9810 works, briefly



#2

Well, my TAS 9810 arrived today. Did you know that the rubber printer roller can disintegrate into goo so liquid that it actually soaks into the paper roll? Well, it can.

The card reader rollers had turned into something like brittle cork. Messy, but easier to deal with than goo.

Took it apart, removed PCVs, cleaned connectors, plugged everything back in and...it works! Came up, and aside from two bad segments in one digit of the "keyboard" register display, everything seemed functional. Bolted it back together and....nothing. Won't turn on, no power fan, no horrible smoking components, just as though it was receiving no power at all.

Fuses OK, power switch OK. No voltages on the nicely labeled measuring points on the power supply.

Any suggestions?


#3

Never mind...there's a short somewhere near the power switch. If I apply pressure to the lower right end of the keyboard, it works fine. Just a matter of tracking it down...


#4

Post a couple of pictures one you get it sorted - that's just a cool looking machine. Glad you got it working.


#5

Well, I probably won't get time to work on the short issue for a week or so, but here are some photos anyway. First, the 2011 quail chicks:

I catch and release quail chicks in my yard every year. The little buggers can run like lighting and make inertialess 90-degree turns. Considering I'm 55, I'm kinda proud I can still catch 'em.

What? Oh, the 9810! Well, the first thing I noticed was when I removed the printer paper:

Yeah, that's going to be a LOT of fun to clean up internally. The card reader, on the other hand, will be easier:

As you can see from this slightly blurry photo, the drive rollers have turned to some crumbly brown material.

The edges of the circuit boards, on the other hand, are pristine. Is this gold plating?

The keyboard/X register has one digit with two dead segments. I don't suppose these LEDs are still made...?

Here's another shot of the display...

Here it is in program mode. I have a simple little program I use to do basic functional testing; it calculates how long it would take to fly a ship from here to Pluto (using 3.7e9 miles as the distance) assuming it could accelerate at 1g halfway then decelerate the remainder. On an HP-67, this program takes several seconds to run. On an HP-41, it takes just under 2 seconds. On the 9810, it's instantaneous.

The primitive version of RPN is confusing: having the result of an operation end up in the Y register is weird; you have to roll the result down to X to perform a subsequent unary operation.

Here are the options the calculator came with.

The exterior of the machine is quite battered: stains, chips in paint and plastic, and so forth. I won't try any cosmetic restoration, though: it has earned its stripes, as it were.

I would like to get a math module, because it feels weird to have an HP calculator without transcendentals. I will also eventually repair the printer and card reader, and although I have Katie's writeup on the latter, I have nothing on the former.

Also, I have no mag cards. Maybe Katie will sell me one or two.

Edited: 15 July 2011, 4:20 p.m. after one or more responses were posted


#6

Looks cool! Thanks for putting those up.

It is indeed gold. Fortunately for us, HP was generous with it back in the day. I doubt all our old treasures would work so well today if they hadn't. I was restoring an old HP DVM a couple of weeks ago and the internal construction was very reminiscent of HP calculators. Lots of gold plating and, interestingly enough, one of those weird gold "comb" connectors like the ones between boards on the older calculators. In the DVM it was used to connect a PCB with a large hybrid circuit (A2D, signal buffers, etc.) to the main board.

Is the LED display a monolithic unit? Or can you swap, say, the "X" and perhaps "Z" displays? That would at least move the problem to a less noticed location and also confirm the problem is the LED, vs. a connector or something like that.

If you score a math module, perhaps you can start doing some curve fits on quail population trends. Always good to have a dissertation topic in your back pocket :-)

Thanks again,
Bob

Edited: 9 July 2011, 1:19 p.m.


#7

Quote:
Is the LED display a monolithic unit? Or can you swap, say, the "X" and perhaps "Z" displays? That would at least move the problem to a less noticed location and also confirm the problem is the LED, vs. a connector or something like that.

The display is a single, easily removed, plug-in circuit board that contains the drivers and all three registers (I believe there's a photo on the Museum's 9810 page). Each LED digit is individually soldered on.

I think it would be relatively simple to unsolder and swap the problematic digit with another one; I've done similar things on Classic series calculators (where the LEDs were in 5-digit segments). I'm not going to risk it here, though, because the best possible outcome it still a digit with missing segments, just in another part of the display, and the worst possible outcome is a display with more problems (if not completely dead).

It's still pretty easy to find an old Classic for spare parts; not so another 9820! Frankly I would be reluctant to try it even if I had several new LED digits for this display!


#8

There are 2 versions of the display board. They are totally interchangable, and differ only in the displays themselves. The rest of the electronics works in the same way.

The older one has 45 individual 7 segment modules fitted to it. The later one has 9 off 5-digit modules. These appear to be electrically the same as the displays used in the classic series handhelds (HP35, etc), but with the pins bent differently. But of course I don't recomend raiding an HP35 to fix an HP9810.

It sounds like you have the older display board anyway. In the absence of any spares, I'd do what others suggest and move the dud displays to a less-used position. Individual segments missing is almost always a problem with the displays themselves, of course.

#9

David,

Take a look at this thread for suggestions on how to remake the printer platen.

There are two different length cards made for the 9810: one about 6 inches the other about 10.5 inches. I have a few extras of the 6 inch cards if you need a couple of test with. I found some large 3M magnetic cards somewhere that I was able to cut down to around 8 inches and use successfully as well. (I'm not sure if I have any of those left or not. PM me on this.)

I need to do some repair work on my 9810 too, it's not booting up. But was 100% functional not too long ago....

-Katie


#10

Katie, thanks for the card offer, but Etienne Victoria is already sending me a couple (from France!), which should be all I ever need.

Hopefully your 9810's problem is as simple as mine was.

#11

It will actually record on any length of card (within reason). I think you can even get 9100 cards to hold a few program steps. There are optical sensors in the card reader that detect when the leading edge of the card has got past the head (start recording) and when the trailing edge is getting near the head (stop recording).

Another common card reader problems is that the bulbs for these sensors burn out. That's easy to cure and the bulbs are still available.

As for the machine that's not starting up, my procedure is basically to check all the PSU outputs, then check if the clock circuitry is running properly. You can power up the clock PCB on its own (either from a 5V bench supply or by plugging it into the machine with the other logic boards removed) and check the muClk and Bitclk signals. You should get 16 of the latter then 1 of the former, 16 of the latter and so on.

After that, I connect a logic analyser to the test connector on the CPU control board, trace the microcode, and sit down with the listing.


#12

Tony,

Thank you for the tips! Isn't there also a POP(not) indication from the 5 volt supply that might be the issue with a machine that doesn't boot up?

-Katie


#13

Yes there is a POP signal (Power On Preset, what is more commonly called a 'rst' signal now). It clears the microcode program counter, basically, to get the microcode starting at location 1717 (split octal).

But if it's stuck you'll find the problem when you try to trace the microcode with a logic analyser and find that it's totally stuck. I suppose it could fail to operate at all, and thus cause the processor not to start cleanly, but I think you'll find that when you try to trace the microcode too.

#14

Hey Dave/Katie,

I have the Mathematics Module. Is there any way to transfer some of the functions you want to a card? This would imply that I GET MY CARD READER GOING!

Just a thought.


#15

Quote:
Hey Dave/Katie,

I have the Mathematics Module. Is there any way to transfer some of the functions you want to a card? This would imply that I GET MY CARD READER GOING!

Just a thought.


Although at this point I know very little about the 9810, I suspect the various plug-in modules contain machine code, not keystroke sequences that could be recorded to a card, sigh.


#16

David,

Thanks for sharing your pictures, they (apart from the wonderful quail picture) look very familiar. I had a similar experience with my unit that required a major clean up due to a liquified printer roller... Thanks to Tony Duell's help it is now again in great condition, including new card reader rollers.

By the way, I should have a spare Math module. If you are interested you're welcome to have it. Just drop me a line at 'vdwateren at netscape.net'.

Congratulations with your aqcuisition and enjoy it!

Best regards, Eric

#17

Much more likely to be an open-circuit than a short. The power
switch directly controls the mains input, and if yuo get a short there you'll blow fuses.

The power switch itself can fail. I've never never had any success with taking them apart and cleaning them, nearly always there are broken plastic parts. Finding a mechanically-similar switch is a problem if you need to replace it.

Getting the swtich out is not totally trivial either. It goes like this:

Remove the complete keyboard assembly and turn it over so that it's resting on the keys. Have the the display window section overhanging the edge of the bench.

Unplug the interconnection PCB between the keyboard and the encoder PCB, then unplug the cable from the other end of the encoder PCB, undo the screws and remove said PCB.

Undo the 6 screws holding the keyboard to tbe bezel. 4 are easy to see at the sides, the other 2 are reached through holes in the keyboard PCB. Remove the keyboard from the bezel.

Undo the 6 screws holding the front strip to the keybaord. Remove the strip. Now you can see the 2 screws holding the switch in place.

The switch can be tested in the obvious way (look at the schematics...)

You may find that the screws were originally anchored with threadlock or similar. I find it best to clean up the threads using taps and dies (4-40 and 6-32 UNC) before refitting them. And don't bother with any threadlock, unless you are shipping the unit in an army tank it's not going to come apart by itself.

When you come to ressemble it, the only difficult part is refitting the bezel. I do it like this :

Put the keybaord into the bezel and fit the 4 side screws (the easy ones to get to) loosly. The bezel should be just lifted off the bench.

With long-nose pliers, fit the rear center screw, fiddling it between the 'switches'. Put a screwdriver through the hole in the PCB, locate the screw and start it into the bezel thread.

Put the remaining screw on the screwdriver tip and 'ram' it into the front hole in the PCB. If it doesn't locate, shake it out and try again. With all screws located, tighten them up evenly.


#18

Quote:
Much more likely to be an open-circuit than a short. The power
switch directly controls the mains input, and if yuo get a short there you'll blow fuses.

The power switch itself can fail. I've never never had any success with taking them apart and cleaning them, nearly always there are broken plastic parts. Finding a mechanically-similar switch is a problem if you need to replace it.


Yeah, looking at the switch carefully I don't see any obvious problems with the solder connections, so it may well be a mechanical issue inside the switch. The switch doesn't
feel broken when you operate it, though, so I'll want to look at this very carefully.

Given the generally horrible cosmetics of this example, I have no problem with replacing it with a functionally identical switch mounted in a hole drilled in the right lower side of the case. I don't have the machine apart now but as I recall the switch has 6 contacts, 4 of which are in use, and a resistor bridges each pare of contacts. Normally I'd call this a "triple pole single throw" type switch, but hopefully you can tell me what it REALLY is...

(Yes, I've ordered the CD with the schematics, but I assume it'll be a couple of weeks to get here...)

Edited: 10 July 2011, 11:13 a.m.


#19

In England we'd call it a 'Double Pole Double Throw' switch. Or 'DPDT', 'Double Pole Form C', 'Double Pole Changeover'. Electrically, it has to be rated to switch mains voltage (250V) at about 2A.

Mains comes in on the middle pin of each side of the switch (which is the 'moving contact'. The contacts that are made when the switch is in the 'on' position feed the mains to the voltage selector and transformer. The contacts that are made in the 'off' position connect a 1M resistor across the output of the mains filter to discharge the capacitors in said filter when the machine is unplugged. I believe this was a prodution modification, early machines don't have this resistor (and you could, if unlucky, get a small shock from the pins of the mains plug after unplugging the machine).

For testing you can bridge the switch by soldering a wire link between the middle pin and the 'end pin' carrying a wire on each side of the switch. That is one link between the grey wire and the white/brown/black wire and a second link between white/grey and the white/brown/red wires. Remember you are dealing with the mains so be careful nothing can sort to things it shouldn't.


#20

I'm on the road now but will check this out when I return, and make sure that I'm not about to smoke the PS.

#21

Lacking an exact switch replacement, one would likely do better bypassing the existing switch electrically. A power cord with in-line switch would replace the function. This would also preserve the possibility of restoring all back to original status, should an exact replacement be found later.


#22

Quote:
Lacking an exact switch replacement, one would likely do better bypassing the existing switch electrically. A power cord with in-line switch would replace the function. This would also preserve the possibility of restoring all back to original status, should an exact replacement be found later.

True: since the switch connects to the rest of the machine via a 4-pin Molex connector, it's easy to wire around. All I need to know is which of the four pins on the mating connector I need to short to turn it "on", and then I can build a shorting plug to keep in on internally, and use the switched power cord as per your suggestion...


#23

It's actually a 5 pin Molex (I think it's Molex and not AMP) connector, the middle pin is not used. It's non-trivial to find this housing now too.

But numbering the pins from either end, the switch connects 1 to 4 and 2 to 5 when it's turned on.


#24

Hm. Well, it would be trivial to replace the 5 pin Molex with a 4 pin version...in fact I could probably just push the pins out of the existing connector and into a 4 pin shell, allowing future restoration of the Historically Correct Part.

#25

When I see stuff like this, I rejoice, because it means that even someone of my extremely limited hardware skills can deal with it:

One wire had corroded and come lose from the middle connector. I will note that the quality of the hand soldering on this switch was not the best: note the "solder ball" appearance which means that the connections weren't heated well when the solder was applied. Still, it works and it was easy to strip a few mm of insulation from the broken wire, clean off the existing solder, and re-solder it.

My power switch now works perfectly.

BTW, getting that last keyboard retaining screw in initially frustrated me: the screws are non-magnetic, so that trick doesn't work. Then I had an inspiration: dip the tip of the screwdriver in platen goo, touch to screw, insert! Works a champ.


#26

My power switch fell apart completely. I salvaged the metal frame that was holding the switch together and mounted a sub-mini toggle switch in that frame. In this way the two screws that were holding the switch in place still are holding it in place.

My problem now is that the machine needs a while to warm up before it starts working. This can take a few minutes if it's cold or about 2 seconds if it's warm. I don't know what's wrong but once it's on it stays and and works 100%.

What the normal delay from when the power switch is turned on to when you see all zeros on the display?

Thanks,
-Katie


#27

Quote:
My power switch fell apart completely. I salvaged the metal frame that was holding the switch together and mounted a sub-mini toggle switch in that frame. In this way the two screws that were holding the switch in place still are holding it in place.

That's probably more elegant than anything I would have done. Hopefully the submini toggle can handle the current.

Quote:
My problem now is that the machine needs a while to warm up before it starts working. This can take a few minutes if it's cold or about 2 seconds if it's warm. I don't know what's wrong but once it's on it stays and and works 100%.

Do you just have to wait, or do you have to turn it on, let it warm up, then turn it off and on again? I can only suspect this is a power supply problem (low voltage?) since none of the electronics should be temp sensitive, unless maybe it's a case of "heat expansion making a contact that should be made". But considering my minimal hardware knowledge, someone else should chime in.

Quote:
What the normal delay from when the power switch is turned on to when you see all zeros on the display?

3 seconds or a fraction under. All the status LEDs on the keyboard light up immediately, and the display follows shortly thereafter.


#28

Thanks for that. The sub-mini toggle switch is rated for 3A 125V, it should be fine.

I let the machine stay on for a long time and now it's working perfectly, even when it's cooled down to room temperature. I suspect that the issue is/was with one or more electrolytic capacitors. Without use, over time they loose their ability to store charge and need to be "reformed". If they're not too far gone reforming can be done at their normal operating voltage -- i.e., "the calculator fixed itself".


Edited: 13 July 2011, 10:15 p.m. after one or more responses were posted


#29

Well, replacing the electrolytics on the PS boards should be pretty simple...


#30

I agree, replacing the caps is simple, but I'm not so sure it's the power supply ones because the voltages were all very steady with very low ripple. Anyway, if it acts up again I work on doing just that.


#31

In other news, I received my O rings from smallparts.com today. Now I have a bag of 100, if anyone needs any spares. Using your instructions I'm hoping repairing the card reader will be relatively simple. SR-60 cards physically fit in the slot with only a mm or so of clearance on each side, so perhaps they will work.


#32

The card reader in the 9810/20 is remarkably robust, the SR-60 cards should work. BTW the o-ring fix to the reader in my calculator has never failed to work in the past 10 years. Not that I was using it much but there's no deterioration or slippage over time.


#33

A question on card reader repair (yeah, I know it's been 10 years):

Access to the drive wheels are blocked by the black plastic card guide:

The bottom end of said card guide is secured by a snap ring:

I have snap ring pliers but that sucker's on pretty tight and I'm worried about damaging it. Your instructions don't mention this, but it would seem a necessary step to getting to the drive wheels...unless I'm doing something wrong.

Hints?

Update: Never mind! The solution is to unscrew the shaft securing the bottom part of the card guide from the drive belt side of the reader. Duh.


Edited: 15 July 2011, 2:09 p.m.


#34

And...success! Wrestling the O rings on was tedious, but patience and many, many tries won out.

I tested the reader feed mechanism by manually running an SR-60 card through it, turning the motor by hand. OK, that works.

Reassembled everything, wrote a small program, stuck the SR-60 card in and pressed RECORD. The card's entire length fed through and stopped.

Turned the calculator off, waited 30 seconds, turned it back on and verified program memory was clear. Inserted the card and pressed LOAD.

Success! The 20 or so program steps read in and executed perfectly.

The SR-60 cards are long, but otherwise fit as though they were made for the machine. Makes me wonder...


#35

Quote:
The solution is to unscrew the shaft securing the bottom part of the card guide from the drive belt side of the reader

That sounds about right. Ten years is a long time for my memory cells to retain something, I'm glad you figured it out!

It's a fast calculator and good to write programs for. It's easier if you have the alhpa/printer ROM to see nemonics for your listings and fun to use the FMT key to print alphabetics from programs. The fan noise is loud though, which detracts from long-term use. I suppose I could slow it down or replace it with a quieter one, but I don't want to take a chance on it overheating and shortening the IC's lifetimes.


#36

It is startlingly fast, much faster than, say, an HP-41 or 42 and about a zillion times faster than the SR-60.

It's amazing how well the card reader works. I haven't had a bad read or write yet. Just need to remember to end programs with END instead of STOP or RTN, because otherwise the card reader will keep on running...

#37

Yes, I had a paper platen roller turn to goo (in an HP9820, which is mechanically identical). Based on some advice in this forum, I used 3M Cold Shrink to replace it. You want the smallest size cold shrink kit, which contains 2 rubber tubes (which you have to force onto the spindle) and the cold shink itself. Put that over the sleeves, pull out about half the support (this will make sense if you are doing it), then put the remaining cold shrink on top as a second layer. Finally trim to length.

Cleaning off this goo is the big problem though. You hae to completely strip the printer. It gets on the printhead, the driver PCB (at the bottom), and even onto the main backplane and keyboard connector. I found propan-2-ol helps in cleaning it up, but it still takes a lot of elbow grease.

For the card reader roller, I use a rather different technique to the one normally suggested here. I remove the old hubs with a puller and then fit a new, machined brass hub with grooves to take a pair of 1" outside-diameter O-rings.

Sometimes I can't get the hubs off without damaging the spindle, so I have to replace that too (1/8" diameter, I use stainless steel).

The only downside to this method is that you need a small lathe.

My datafile articles on the HP9800s do cover repairing the printer and card reader, but in part 8, which will (hopefully) be published at the end of next year. I don't suppose you want to wait that long :-). But I will be happy to talk you trhough any repairs.

There are some pictures in my flickr account of repairing the printer and card reader. Actually listed as for an HP9820, but it's the same.


#38

I can't locate your Flickr images using your name or various permutations of "HP 9820". Pointers...?


#39

Try tony_duell :-)

I don't see why this URL should be banned, the link is :

http://www.flickr.com/photos/tony_duell/


Possibly Related Threads…
Thread Author Replies Views Last Post
  HP Prime works from the traditionally TI community... debrouxl 1 1,670 12-06-2013, 09:00 PM
Last Post: Erwin Ried
  William Wickes briefly introduces using the HP 48S/SX Peter Murphy (Livermore) 11 3,208 10-28-2012, 09:46 PM
Last Post: Luiz C. Vieira (Brazil)
  HP48GX everything works perfectly except the 'ON' key? Bruce Larrabee 5 2,081 10-22-2012, 05:41 AM
Last Post: Bruce Larrabee
  Program works in emulators but not on HP-15c hardware Marcel Samek 32 8,008 05-11-2012, 08:46 PM
Last Post: Jeff O.
  15c LE - PSE works and doesn't! Please explain! Alexander Oestert 6 2,043 09-28-2011, 03:23 PM
Last Post: Alexander Oestert
  Original 15C Keyboard Test Works With 15C LE!!! DigiGal 5 2,251 09-26-2011, 07:33 PM
Last Post: M. Joury
  Removing 9810 printer snap rings (circlips) David Ramsey 5 1,852 09-06-2011, 01:32 PM
Last Post: David Ramsey
  WP 34S: New Release in the Works Marcus von Cube, Germany 22 5,693 08-30-2011, 03:48 PM
Last Post: Marcus von Cube, Germany
  Which cold shrink kit for 9810 printer platen? David Ramsey 1 995 07-19-2011, 04:06 AM
Last Post: Tony Duell
  9810 on TAS David Ramsey 25 5,998 06-30-2011, 04:16 AM
Last Post: Tony Duell

Forum Jump: