At 11:12 pm 22/08/2001 +0100, you wrote:
>> >> Hmm, possibly. I suspect it's going to involve many hours of slaving
over a
>> >> hot soldering iron, and TBH my soldering's just not that good. Besides,
>> >
>> >Well, this is good practice, then :-)
>>
>> True :) It does seem a shame to practice on such a rare piece of kit,
>> somehow...
>
>Err, are PETs rare? They seem pretty common compared to the stuff I
>learnt computer repair on (Philips P850). I've never thought of the
>8032SK as being a particularly rare PET either (It was the first CBM I
>obtained, which might affect my judgement here).
It depends which one... I've not seen many 8032SKs about (I still curse the
day I had to hand the working one I was using back to its rightful owner),
probably because they're either rotting away forgotten in lofts, or
collecters who have them won't give away/sell them. I don't blame them, the
8032SK is an inspirational piece of design, even now. Well, until you look
under the cover and see that it's a right hotch-potch...
>> >> Possibly. The trouble is - where can the problem lie? The machine is
either
>> >
>> >What test equipment do you have?
>>
>> Erm, a multimeter...
>
>Analoge or Digital (I use both, FWIW, as each type has its advantages).
Digital. I used to have an analogue, but it went missing many years ago.
>Any ideas as to the input resistance (or sensitivity), just in case it's
>seriously loading the signals.
'fraid not. I've got the spec sheet for it, but it's up at my Liverpool
pad, and thus inaccessible for the next couple of weeks.
>> I'd have to buy a 'scope, and they're not cheap...
>
>You can get _good_ second-hand 'scopes for less than a poor quality new
>one. Stewarts of Reading often have Tektronix 'scopes for a few hundred
>pounds (something like a 465 or a low-end 7000).
Which would you recommend? I'd pay ?200-?300 for something that's going to
do what I need, which is borderline new one (so long as I don't go for the
one and only model Maplins still do). I *may* be able to wheedle one out of
EATL (the place I briefly worked at so many years ago), I'll have to tickle
my contacts...
>An older, vavled, Tektronix 500-seires is another possibility (it's what
>I use), but be warned that collecting and repairing classic 'scopes can
>be as adictive as collecting classic computers, and you'll end up needing
>a 'scope to fix the 'scope, and so on ;-)
LOL. I'm running out of space - fast - as it is; no way do I need yet more
bulky gear!
>I notice you're in the UK, just out of curiousity where are you (I'm in
>SW London), just in case there's somebody near who could at least do a
>few tests for you.
Just up the road from Tower Bridge (E1) mostly; when I'm not down here, I'm
up near Liverpool. At the moment, most of my collection is here with me in
my 18'x12' flat; I shall be moving a good chunk of it up north soon,
however, it's getting to the point where I can't move in here!
>> >Start by checking all the power rails (at the pins of the chips) with a
>> >voltmeter. Get the PSU working first.
>>
>> Done & tested. Power is OK.
>
>Excellent. You'd be amazed how many problems are PSU-related, though...
I can imagine. 12v will fry a TTL chip, won't it? All the working PETs I've
got have some screen wobble, so it's possible that the PSUs are getting
borderline with age, unless that's what they always did - except through my
rose-tinted retrospecs.
>
>> >Then use a 'scope, logic probe or logic analyser (yes, OK, I have
>> >somewhat fancy test gear) to check the clock input to the 6502.
>>
>> Actually, I used the voltmeter for that; it reads approx half the normal
>> input voltage, which *suggests* that its OK.
>
>Did you discover that trick yourself? If some, I am fairly impressed.
Much as I would like to take credit, I can't. A chap on the cbm-hackers
mailing list suggested it.
>I've used it many times to check clock signals, etc when all I had was a
>multimeter....
This was before I discovered the frequency measurement available on my
multimeter. However, it looks like it's got a maximum of 99.9Hz, which will
not be suitable :(
>> >If
>> >missing, fix the clock circuit. Check the reset pin while you're at it,
>> >to ensure the machine is not stuck in the reset state. Without good
>> >power, clock, and reset, the machine is not going to do a lot.
>>
>> Reset is OK too, so theoretically (and assuming the clock really is
>> ticking, and not just stuck at a constant half voltage) the CPU is getting
>> what it'll need.
>
>OK... Is RDY mostly high at the CPU (or could this darn thing be stuck in
>a wait state)?
IIRC, I checked that accidentally (got the wrong pin when checking the
voltage), and it seemed to be stable at 5.23v; depending on how often it's
supposed to drop, I don't know if that's a good or bad thing.
>> >Then use the same instrument(s) to check the phi2 (bus clock output) of
>> >the 6502. And look for activity on the address/data buses.
>>
>> The clock outputs register lower voltages than the clock in (0.93v and
>> 0.75v). However, that tells me pretty much bugger all, I will need some
>> better kit to really examine those.
>
>Those sound low. I'd expect them to be 50% duty cycle square waves with a
>5V or so amplitude. But as you say, you need a 'scope to see what is
>really going on.
I'll give Stewarts a call tomorrow, if I remember. Are they likely to have
logic probes as well?
>> >When you've got this far, it's then time to see just what ROM(s) are
>> >being selected (logic probe, etc on the CS/ pins) and thus what the CPU
>> >is doing. Unless, of course you have a logic analyser that can look at
>> >all 16 address lines at once.
>>
>> This is realms of the unknown for me - I shall have to go look up logic
>> probes/logic analysers... I assume the same sort of checks can be done with
>> most chips, e.g. RAM, ULAs and so forth.
>
>Be warned a new logic analyser is an expensive instrument, and is likely
>to be way out of your price range. But a lot of older logic analysers,
>which 'only' go up to 100MHz are available second-hand. While they're not
>all that useful for many modern chips (I will not say 'useless', becuase
>I've used them to debug such systems), they are idea for 99% of classic
>computers. You might find one for \pounds 50.00 or less if you hunt
>around (say at radio rallies). Something like a Gould K100D would be
>ideal, I think.
Maplin do a single-point probe for about ?50; Farnells do a couple more,
but their full logic analyser is about ?500 (and looks suspiciously like an
oscilloscope with the knobs replaced with a keypad). I'm not out to spend a
grand on test kit *just* yet; especially not if there's some good s/h gear
around.
>You need a 'scope as well -- a logic analyser is only useful if the
>signals are known to be logic-level signals.
Did anyone ever do a combined logic-analyser/'scope?
>
>IIRC, the monitor needs drive signals from the 6845 on the mainboard to
>get the horizontal output stage running (and this whistling). So if there
>is a video section fault, you won't get any display or whistle. I would
>ignore the monitor problem for the moment and debug the logic side...
>
That makes sense. I'd guess that the monitor side is mostly analogue anyway?
Thanks - you've definitely got me hooked onto this electronics lark. Mind
you, that wasn't precisely difficult, I nearly went that way at school, but
side-tracked into software instead.
Cheers!
Ade.
--
B-Racing: B where it's at :-)
http://www.b-racing.co.uk
Received on Wed Aug 22 2001 - 20:07:40 BST