apple Lisa2. Any advice on non-working floppy drives?
ard at p850ug1.demon.co.uk
Sun Dec 5 15:09:22 CST 2010
> Tony, thanks for doing that bit of research.
> While I can see your point, I feel board swapping does have its place in
> some circumstances but let's put that aside. I don't want to fan flames (-:
Well, I might as well put forward my objections to board-swapping...
Board-swapping really has 2 meanings. One is replacing a PCB (or other
module) where you know exactly what's wrong with it eithe becasue you
can';t get the individual spare component or because you don't have the
skill/tools to fit it. For example, replacing a PCB witha failed
BGA-packaged device if you only have a normal soldering iron. Or
replacing a floppy drive with a damaged head becuae (a) you can't get a
spare head and (b) you don';t ahve the alignmet disk anyway.
Now that I dislike doing, becuase I prefer to replace as little as
posisble both to keep my machines as origianl as possible and to cut down
on waste. But it's less objectionable than the other meaning of 'board
Which is, of course, to repalce parts mroe or less at random until the
machine appears to work again. I have plenty of reasons for disliking it
1) Particualrly in the case of classic computers, you may well not have
any known-good parts. Swapping one possibly defectinve part for another
possibly defecrtive part is not going to tell you very much.
2) Defects in the amchine can damage the parts you're fitting. The
classic (but by no means only) example of this is a power supply problem
that is causing one of the supply rails to be overvoltage. You can end up
blowing every PCB you fit.
3) Even if the machine appears to work after you've replaced a certain
part, you dont know that was the problem, and that it's going to stay
working. It may have been a bad connection on one of the connectors that
has been temporarily fixed when you removed the old PCB and fitted the new
one. Or worse still, the problem could be in a totally different part of
the machine. Let me explain. Suppose you have 2 parts A,B that are
connected. And supposing there is some tolerance on a signal between them.
In fact let's say that it's a pulse train and it's the frequency that's
important. Module A outputs a nominal 500kbps pulse train (let's say it's
a disk drive ;-)). Module B receives it amd has some kind of PLL circuit
that has to lock to it. Now, suppose that A is running slowly, but within
tolerance. The PLL on B is drifiting in the opposite direction. In the
end, the system fails to work. You replace the drive first, and by chance
you pick one that's still withing tolerance, but towards the faster side.
The PLL can lock to that, so the system works again. But as you've not
cured the real fault -- the fact that the free-running freqeuncy of that
PLL is drifitng up, the machine fails again after a short while. Think
that sounds unlikely? I've seen it often enough for it to be a real problem!
4) Again particualrly for clasisc computers, you don't knowe that the
replacement part is the same revision level as the one you are replacing
and you don't know it's going to work correctly with the rest of the
machine. Yes, seen that one too. It led me a merry dance in a PDP11/45
until I discoered that the later version of one the CPU boards needed an
extra clock signal to be wire-wrapped on the backplane.
I learnt how to find faults _many_ years ago. And I learnt that the best
way to do it -- heck the only way to really do it -- was to collect
evidence as to what the problem was (that is, make measurements) then
think about that evidence and then (and only then) make any repairs.
Strangely that method still works today. And machines I've fixed, even if
I say so myself, seem to keep working...
> Yes, a checkout to see if voltages were where they should be on the LisaLite
> boards was going to be my next step (and also a check of the cables). I
I hate to say this, but that's the first thing I would have done, if only
to prevent further damage.
> just wanted to see if there was any common design flaw/fault people know if
> in either the boards or the drives which caused the symptom before I went
I am not a great beleiver in 'stock faults' either. Yes, if there is a
known weak spot, I'll check that first, but I won't be suprised if that's
no the fault or at least not the only fault. being able to fix, say, 95%
of machines by stock faults is great if you have a lot of such machines
to repair and can get most of them off the bench quickly. It's less
useful when you have just one machine to repair, and when murphey's law
dictates that itwill be in the 5% :-)
> any further. Collective knowledge might have been a shortcut to the problem,
> hence the post.
OK, rant over, let's get back to this machine...
> I have a manual with some schematics and tonight I'll check and see if the
> LisaLite board is there.
As I said a Google search found it fairly easily. The Lisalite board
looks quite simple (the schematic is just 1 sheet, and I recognisd all
the chips on it). I didn't do much more than glance at the schematic, but
it seemed to be a dairly simple PWM generator using TTL coutners and '85
comparators. I think the control vaule is bit-serially loaded into a
shift register on the board.
> >Firstly, since you have no idea as to the health of any
> > of the parts, you can't deduce anything from the fact that swapping them
> > out makes no difference.
> Not entirely true, as the Lisa 2 startup checks carry right through until it
> asks for a disk for inserted. So I can assume most parts are working.
Can you? I have no idea what these startup checks actually test, but I
would have thouht it as possible for one of the I/O chips that links to
the drive to havee failed but in a way that it still passes the tests.
> Incidently, the docs I have suggests this routine also checks the LisaLite
> controllers so (according to the machine anyway) these are ok. It's hard to
I wonder how? I couldn't see any way for the machine ot check that the
PWM signal (the only signal truely sourced by the Lisalite board) is
present and corret.
> believe that THREE drives all have the same fault though. All inputs into
> the drive are from the LisaLite board. Maybe the diagnostic checks are not
Not really. While the only external cable from the drive does, indeed,
plug into the Lisalite board, the shcmeatics show the most of the signals
are simply passed (with no buffering or anything) between the 'Twiggy'
connector back to the Lisa I/O board and the Sony drive connector. I
think it's likely that any problems iwth those signals are not caused by
the Lisalite board (broekn PCB tracks are not common).
> as thourough as they should be.
We don;t (or at least I don;t know) what hte diagnostics actually check.
There's also the issue that you're using a defective system to diagnose
itseld, and while most diagnostics are written assuming that anythign
that hasn't been checked could well be defective, some are not. I've seen
diagnostics that basically assume that the machine is working correctly,
and which don't help _at all_ if there is a fault.
I really do think you need to start making some measurements on the drive
connecotr and see just what is, and is not, correct there.
More information about the cctech