BBC A3000 + Watford Electronics ARM3 - installation woes

Arc/RPCs, peripherals, RISCOS operating system & ARM kit eg GP2x, BeagleBoard
shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Sat Jul 29, 2017 8:46 pm

Hey there,
I'm currently attempting to install a Watford Electronics ARM3 upgrade into a friend's BBC Archimedes A3000.

Installing the PLCC socket for the CPU has worked nicely after a failed first attempt (using hot air, burning the plastic to a slight crisp) and the machine works nicely with the original CPU now in a socket (as long as you don't push it down too far). Photo here: http://i.imgur.com/HFhhaNv.jpg
However, when I install the ARM3 upgrade I get the usual magenta-blue selftest and the computer appears to attempt to jump to the ROM (RISC OS3) with the background going solid black. After that it appears to be dead ... even pushing reset doesn't do anything for the first 30 seconds or so (after that it either does a delayed reset or reacts to the push of the button). Pushing the toggle switches (capslock, etc.) doesn't light up the corresponding LED and my "quicktest" for confirming the CPU as working by making a bunch of keys stick down (black screen with red border after the selftest) shows no reaction either.
I've tried reseating the upgrade certainly 50 times or so to get a bootup - with no success so far.

The machine is upgraded to 4MB and has an addtional IDE controller + CF adapter (can't recall the brand currently, sorry - I think it's a Simtec one for the expansion headers near the top left).

Any ideas why the machine is not working? Has anyone of you had success with above upgrade? I should mention the Watford Electronics ARM3 is an "Issue 2" - I guess that refers to an "MK2" since I found some info online that the first variant of that upgrade wasn't compatible with A3000 computers.
Last edited by shock__ on Sun Jul 30, 2017 12:30 am, edited 1 time in total.

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Sat Jul 29, 2017 9:47 pm

Hi,
Welcome to the forum!

Great to see someone doing this modification. Could the problem you've experienced be the same ALE issue paulv had when installing ARM3 to an A3010?

Full details are documented in his forum post here: viewtopic.php?f=16&t=6833

Also see his detailed write up here: http://www.retro-kit.co.uk/page.cfm/con ... processor/

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Sun Jul 30, 2017 10:08 am

Tried that (ALE isn't connected from the ARM2 plug to the ARM3) - no improvement.

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Sun Jul 30, 2017 10:42 am

Couple of questions, have you removed every expansion, to get back to the base 1mb RO3 system (just in case of a secondary issue with hdd or RAM which wasn't evident with ARM2)?

And are you switching on holding R, to reset CMOS? (this is needed the first time after upgrade, to ensure the cache-on flag is set correctly).

From what you say, the point of failure seems to be immediately after the self test. This means the RO3 ROM is fine, as is the ARM3, because it has executed the self test code and confirmed ROM and CMOS checksum. After this point, expansion ROM/podules are queried and drivers loaded to RAM, and the CMOS settings are applied.

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Sun Jul 30, 2017 1:33 pm

I've tried my luck with and without the expansions, I have gone back to fully stock for now (1MB, no IDE). No change.

Thanks about letting me know to reset the CMOS by holding R - that's something I would have missed. That doesn't seem to help either I'm afraid (as said getting the "key stuck" screen doesn't happen either). I've also tried 'delete' which doesn't do anything either.

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Sun Jul 30, 2017 4:52 pm

What happens if you leave the computer on for 30-60 seconds? If it has failed power on self test, then depending on the error, it may proceed to boot after a delay (delay is due to flashing codes out on the floppy drive).

Also just check the floppy drive is plugged in - have you seen any self-test failure codes being flashed on the drive light?

If no fail codes, you can force an error to check the ROM is getting as far as the end of the self test. Check if LK28 is fitted. If so, you can remove the shunt, which will disconnect the 24MHz clock from VIDC - and cause a self test fail on powerup. This should be flashed out on the floppy drive - and at least that will confirm absolutely that the ARM3 and ROM are working fine. If LK28 isn't fitted, you can temporarily desolder either end of R138 to get the same effect (or you can fit LK28, but in that case, make sure to cut the track between the pins first).

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Sun Jul 30, 2017 10:33 pm

I had tried the LK28 procedure earlier - results were just as you mentioned. No video, floppy blinking.

About letting the computer run for a bit, I tried that ... it either does a delayed reset if I pressed the button earlier ... or just sits there (waited up to 5 minutes - after that it would instantly react to pushing reset).

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Mon Jul 31, 2017 11:01 pm

shock__ wrote:I had tried the LK28 procedure earlier - results were just as you mentioned. No video, floppy blinking.

About letting the computer run for a bit, I tried that ... it either does a delayed reset if I pressed the button earlier ... or just sits there (waited up to 5 minutes - after that it would instantly react to pushing reset).

Good to know the LK28 trick works - so there is no problem with the ARM3 executing code from the ROMs.

Is the keyboard definitely working OK? (I've had keyboard related startup issues before)

When you say the computer reacts to pressing reset - does it boot up OK after the delayed reset?

Another thing to try, if you have the RISC OS 2 ROMs (or a way of burning them), is to set the links for RO2 and boot using that. RO2 doesn't carry out a self check and will run with multiple failing components, so you'll soon work out what's wrong :)

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Mon Jul 31, 2017 11:39 pm

I found a 4x128K image for RISC OS 2.00 ... would I have to change any of the ROM related jumper links when using 29F010 for ROMs?

Also, whenever the machine reacts to the reset, it will go through the selftest again and then jump back to the solid black screen.

Keyboard works fine when using the original CPU.

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Tue Aug 01, 2017 12:03 am

shock__ wrote:I found a 4x128K image for RISC OS 2.00 ... would I have to change any of the ROM related jumper links when using 29F010 for ROMs?

I've not used these before, you'll need to confirm with the circuit diagram. I've used 4 x 27c801's successfully in the past to burn RO3 (4x512kb images) and just doubled up the content to fit the eprom size.

But remember you'll need to ensure the data is correctly striped across the images, so that for each 32bit word of data, the first byte comes from rom 0, the second byte from rom 1, third from rom 2 and fourth from rom 3.

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Tue Aug 01, 2017 12:31 am

steve3000 wrote:But remember you'll need to ensure the data is correctly striped across the images, so that for each 32bit word of data, the first byte comes from rom 0, the second byte from rom 1, third from rom 2 and fourth from rom 3.
That's good to know because the ROMs I found apparently are split "flat" rather than in a bytewise fashion (or at least I'm getting plain ASCII text still recognizeable - even tho they're supposedly dumped directly from ROMs and named according the the IC footprint they're supposed to go into). I'll investigate that.

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Tue Aug 01, 2017 6:33 am

shock__ wrote:That's good to know because the ROMs I found apparently are split "flat" rather than in a bytewise fashion (or at least I'm getting plain ASCII text still recognizeable - even tho they're supposedly dumped directly from ROMs and named according the the IC footprint they're supposed to go into). I'll investigate that.

They'll be data dumps of the ROM area taken from within RISC OS then, rather than being read from an EPROM reader.

Emulators will accept these, but some emulators require the ROM files to be named with physical ROM numbers, which is misleading.

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Tue Aug 01, 2017 11:20 am

Minor success!

With the RISC OS 2.01 downgrade the machine now responds to the "sticky key quicktest" and outputs the amount of memory found on the screen. Attempting to boot (without the keys stuck) it does blink the cursor once and then appears to crash/get stuck.

I'll investigate further from here ... my ROMs are 150ns, that should be borderline within the limits for the 10MHz bus, right?

EDIT1:
Okay, with the original CPU the computer boots to the desktop.
With the ARM3 it hangs for roughly 30 seconds and then gives me the following message:
"Branch through zero at &03811F1C (Error number &80000005)"
followed by a * prompt in which I can enter _some_ keys (which don't exactly correspond to the keyboard symbols).

EDIT2: Re-applying the ALE patch brought no improvement, leaving it there for now.

EDIT3: Apparently the machine has a keyboard issue related to the connector ... I'll get into that at a later point.
Would resetting the CMOS give me any kind of feedback?

EDIT4: Holding T made the screen loose sync, resetting via R worked. "Branch through zero" error remains.
However calling the Desktop from the * prompt via "DESKTOP" makes it pop up ... so I think we're getting there.
Basic works as well from the desktop.

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Tue Aug 01, 2017 7:23 pm

Major success!

Unplugging the ROM module "ARM3Support" (previously dormant) which was somehow previously installed the machine now boots through to the desktop when using the ARM3 upgrade together with RISC OS 2.01.
Going back to RISC OS 3 (very likely 3.11) the problem remains as it was ... selftest followed by a black screen.

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Tue Aug 01, 2017 9:09 pm

shock__ wrote:Major success!

Unplugging the ROM module "ARM3Support" (previously dormant) which was somehow previously installed the machine now boots through to the desktop when using the ARM3 upgrade together with RISC OS 2.01.
Going back to RISC OS 3 (very likely 3.11) the problem remains as it was ... selftest followed by a black screen.

Great test results and very interesting indeed!

ARM3support switches on the ARM3's instruction cache (which runs at the ARM3 clock speed of the on-board oscillator/2 = 25MHz for the WE card). Without this module, under RISC OS 2 the ARM3 is effectively running as an ARM2, at the host clock speed.

Is your ARM3 board known to be fully working (eg. tested in another Archimedes such as A3x0 or A4x0)? Do you know the history of it - has it been overclocked in the past?

I've not come across a failed ARM3 before, doesn't mean it can't happen, but I'd check the 50mhz oscillator and the divider. If either have failed, this could be the problem.

User avatar
IanS
Posts: 477
Joined: Mon Aug 31, 2009 6:02 pm

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby IanS » Tue Aug 01, 2017 9:19 pm

shock__ wrote:I've tried my luck with and without the expansions, I have gone back to fully stock for now (1MB, no IDE). No change.

Thanks about letting me know to reset the CMOS by holding R - that's something I would have missed. That doesn't seem to help either I'm afraid (as said getting the "key stuck" screen doesn't happen either). I've also tried 'delete' which doesn't do anything either.


Del is probably better. It resets evrything.

http://www.riscos.com/support/developer ... types.html

(I'm prettry sure the R/Del difference also applies to earlier RISC OS versions).

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Tue Aug 01, 2017 9:23 pm

shock__ wrote: my ROMs are 150ns, that should be borderline within the limits for the 10MHz bus, right

8mhz on the A3000, so you're fine. :)
shock__ wrote: EDIT3: Apparently the machine has a keyboard issue related to the connector ... I'll get into that at a later point.
Would resetting the CMOS give me any kind of feedback

Quite common to see A30x0 keyboard connector damage especially if there's been any battery leakage. This can lead to start up problems if the keyboard is sending spurious data, but you'd see that with ARM2, so probably not the issue here. New connectors are available, cje sells two types, and I recall a couple of threads on here about obtaining the correct connectors from various suppliers.
shock__ wrote: EDIT4: Holding T made the screen loose sync, resetting via R worked. "Branch through zero" error remains.
However calling the Desktop from the * prompt via "DESKTOP" makes it pop up ... so I think we're getting there.
Basic works as well from the desktop.

So have you had the desktop eventually running OK under ARM3? How long did it work for? Did you get any applications to run, or access the floppy drive? What happens if you switch the cache on /off manually using *cache on or *cache off?

Definitely think this sounds like a sticky oscillator or failing divider.

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Tue Aug 01, 2017 9:26 pm

IanS wrote:Del is probably better. It resets evrything.

http://www.riscos.com/support/developer ... types.html

(I'm prettry sure the R/Del difference also applies to earlier RISC OS versions).

Good point, power on+Del works under RO2, you may have a bit more *configure-ing to do after though, but it ensures everything is reset.

SteveBagley
Posts: 129
Joined: Sun Mar 15, 2015 8:44 pm

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby SteveBagley » Tue Aug 01, 2017 9:51 pm

Just a thought but did the A3000 ever ship with MEMC1 (or has one been used to repair it)? That wasn't compatible with the ARM3 iirc.

Steve

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Tue Aug 01, 2017 10:04 pm

SteveBagley wrote:Just a thought but did the A3000 ever ship with MEMC1? That wasn't compatible with the ARM3 iirc.

I don't think they did...but even if they had done, I recall reading the problem with MEMC1 and ARM3 was that it effectively forced cache cycles to the same timing as bus cycles, so it should work with ARM3, but there would be no speed up.

EDIT: I've actually just found a MEMC1 and associated PAL IC in my chip box, so I'll plan to try swapping out the MEMC1a on my A440 (with ARM3) when I've next got it in pieces, just to confirm it's ok...

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Tue Aug 01, 2017 11:57 pm

I've just ordered a new 74ACT74 and still have a 50MHz oscillator in my spares box, so I should be good to go in case the clock circuit is indeed broken.
I'll have to wait a few days tho since I don't own a scope that can reliably measure frequencies above 10MHz and will have to borrow one from a friend (I'm more of an 8bit/x86 person, since I never properly introduced myself, therefore a 20MHz scope usually is enough for me).
About the past of the accelerator - I got the info that it was bought from a reliable seller from the UK - sadly I have no means to test it in another machine, as Archimedes (especially anything but A3010) machines are quite uncommon here (Germany).

Just to summarize:
ARM3 + RISC OS 2.01 works, gives an error on boot "Branch through zero at &03811F1C (Error number &80000005)" and needs manual starting of the Desktop, seems to work fine otherwise - however when unplugging the ARM3Support ROM module the computer boots through to the desktop and can be used without any issues. Simtec IDE controller won't work tho (IDEFS stays dormant)
ARM2 + RISC OS 2.01 works without any issues. Simtec IDE controller won't work tho (IDEFS stays dormant)
ARM3 + RISC OS 3.11 passes the selftest and then hangs with a solid back screen, keyboard LEDs won't work, neither will it detect sticky keys.
ARM2 + RISC OS 3.11 works nicely without any issues. Simtec IDE controller works.

Additionally there seems to be some remaining battery damage regarding the keyboard connector which leads to the keyboard working unreliable or just when pushing the contacts into the connector. There's also some slight visual residue on 2 ICs and inductors near the battery. The original battery had previously been removed and the battery damage cleaned up by someone else, who did a mostly good job but didn't quite get everything (personally I prefer to remove and replace 'compromised' parts unless spares are hard to get).

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Thu Aug 03, 2017 2:17 pm

shock__ wrote:Just to summarize:
ARM3 + RISC OS 2.01 works, gives an error on boot "Branch through zero at &03811F1C (Error number &80000005)" and needs manual starting of the Desktop, seems to work fine otherwise - however when unplugging the ARM3Support ROM module the computer boots through to the desktop and can be used without any issues. Simtec IDE controller won't work tho (IDEFS stays dormant)
ARM2 + RISC OS 2.01 works without any issues. Simtec IDE controller won't work tho (IDEFS stays dormant)
ARM3 + RISC OS 3.11 passes the selftest and then hangs with a solid back screen, keyboard LEDs won't work, neither will it detect sticky keys.
ARM2 + RISC OS 3.11 works nicely without any issues. Simtec IDE controller works.

So from the above list, it looks like all is good with ARM2, as you said.

For ARM3 under RISC OS 2, it's not clear if the ARM3 cache is actually being enabled successfully when it reports the "Branch through zero" error. So when you manually start the desktop, it may still be running without cache. Can you try unplugging ARM3Support, restarting the computer through to the desktop, then press f12 and try *RMREinit ARM3Support (does this work?), then try *Cache On (does this work)?

If the above works, can you enter *BASIC and write a short timed loop like this:

Code: Select all

t=TIME:FOR i=0 TO 10000:n=SQR(i):NEXT:PRINT TIME-t

And compare the time taken with *Cache Off to the time taken with *Cache On (which should be much faster).

And it's perfectly normal to see the IDE drivers are not RISC OS 2 compatible, given that RISC OS 2 support was dropped pretty quickly by most suppliers.

shock__
Posts: 15
Joined: Sat Jul 29, 2017 9:21 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby shock__ » Thu Aug 03, 2017 8:45 pm

Good news everyone!

It seems you were spot on steve3000 ... the oscillator circuit was dead (one of the flip flops of the 74ACT74 was bad). Exchanged the 74ACT74 with a new one (and socketed and replaced the oscillator in the same go) and boom - RISC OS 3 is working with the ARM3 upgrade.
The ARM3Support ROM module gets activated straight ahead with RISC OS 3 (on RISC OS 2.01 it stayed dormant and hung when trying to active it) and the "cache on/off" commands work. I'll "benchmark" the machine to confirm all is good at a later point because of the keyboard only working extremely unreliable. Didn't check whether RISC OS 2.01 boots through because this machine is intended to be used with RISC OS 3 exclusively.

There still seems to be an issue with the keyboard connectors being wonky, but I'll open a seperate thread for that if required.

Thanks a lot for all your input and ideas folks :)

EDIT: above Basic programm outputs 34 with 'cache on' and 208 with 'cache off' - seems good to me :)

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

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby steve3000 » Thu Aug 03, 2017 10:59 pm

shock__ wrote:Good news everyone!

It seems you were spot on steve3000 ... the oscillator circuit was dead (one of the flip flops of the 74ACT74 was bad). Exchanged the 74ACT74 with a new one (and socketed and replaced the oscillator in the same go) and boom - RISC OS 3 is working with the ARM3 upgrade.

Great news, glad you've got it all working well now :D
shock__ wrote:Thanks a lot for all your input and ideas folks :)

That's what we're here for. :mrgreen:

enthusi79
Posts: 13
Joined: Mon Jan 11, 2016 10:08 am

Re: BBC A3000 + Watford Electronics ARM3 - installation woes

Postby enthusi79 » Wed Aug 30, 2017 9:57 am

As the one benefiting from shock's awesome work/skills, I can report that all works perfectly now!
Thanks alot. There are not many - certainly not in germany- that could and WOULD go through such a hassle ;-)
I am attaching some specs-info screenshots :)
Thanks alot to shock and also this community here!
Attachments
DSCF3799.JPG
DSCF3798.JPG
DSCF3797.JPG
DSCF3796.JPG
DSCF3795.JPG


Return to “hardware”

Who is online

Users browsing this forum: No registered users and 1 guest