Partially Working A4000 - Help!!

discuss the archimedes & risc pc, peripherals and risc os/risc os on pi
User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Sat Jun 30, 2018 7:40 pm

Another three years have passed and my inability to fix this A4000 system still plays on my mind.

In the mean time, I've acquired some better test kit, so I might have another try with this.

My preferred technique now for debugging a Beeb is to do a data bus capture, and feed this into into the "6502 protocol decoder" tool I wrote. I'd like to try something similar with the ARM, possibly by snooping the address/data bus on the ROM (as it's easy to connect to this).

Am I correct in thinking that the A4000 uses an ARM250, which doesn't have any kind of cache? So snooping the address bus should yield useful information?

Is there any way to bypass the soldered-in DRAM (analogous to link S25 on the Beeb)?

Does anyone have a fatally battery damaged A3010/3020/4000 board they would be willing to donate or sell for a nominal amount, that I could use for spares?

Dave

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Sun Jul 01, 2018 7:42 pm

This afternoon has been interesting to say the least!

I brought the A4000 down from the loft, powered it up, and it complained there was no keyboard attached (which there was). After closer inspection, the keyboard connector was very corroded from the previous battery damage. So I carefully desoldered it and fitted a new connector.

After doing this, the system seems more reliable than before, and is happy running from 5V again.

So I refitted the battery, cleared the CMOS Ram (pressing Delete on power up), and I'm now able to get to the RISC OS Desktop.

Once in the Desktop, the system is 100% reliable and will run for hours. I've even been able to play Zarch. I've also been able to run the Test/Diagnostics disc, which I've never managed to do before. I can't really explain what's changed in the last 3 years, but something has. Or was a corroded keyboard socket somehow to blame all along?

The A4000 soak test on the diagnostic disc mostly passes, but fails on two specific tests:
- The write protect on the floppy not being effective
- The CMOS Ram test fails (specifically writing location &85)

So it's possible there is an issue with the CMOS Ram, but apart from the diagnostic failing it does seem to be keeping it's configuration. Or maybe the diagnostic is broken. That location stored the monitory type, which seems to be retaining it's value. So maybe the diagnostic is faulty?

Anyway, the main outstanding issue is that system will not boot reliably from power up. You need to press the hard reset key a couple of times. Once the system has reached the desktop successfully once, that it will perform flawlessly until it is powered down.

So just possibly there is an issue reading the configuration from the CMOS Ram on first power up (though no checksum error is reported). Or more likely there is a problem with the reset circuitry. The symptoms are now identical to Steve's A5000:
viewtopic.php?f=16&t=9378

Dave
Last edited by hoglet on Sun Jul 01, 2018 7:43 pm, edited 1 time in total.

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Sun Jul 01, 2018 10:11 pm

Well, the soak test for the last few hours didn't go so well.

The A4000's power supply seems to have suffered a rather catastrophic failure, taking with it:
- the 2A internal fuse
- the main switching transistor on the HT side
- a smaller transistor
- two high wattage 2K2 resistors
- a smaller 10R resistor
- and possibly the HT winding of the transformer

The good news (if there is any) is that it seems not to have taken the A4000 main board, as this still boots when powered from the Bench PSU.

Drat!

Does anyone know if the A4000 power schematic is available anywhere?

Dave
Last edited by hoglet on Sun Jul 01, 2018 10:12 pm, edited 1 time in total.

steve3000
Posts: 2160
Joined: Sun Nov 25, 2012 12:43 am
Contact:

Re: Partially Working A4000 - Help!!

Post by steve3000 » Mon Jul 02, 2018 1:42 am

Yikes, sorry to hear about your A4000 PSU. Glad the main PCB is ok though. I don't recall seeing the schematic for the A4000 PSU in the TRM (the A3010/A3020 on-PCB PSUs are shown, but not sure these would be useful) although it's been a while since I looked so could be wrong.

I never got round to further repairing my A5000 for various reasons, but still suspect the CMOS chip. I have however since repaired a couple of other A5000s with various issues and fitting a new CMOS chip did solve a failed startup problem on one of these... So I'd recommend trying this, especially if you already have suspicions about the CMOS chip.

Promoted by your resurrection of my old repair thread, next in my list of projects for the summer will be to dig out my old A5000 and attempt the CMOS RAM change. :)

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Mon Jul 02, 2018 7:14 am

steve3000 wrote:
Mon Jul 02, 2018 1:42 am
Yikes, sorry to hear about your A4000 PSU. Glad the main PCB is ok though. I don't recall seeing the schematic for the A4000 PSU in the TRM (the A3010/A3020 on-PCB PSUs are shown, but not sure these would be useful) although it's been a while since I looked so could be wrong.
Thanks Steve.

I've just found CJE are selling original A4000 power supplies on eBay for £24, which seems quite reasonable so I have bought one. I will try to fix the old one as well if that is possible.

I'm putting an order in to Farnell this week, so I'll add on a replacement CMOS Ram. I'd actually be quite surprised if that fixed my problem, but it's very inexpensive so worth trying.

Dave
Last edited by hoglet on Mon Jul 02, 2018 7:15 am, edited 2 times in total.

aotta
Posts: 247
Joined: Fri May 26, 2017 9:57 am
Location: Italy
Contact:

Re: Partially Working A4000 - Help!!

Post by aotta » Tue Jul 03, 2018 11:09 am

Fyi, on the bay there are a couple of A4000 psu now.. Not at cheap price, but original and working...

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Tue Jul 03, 2018 11:40 am

aotta wrote:
Tue Jul 03, 2018 11:09 am
Fyi, on the bay there are a couple of A4000 psu now.. Not at cheap price, but original and working...
Thanks, I now have one of these, and actually I thought £24 was a pretty reasonable price (there is one still left at this price).

I've also traced a schematic of high voltage side, so should be able to fix my existing one for a few pounds as well.

Dave

aotta
Posts: 247
Joined: Fri May 26, 2017 9:57 am
Location: Italy
Contact:

Re: Partially Working A4000 - Help!!

Post by aotta » Tue Jul 03, 2018 12:27 pm

And i always go for a few pounds more, if it let me avoid working with i psu and crt monitor.. 😉
I am working on repairing my a4000 too, but my focus is now in the ide connection.. 😖

TynHau
Posts: 17
Joined: Mon Oct 01, 2012 10:55 pm
Contact:

Re: Partially Working A4000 - Help!!

Post by TynHau » Fri Jul 06, 2018 10:51 am

hoglet wrote:
Tue Jul 03, 2018 11:40 am
there is one still left at this price
Not anymore, there isn’t! Thought I’d buy one myself while I can. Of course this now means the one powering my A4000 will never fail. :roll:

aotta
Posts: 247
Joined: Fri May 26, 2017 9:57 am
Location: Italy
Contact:

Re: Partially Working A4000 - Help!!

Post by aotta » Sat Jul 07, 2018 10:17 am

I hope mine will work forever too! Anyway, psu is one of the simpliest parts to replace, and i used sometimes picopsu inserted into original psu.. Less weight and more efficiency!

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Thu Feb 20, 2020 7:06 pm

And another two years has passed.... This has to be a record for the longest active debug thread!

I've been messing with this A4000 again; symptoms broadly unchanged (do read the thread from the start if you need to catch up).

I've tweaked the PSU voltage down to exactly 4.99V (it was 5.10V) and this is improved reliability a tad. As reported previously, whatever the fault is, it's very sensitive to high power supply voltages.

I've also connected a "Post Box" to the test connector, as reported in this thread:
viewtopic.php?f=16&t=17478#p260399

Nothing unexpected ever shows up during the POST Test, which always passes.

Booting to the desktop is now fairly reliable (>95%) and Zarch will play for hours.

I can also draw random triangles forever without (yet) seeing a crash.

Code: Select all

10 MODE 27
20 REPEAT
30 GCOL 0,RND(256)-1
40 PLOT 85,RND(1280)-1,RND(1024)-1
50 UNTIL FALSE
But I can still make it crash very quickly with CLS:

Code: Select all

10 MODE 27
20 REPEAT
30 CLS
40 UNTIL FALSE
I wanted to try a different screen mode, specifically MODE 28 (640x480 in 256 colour), but I get a "Bad screen mode" error. Is this expected?

Dave

SteveBagley
Posts: 250
Joined: Sun Mar 15, 2015 8:44 pm
Contact:

Re: Partially Working A4000 - Help!!

Post by SteveBagley » Thu Feb 20, 2020 7:18 pm

hoglet wrote:
Thu Feb 20, 2020 7:06 pm
I wanted to try a different screen mode, specifically MODE 28 (640x480 in 256 colour), but I get a "Bad screen mode" error. Is this expected?
You probably need to modify the amount of memory set aside for the screen in the task manager (or by using *configure screensize -- I think) 160K should be enough.

Steve

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Thu Feb 20, 2020 8:07 pm

SteveBagley wrote:
Thu Feb 20, 2020 7:18 pm
You probably need to modify the amount of memory set aside for the screen in the task manager (or by using *configure screensize -- I think) 160K should be enough.
Thanks Steve.

320K did the trick.

It's now happily drawing triangles in MODE 28.

Looks like the fault is quite specific to CLS. As per an earlier post, it seems to be the combination of heavey use of STM and Interrupts being enabled. I think tomorrow I'll rewrite my test version of CLS to not use STM, and see if that changes the behaviour.

Dave

steve3000
Posts: 2160
Joined: Sun Nov 25, 2012 12:43 am
Contact:

Re: Partially Working A4000 - Help!!

Post by steve3000 » Sat Feb 22, 2020 8:39 am

Yes, this is an enduring repair thread, well done for not giving up.

I'm still wondering about the RAM, do you have access to one of cmorely's A4000 RAM kits? With these you can bypass the onboard RAM...

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Sat Feb 22, 2020 9:06 am

steve3000 wrote:
Sat Feb 22, 2020 8:39 am
I'm still wondering about the RAM, do you have access to one of cmorely's A4000 RAM kits? With these you can bypass the onboard RAM...
Unfortunately I don't have one of these.

I'll be seeing Chris today at the SW RISCOS show, so I'll ask if there are any still available.

Dave

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Sun Mar 01, 2020 7:55 pm

I've now received a 4MB RAM kit from Chris, which I have fitted to the A4000.

Unfortunanetly, the weird behaviour is unchanged, so we can rule out a subtle RAM fault I think.

This pretty much just leaves to ARM250.

If anyone has a scrap A3010/A3020/A4000 board that is battery damaged beyond repair, then do get in contact with me!

Dave

SteveBagley
Posts: 250
Joined: Sun Mar 15, 2015 8:44 pm
Contact:

Re: Partially Working A4000 - Help!!

Post by SteveBagley » Sun Mar 01, 2020 8:43 pm

Just a thought but I’m assuming fitting extra RAM off the motherboard still leaves the onboard RAM connected to the address and data bus and just disconnects the enable lines to the onboard RAM. Could the onboard RAM still be loading the address/ data bus weirdly and causing an issue?

Steve

User avatar
hoglet
Posts: 9096
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Partially Working A4000 - Help!!

Post by hoglet » Sun Mar 01, 2020 9:11 pm

SteveBagley wrote:
Sun Mar 01, 2020 8:43 pm
Just a thought but I’m assuming fitting extra RAM off the motherboard still leaves the onboard RAM connected to the address and data bus and just disconnects the enable lines to the onboard RAM. Could the onboard RAM still be loading the address/ data bus weirdly and causing an issue?
That's always a possibility, and it could be any other device, not just the RAM.

I still don't understand how a complex game like Zarch will play indefinitely without a crash, yet a few CLS commands will always crash.

Dave

User avatar
myelin
Posts: 854
Joined: Tue Apr 26, 2016 10:17 pm
Location: Mountain View, CA, USA
Contact:

Re: Partially Working A4000 - Help!!

Post by myelin » Sun Mar 01, 2020 10:26 pm

SteveBagley wrote:
Sun Mar 01, 2020 8:43 pm
Just a thought but I’m assuming fitting extra RAM off the motherboard still leaves the onboard RAM connected to the address and data bus and just disconnects the enable lines to the onboard RAM. Could the onboard RAM still be loading the address/ data bus weirdly and causing an issue?

Steve
This was the case with my A5000 - I had to desolder one RAM chip to get the data line test to pass; it was messing with D31.
SW/EE from New Zealand, now in Mountain View, CA, making BBC/Electron hardware projects for fun.
Most interesting: Arcflash, FX2+PiTubeDirect Tube/Cartridge adapter, USB cart interface.

sP1d3r
Posts: 45
Joined: Fri Aug 23, 2019 2:40 am
Contact:

Re: Partially Working A4000 - Help!!

Post by sP1d3r » Tue Mar 24, 2020 12:39 pm

:D Although I know nothing about the problems with the A4000 that've so far been described, the basic problem sounds to me like it may be connected with a faulty Zener diode or a component that a zener enables.

Hope of help.

Post Reply

Return to “32-bit acorn hardware”