Date : Sun, 14 Aug 2011 18:32:17 +0200
From : rick@... (Rick Murray)
Subject: Advice on restoring a dead Beeb?
On 14/08/2011 15:14, Peter Coghlan wrote:
> I'm not sure if your reply was meant to go to the list.
Yes. :-/ You might want to repost your message so we know what was said?
> Feel free to quote this on the list
Will do.
>> I meant optional as in the sense of it not being a *required* part of
>> the reset sequence (resetting after power-up without holding Ctrl, for
>> instance).
> Ah, I see. I somehow assumed we were discussing power up only but on
> rereading the original post, I see Chris says "reset" which suggests
> powering on or pressing break.
Yup, it's a bit ambiguous, but useful for diagnostic as the reset
appears to start, but not complete.
>> Thought: Does your keyboard have any links made in the location (usually
> I don't want to go messing with keyboard links right now
...what I meant was - are there links to configure the Been to auto
shift-boot?
> but I've just tried holding down shift and powering on. This results in
> the following: "BBC Computer 32K", the VDU7 beep and "Acorn DFS" followed
> by a hang while the machine attempts to access a floppy drive that is not
> connected.
Right. So it VDU7s. Okay, so we can rule out waiting on floppies.
> Hopefully Chris will get back to us and say what is on the screen as
> this might help a lot to nail down the problem.
Indeed - are we getting as far as initing the display? Flashing cursor?
Stuck? Gibberish?
> Sure, but I don't think Chris has said what is written on it. He does
> say ROM but it is not clear if he is distinguishing between ROMs and
> EPROMs either.
No, he identified the MOS by removing it, as the only labelled one was DFS.
>>> everything is connected in parallel to the data bus, almost anything
>>> could be the source of the problem, whether it is accessed or not.
>> Take the CPU chip out, you'll see the difference.
> The problem is that when something faulty is loading the bus more than
I would have thought a bad bus loading would have stalled the reset from
the beginning, with the CPU receiving bad opcodes.
That's what I meant about removing the CPU - then you'd see the
behaviour of *nothing* happening. That we get to keyboard LEDs and such
implies a good number of instructions executed, hopefully correctly.
> I eventually found the problem was due to a faulty 74LS139 (IC30) which
> was not driving one of the paged ROM select lines high properly.
Whoa. That must have been sweat & tears to diagnose!
> is more devious than we give it credit for. For example, even though
> we can probably assume that because the keyboard LEDs have been set up,
> memory has been zeroed, it is also possible that it hasn't.
True. But armed with a proper diagnostic of the machine (ie on our bench
with a logic probe...), we at least have the ability to work through the
startup bit by bit.
> We don't know that random garbage hasn't been written to part
> of the memory for example.
True - but does this matter much in the initial startup? I'd need to
look at the code to see. ;-)
Perhaps more likely is a hardware init failure due to decode failure.
But, this is speculation...
> monitor, the colour encoder in the beeb is a little cheap and cheerful
> and the inclusion of its output with the monocrome signal can cause
> significant reduction in quality on a decent monochrome monitor due
> to unwanted patterning.
Ah, that makes sense.
> I do think it would have been nicer to provide a PCB mounting switch
> or at least a pair of jumper pins instead of a pair of solder pads
Indeed, given the pins for setting (EP)ROM decoding in odd circumstances.
> Yes indeed but I would like to know what is on the screen and have
> some clarification on some other points before putting too much effort
> into figuring out exactly how far we are getting into the OS.
As I mentioned above, anything else is really speculation. We *need* to
know, bit by bit, in *detail* how the machine behaves from start-up.
Chris - if you are reading - do you have a video camera, decent digital
camera or mobile phone? Could you record it powering up twice - once
aiming at the keyboard LEDs, and once recording what (if anything) is
shown on-screen? Drop it on YouTube, let us know the URL. Then we can
see/hear its behaviour.
> Yes. On the other hand, it was a good move to check the power supply
> voltages were correct.
Certainly. And he measured at the connections on the board so it would
have been a power supply under load.
> disconnecting the power supply and checking it connected to a dummy
> load only before attempting to power on the logic. This is in case the
> voltages are high and would damage the electronics.
Hmmm... It's not as crazy as it sounds. I've never had a *computer*
power supply go wrong, but the power supply in my MDFS went
catastrophically wrong one day when I switched on. From the behaviour, I
can only surmise that some failure mode permitted live and earth to
become joined. This blow the fuse in the MDFS's plug, every single fuse
in the extension leads to the wall socket, and the 30A ring main fuse
exploded with enough force to blow the plastic cover off of the fuse box
and shatter the little brown bakalite holder.
In addition to that, it fried the line drivers in the Econet cards in
the A3000, the A5000, and the A310, plus the Acorn clock box on net #1
(RISC OS) and the line drivers on two of the ports of the SJ bridge.
Astonishingly, the slow network (#2) and the BBC micro were unaffected.
The SJ transient suppressor was connected to net #1 and earthed, I don't
know if this helped or hindered, but let's just say I have a
half-functional bridge and needed to replace a number of network cards.
The MDFS itself? I rescued the harddisc and the floppy drives. The
harddisc is still around (don't have much call for a SCSI harddisc, and
it didn't work in the FileStore) and the floppies were reconnected to my
Beeb. The main board showed signs of burnt transistors and exploded
capacitors.
I consider myself lucky that everything didn't get wiped out - the line
drivers valiantly took the hit. And - here's the bladder weakening one -
I would have turned it on using a metal key in a metal keyswitch that
was earthed. Thank god for rubber soled shoes and a rubber backed carpet...
Hundreds of MDFSs are around, so I guess rather than some sort of flaw,
I was just unlucky for suffering a nasty component failure.
> neither particularly rare nor valuable so I would not normally suggest
> going to that much trouble to test a beeb power supply.
I'd worry more about myself than the Beeb. When I was young and in
boarding school the big old telly in the dining hall failed (dot in the
middle of the screen). So we took it down took off the moulded cardboard
(?) cover that clipped into place, and looked to see if anything had
come loose or was visibly "pffft"ified. Somewhere along the way I had
the bright idea of moving out of the way that massive dusty red wire
that always seemed to be in the way.
Evidently my arms flew outwards, nearly punching my friend unconscious,
and I backflipped clear over one table and landed on (or rather through)
the one behind. I, of course, remember nothing between reaching into the
telly and having a hell of a headache and finding myself in the sanatorium.
I guess I ought to say "ouch". ;-)
Accordingly, one of the few things I will *not* go near is the inside of
a monitor/TV any further than basic focus/dimensional adjustments.
Doubly so with TVs where mains voltages turn up in all sorts of freaky
places - ever seen a TV that said "LIVE CHASSIS" on it? That means the
bit that looks like it ought to be 0v/GND is... very much not.
Anyway - for checking power supplies in unknown kit, I meter it at the
500V AC range and work my way downwards. And if I am at all unsure, I'll
drag out the extension flex and power it up in the middle of the
driveway (so if it blows...).
Thanksfully most of the kit I've obtained in the past has a bog standard
wall-brick supply. Indeed many of the things are 'standard' enough I can
run it through one of my own bricks.
Best wishes,
Rick.
--
Rick Murray, eeePC901 & ADSL WiFI'd into it, all ETLAs!
BBC B: DNFS, 2 x 5.25" floppies, EPROM prog, Acorn TTX
E01S FileStore, A3000/A5000/RiscPC/various PCs/blahblah...