-----Oorspronkelijk bericht-----
Van: cctalk-bounces at
classiccmp.org
[mailto:cctalk-bounces at
classiccmp.org] Namens Tony Duell
Verzonden: donderdag 4 maart 2010 21:33
Aan: cctalk at
classiccmp.org
Onderwerp: HP2631B printer
You may rememebr that a few weeks ago I asked ofr help on the
encoder of an HP2631B printer. I've not sorted it all out,
and will describe what I did...
The initial problem was a defective 2114 RAM (!) on the
procesosr board.
After replacign that, the printer rammed the carriage into
the side of the mechanmism, blowing the 2 motor driver
transistors and an overvoltage zenenr on the -20V line.
Furtehr tests showd there were no signals coming from the
shaft encoder on the carriage movement leadscrew.
I took the machine apart. Really apart. Not having the
special tools to remove the encoder form the leadscrew, I
found I could take them out together, You loosen the cap
screw in the collar on the carriage motor shafter, free the
ribbon cable from the cips onthe rear paper guide, and tkae
out the 3 screws from the encoder mounting plate. The whole
lot this comes out of the right hand side of the mechamism. I
could then remove the end cover from the encoder (carring the
IR emitter) and the little cover over the cable connections.
The HP service manual warns you not t oremove the carriage
rails or platten bar, since 'an important adjustment will be
upset'. I reckoned that my pritner was in a pretty bad state
already, and little could make it any worde, so I took it all
apart. I am wondering what the HP manaul is talking about.
The 2 carriage rails have their ends turned down to fit in
holes i nthe side plate, so theres no way they can be moved
by more than a few thou. And these ends are not eccentric, I
checked. The platten bar is locared by dowel pins to each
side plate. It can't move either. As far as I can see there
is no problem at all with completely stripping the mechanism.
This meant I could deal with that substance well-know to all
classic computer people -- sound-deadening foam that turns to dust.
Back to the enconder. The person who said 'sounds like the IR
emitter is out' was vey close. The IR emitter was indeed not
glowing. The reason was that it's conencted to the rest of
the encoder by 2 little pin sockets on the emeltter assembly
(encoder end cover) which fit onto wires coming out of the
enocderr boddy -- one of them seems to be the end of a
current limiting resistor... Anyway, one of those wires was
bent and not connecting to the emitter assembly. Easy to fix
when you know where to look!
I crelaced the shorted zneer diode on the PSU board, and did
some simple checks on the rest of the electronics. The DIP
switches on the priner logic PCB and HPIB conenctor PCB were
not reliable. Since an HPIB address swithc which doesn't set
the address you expect is going to be a curse, I replaced
them. I managed to re-stake the puchbutton assemblies on the
control panel PCB --
http://www.parts.agilent.com indicates
that type of switch (plastic housing heat-staked to a PCB
with gold contact pads) is still avaialbe, but only if you
return the instrument to Agilent for repair. Sorry, but no way...
With everything back together (apart from the motor driver
transistors), it was time to give it a go. I could now see
that the encoder was producing pulses, that the end sensors
worked, and that it wasn't trying to turn on both motor
driver transsitors at the same time. Powered down, fitted the
(expensive, 25A) motor driver transistors, and tried again
with the lesdscrew nut unscrewed from the carriage. The idea
was that if the motor 'ran away', it wouldn't slam the
carriage into the side plate.
Tjhis time on power-up the mtoor ran -- but at a sensible
speed, there were plento of pulses from the encoder, and by
just touching the leadscrew nut I found it was indeed trying
to drive the carriage to the left.
Time fore the real test. I ffittd the 3 screws holding the
leadscrew nut to the carriage, and tired again. The carriage
went to the home position, and the macjhine gave a long beep.
The frontpanel buttons did nothing apart from reset (which
repeated the initialisation) and On-Line which caused it to
beep again. I tried frobbing the paper-out switch, it made no
difference, so I guessed I had a real fault.
I spend 2 hours looking at signals. The peocessor was clearly
running.
The end sensor signals were fine. The enocder, position
counters, direction flip-flop, and so on all seemed to be
doing the right things. I was beginning to think i had a
nasty fault in the custom HP procesosr chip. And yet, it was
running the firmware, at least enough to run the carriage to
the home position, sound the beeper, and so on. Checking what
the processor ws trying to do to the pritner logic PCB
indicated it was reading the sensors and writng to the
carriage motor register, which made sense. It wasn't
randoming acccessing all the ports.
What had I missed? I went back to the paper out signal. It
was high (indicating out-of-paprr) at the input pin of the
3-state buffer on the printer logic PCB. It was low on the
motor harness pin on the PSU board (that makes sense, there's
a NOT gate on the PSU board which inverts this sigal). But it
didn't change state when I frobbed the microswitch.
Aha...
Although the microswitch is hidden inside the printer
mechanism, I managed to disconenct one of the faston
terminals from it (the switch is closed when out of paper).
This time when I powered the machine up, it homed the
carriage and didn't beep. I could do linefeeds and formfeeds
from the panel, the on-line button worked, and the self-test
seemed to be trying to print something (I'd not fitted the
printhead at this stage.
So the microswtich was faulty. Strangely it was stuck closed
(most switch porblems cause them to not make contact). I
removed the printer mechanism again, turned it over, and
removed the rear paper guide (4 screws). 2 mores screws
released the microswitch from the guide. And it didn't
'click' when I pressed the actuating lever. Unfortunately,
although it's a standard V3 size switch, the actuator is
unusual, so getting a replacement would be nnon-trivial. With
nothing to lose, I drilled out the rivet holding the swtich
together, took off the cover and remvoed the contacts. I then
fount that other substance well-known to classic computer
types -- grease that turns to cement. Cleaned it off,
cleanded thee contacts (well, while I had it apart) and
resassmbled it. Now it clicked. And an ohmmeter showet it was
woring electrically too.
Put it all back together again. Now it will initialise and
respond to the control paenl -- provided there's paper in it.
Time to fit the printhead (trivial), and it now makes that
well-known buzzing that everybody who's ever been near a
dot-matrix printer would recognise. Will it print anyhting
sensible? Well, let's try the ribbon. Which is jammed. The
ribbon cartridge is heat-staked together, but the hold trick
of pulling htr ribbon out and widing it back in got it free
enough to work. Now the self-test prints a character set --
and it looks quite sensible,
Tiem to try it with a computer. I grab my HPIB test set-up
(HP71 + HPIL module + HP82169 HPIL-HPIB interface) and cable
it all up. Set the printer address to 4 (yes, I used a PET in
the old days...) type PRINTER IS 4 and then PRINT. The darn
thing does a formfeed (!). Then try PRINT"0123456789". It
prints "0000444488" andanotehr formfeed. Clearly the 2 least
significat bits (bits 1 and 2 in HPIB terminology) weren't
gettign through (an unconnected HPIB line is high, which
corresponds to logic 0 on this bus). Hence the CR character
was becoming a formfeed...
Fortunately I'd picked an HPIB address where this wasn't a problem,
I hoped the HP custom PHI HPIB chip hadn't failed. I
disconnected the HPIB cabel from the HP82159 and removed the
HPIB interface PCB together with the connector PCB and HPIB
cable from the printer/And then did continuity checks from
the free end of the HPIB cable to the pins on the
3448 buffer chips on the HPIB PCB. Fortuneately bits 1 and 2
were indeed open.And a few morre detaild test showed it was
nothing more than dirty contacts on the H{PIB socket. A
cotton bud and propn-2-ol cured that.
And then it printed properly -- at last. The last job was to
fit the cover and platten knob, which was trivial -- at least
after everything else.
-tony