2 truly represent different states or are an artefact of my logic probe, but the results are consistent and reproducible every time.

Going a bit further tonight...

7) Address lines A2-A25 on MEMC are all pulsing away merrily, in different patterns. A0 and A1 are stuck low. On IOC, LA2-6 are pulsing away, as are LA16,17,19,20,21. Pin 66 (LA18) is stuck low and not pulsing.

If I'm reading the circuit diagram correctly, Pin 66 of IOC should be connected to Pin 13 of IC31. There's no continuity between these pins, whereas there is between, say, Pin 14 of IC31 and Pin 1 of IOC.
Tracing this down, there doesn't seem to be continuity between Pin 1 of LK17 and the via immediately below it; on another A3000 board there is. There is some corrosion under LK17 and 18, so I'll change them and see what happens...

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000

Post by Boydie » Fri Nov 01, 2019 10:07 pm

Well, I removed LK 17&18. There was definitely no continuity between LK18 pin1 and the via, but the damage must have been very subtle because there was nothing visible, even with a magnifying glass.

Replaced LK 17&18. Patched between LK18 and the via. Got continuity between IOC p66 and IC31 p13. IOC p66 now had activity on the logic probe.

All looked good. Hooked it up to a display and powered up...

Nothing, zilch, nada, not an electronic sausage.

Then realised it would may just possibly help if I removed the reset link from SK3. Oddly enough, it did. One CMOS reset later and I've been rewarded with a beep and a desktop! :D

Yup, I'm an idiot.

Many thanks to everyone for the advice and support, particularly in Arkleyjoe's thread - it was a perfect lesson in troubleshooting.

Boydie
Posts: 437
Joined: Sat Oct 24, 2015 9:25 am
Location: Sunny Wigan
Contact:

Re: Very dead A3000 (now very much alive)

Post by Boydie » Sat Nov 02, 2019 5:33 pm

One last fault found - broken trace between L7 and pin 27 of the 8051, which meant mouse Y-axis wasn't working.

Now appears to be fully functional - Econet and User Port/MIDI cards are recognised, as is 2Meg RAM upgrade. Floppy drive access works fine. Mouse is fully functional. Even the RTC seems to have survived unscathed - just need to bypass R7, which is shot, to get the battery working...

I think it's safe to say I've learnt a lot from this. I may even repair my spare A3000 board (which is in a worse state), just for the fun of it!

Many thanks again. I'd been putting off repairing this machine for the best part of a decade.

Kazzie
Posts: 1552
Joined: Sun Oct 15, 2017 8:10 pm
Location: North Wales
Contact:

Re: Very dead A3000 (now very much alive)

Post by Kazzie » Sat Nov 02, 2019 6:23 pm

I'm glad you found success, even if I wasn't able to do much more than offer moral support! =D>
BBC Model B 32k issue 7, Sidewise ROM board with 16K RAM
Archimedes 420/1 upgraded to 4MB RAM, ZIDEFS with 512MB CF card
Acorn System 1 home-made replica

Post Reply

Return to “32-bit acorn hardware”