Threaded Mode | Linear Mode

robert rozee · 02-17-2014, 07:46 AM

(02-16-2014 07:29 PM)dramsey Wrote: Cool. Yes, that's the transistor. I will start testing and see what I can come up with.

don't bother removing them from circuit. just use a multimeter set to the diode-check range (if it has one) to find the transistor that in-circuit measures different to the rest. the dud one will be fairly obvious, with a near dead-short between emitter and collector. the next step will be to remove this transistor, reassemble, and test. with the shorted transistor removed the 'stuck' columns should turn into blank columns.

as for replacements, a bc557 (general purpose PNP) is likely to be fine. by the sounds of it the application is nothing special.

cheers,
rob :-)

dramsey · 02-19-2014, 03:19 AM

(02-16-2014 09:01 AM)Tony Duell Wrote: There are several ways to repair the card reader (and BTW I have never seen a good original roller after all these years!). The method I use is anything but simple, but it does work, and it is likely to stay working.

Basically, I strip the reader and remove the old roller hubs if I can. That needs a special puller tool that I made, any other method is likely to do damage. Sometimes the old roller will not shift without the spindle getting bent, even with a puller.

Anyway, I then turn a new hub, 7/8" in diameter with 2 grooves in it for O-rings (3/4" ID, 1" OD, 1/8" thick). Fit that onto the spindle with a grub screw to hold it -- I mill a flat on the spindle for the screw to go against. Then put it all back together.

Of course if the spindle was bent I have to make a new one of those too.

Doing this takes an afternoon or so, but it's going to stay working, and next time the reader fails it's just a matter of changing the O-rings.

Oh, I also suggest replacing the O-ring that's used as a drive belt on the reader. They do tend to slip after all these years.

Sadly I have neither the equipment nor expertise for this level of mechanical work. Still, I repaired my existing 9810 card reader with Katie's simpler process a couple of years back, and it still works fine (I just checked). Now that I have dozens of 9800-series mag cards, I may even try some longer programs.

dramsey · 02-23-2014, 04:32 AM

Well, my new 9810 is now 100% completely functional.

Rebuilding the card reader was easy enough-- it's much easier to work on the 9810/20 readers than it is to work on a 65/67/97 reader. I used Katie's method since I have a huge bag of the proper O-rings. I did have to steal the drive O-ring off my existing 9180 since I only had one of those (but I've order more of them, just in case.)

Interestingly, after I finished all this, I found a complete card reader with electronics board and what appear to be the original rubber rollers in one of my parts lockers. I don't think I've ever tested it; would have been nice to find it while I had everything apart. The rubber rollers seem to be OK, weirdly. Oh well, I'm sure I'll remember it next time.

My original 9810's printer platen had disintegrated into a mess of gooey tar. I have it mostly disassembled but eventually gave up working on it; I was never able to figure out how to remove all the tar from the delicate rubber paper advance belts (are there any modern parts that could replace these?) and the parts still reside in various plastic bags. However, the associated controller board didn't have the stuck transistor my new controller board did, so plugging it into the new calc resulted in a fully functional printer.

(Somewhere in England, Tony is shaking his head...)

I labelled the faulty board and put it in the parts pile.

So now I have:

-- One fully functional 9810 (with Printer Alpha!)
-- One mostly functional 9810 (a couple of dead segments in the bottom line of the display; card reader needs new drive O-ring, unlikely ever to have a working printer)
-- One parts 9810 (doesn't really work at all)
-- A pile of 9810 boards and the mysterious card reader
-- Insane pile of manuals, documentation, and mag cards

It's not even my favorite HP. But maybe I can bring it to wherever we have the meeting this year.

**Katie Wasserman** · 02-24-2014, 05:37 AM

I'm sure Tony's not pleased, but by swapping boards you're being true to HP's repair methodology as outlines in their service manuals. As far as fixing your broker printer mechanism goes .... I was able to clean off all the goo from the parts and make a new platen from scratch in my 9810. Here's an old thread discussing that and other ideas.

dramsey · 02-24-2014, 08:21 PM

Katie, the referenced thread mentions that you cleaned the goopy stuff, but doesn't say how.

When I was trying it, I was using 100% alcohol (sold as a cleaner at hardware stores) and acetone. The latter works a LOT better but as you know can dissolved rubber and some plastics.

Alcohol works, but slowly. I was nervous about leaving the thin, delicate rubber belts in an alcohol bath for the day or so it would take to work. That's about how long it took to get the goop off the printhead, and honestly I'm still not sure about it.

Any hints would be appreciated.

**Katie Wasserman** · 02-25-2014, 01:35 AM

I really don't remember is was a long time ago. But I typically would use 99% isopropyl alcohol first. If that doesn't work I sometimes use Goof Off or something equally horrible. But I would only use that on metal parts.

I do remember that it took me a very long time to get all the goo off, so I probably just used alcohol soaked gauze pads.

robert rozee · 02-25-2014, 08:05 PM

try straight d-limonene, or if you prefer a mixture of 20% d-limonene and 80% IPA.

dramsey · 02-25-2014, 11:10 PM

Ah, "orange oil", eh? So that works well in your experience?

Tony Duell · 02-26-2014, 08:11 AM

(02-23-2014 04:32 AM)dramsey Wrote: Well, my new 9810 is now 100% completely functional.

Excellent, even if you did it in a way that I do not approve of!

(02-23-2014 04:32 AM)dramsey Wrote: Rebuilding the card reader was easy enough-- it's much easier to work on the 9810/20 readers than it is to work on a 65/67/97 reader. I used Katie's method since I have a huge bag of the proper O-rings. I did have to steal the drive O-ring off my existing 9180 since I only had one of those (but I've order more of them, just in case.)

The HP9810/20 card reader is a lot larger and thus a lot less fiddly to work on than
the ones in the handhelds, but I find it takes me longer because I do make the new
roller hub. The reason I do that is that I am worried the new O-rings could slip sideways on the original hub, there is nothing to hold them. By turning a hub with a couple of grooves in it, I know the O-rings will stay in place.

(02-23-2014 04:32 AM)dramsey Wrote: My original 9810's printer platen had disintegrated into a mess of gooey tar. I have it mostly disassembled but eventually gave up working on it; I was never able to figure out how to remove all the tar from the delicate rubber paper advance belts (are there any modern parts that could replace these?) and the parts still reside in various plastic bags. However, the associated controller board didn't have the stuck transistor my new controller board did, so plugging it into the new calc resulted in a fully functional printer.

Yes, the printer platens decay too. I use the method suggested by Katie et al, that is to use 3M Cold Shrink. The smallest size kit of that contains a pair of rubber sleeves and the Cold Shrink itself. You force the rubber sleeves onto the original spindle, then put the Cold Shrink over it -- 2 layers. The only hard part is trimming the new roller to length.

(02-23-2014 04:32 AM)dramsey Wrote: (Somewhere in England, Tony is shaking his head...)

Hmm... That's the polite way of putting it.

(02-23-2014 04:32 AM)dramsey Wrote: -- One mostly functional 9810 (a couple of dead segments in the bottom line of the display; card reader needs new drive O-ring, unlikely ever to have a working printer)

The only part of the printer you can't really do anything about if it fails is the printhead. Every other part can either be obtained or repaired. So don't give up on this one. Apart from the gooey platen, what other problems does it have?

(02-23-2014 04:32 AM)dramsey Wrote: -- One parts 9810 (doesn't really work at all)

Again, there are very few HP9810 problems that can't be fixed. ROM failure is the main one, actually. All other chips were standard at the time, although some are going to be hard to get now (the RAMs, and some of the more obscure TTL parts). And IMHO debugging one of these machines is fun!

(02-23-2014 04:32 AM)dramsey Wrote: It's not even my favorite HP. But maybe I can bring it to wherever we have the meeting this year.

It is one of my favourite HP's. I think I'll put the HP9830 first, then the HP9100, then this one. It's a very interesting design.

Tony Duell · 02-26-2014, 08:15 AM

(02-24-2014 05:37 AM)Katie Wasserman Wrote: I'm sure Tony's not pleased, but by swapping boards you're being true to HP's repair methodology as outlines in their service manuals. As far as fixing your broker printer mechanism goes .... I was able to clean off all the goo from the parts and make a new platen from scratch in my 9810. Here's an old thread discussing that and other ideas.

Actually, not quite true-to-original. The official way was to replace the entire printer, the boards were not replaced separately. It may have made sense then (although replacing a printer because a transistor or cheap TTL buffer IC has failed seems a bit excessive, tracing this fault wouldn't take long), but today it makes no sense at all, you can get the transistors and buffer ICs easily, you can't get complete HP boards.

Tony Duell · 02-26-2014, 08:17 AM

(02-25-2014 01:35 AM)Katie Wasserman Wrote: I really don't remember is was a long time ago. But I typically would use 99% isopropyl alcohol first. If that doesn't work I sometimes use Goof Off or something equally horrible. But I would only use that on metal parts.

I do remember that it took me a very long time to get all the goo off, so I probably just used alcohol soaked gauze pads.

I used propan-2-ol (isopropyl alcohol) to clean up mine. Yes, it did take some time, but it did work. I know I left the printhead soaking in the stuff, probably the metal parts too. I can't remember what I did with the belts, but I know the originals went back in. Flat belts like these are not easy to find, although surely somebody must still make them. I will have a look around...

dramsey · 02-27-2014, 05:33 AM

(02-26-2014 08:11 AM)Tony Duell Wrote: The HP9810/20 card reader is a lot larger and thus a lot less fiddly to work on than the ones in the handhelds, but I find it takes me longer because I do make the new roller hub. The reason I do that is that I am worried the new O-rings could slip sideways on the original hub, there is nothing to hold them. By turning a hub with a couple of grooves in it, I know the O-rings will stay in place.

I haven't had any problem with them slipping off yet; of course, these machines don't really get used that much. I've used the same technique (different O-rings) to repair an SR-60A card reader.

(02-26-2014 08:11 AM)Tony Duell Wrote: The only part of the printer you can't really do anything about if it fails is the printhead. Every other part can either be obtained or repaired. So don't give up on this one. Apart from the gooey platen, what other problems does it have?

I really don't know, Tony. Since the printer parts were resting in a bath of Lovecraftian Chthulu-goop, I wasn't able to perform any tests. Although I know now that the driver board works perfectly since it's in my new 9810.

(02-26-2014 08:11 AM)Tony Duell Wrote: It is one of my favourite HP's. I think I'll put the HP9830 first, then the HP9100, then this one. It's a very interesting design.

I have all of these machines. The 9830 was the one that was closest to "perfect condition" when I obtained it, although it had a problem in that it would not reliably write tapes. This was fixed with a board-swap from a dead 9865 external drive (sorry). It has the 9866(?) thermal printer on top, which means I never need to worry about anyone stealing it.

I notice the keyboard on the 9810 is one of those wherein pressing a key simply moves a metal disk close to a printed coil on the underlying PCB, presumably registering a key stroke through some sort of inductive coupling. I imagine this makes the keyboard very reliable mechanically, and I also imagine the supporting circuitry is kinda complex.

Tony Duell · 02-27-2014, 08:00 AM

(02-27-2014 05:33 AM)dramsey Wrote: I notice the keyboard on the 9810 is one of those wherein pressing a key simply moves a metal disk close to a printed coil on the underlying PCB, presumably registering a key stroke through some sort of inductive coupling. I imagine this makes the keyboard very reliable mechanically, and I also imagine the supporting circuitry is kinda complex.

The keyboard is quite interesting. There are 2 spiral PCB tracks under each key which form a pulse transformer. The keys are connected up in pairs with the 2 transformers in antiphase. So that normally, when a drive pulse is sent through the primary windings (tracks), the induced pulses in the secondaries cancel out.

When a key is pressed, the metal disk damps that transformer, so it produces little output. Therefore the signal from the other transformer gets through and is detected.

The primaries are wired up in matrix, there's a scan counter and decoders to send a pulse through each pair in turn. The secondary pairs are all wired up in series, and connected to the input of a comparator chip that detects the induced pulses.

The same keyboard design is used in the HP9810 and HP9820 (and I assume the HP9821), but AFAIK in nothing else. The HP9830 uses a conventional set of (Cherry) keyswitches.

Tony Duell · 02-27-2014, 08:09 AM

(02-27-2014 05:33 AM)dramsey Wrote: I have all of these machines. The 9830 was the one that was closest to "perfect condition" when I obtained it, although it had a problem in that it would not reliably write tapes. This was fixed with a board-swap from a dead 9865 external drive (sorry). It has the 9866(?) thermal printer on top, which means I never need to worry about anyone stealing it.

I am going to guess either the 09865-66562 (Blue/Red) or 09865-66563 (Blue/Orange) PCBs (As you must have noticed, the handles on these boards give the last 2 digits of the part number), probably the former. Unreliable writing (as opposed to not writing at all) might be something as simple as the write clock being way off-frequency.

Even HP expected you to repair the tape control boards (but not the drive) to component level in the field.

dramsey · 02-28-2014, 05:25 AM

So the 9830 actually used Cherry key switches? Makes me wonder if modern Cherry MX switches would fit.

The 9826 I have uses Maxiswitch key switches...in fact it uses the exact same switch used on the Apple //e keyboards. I like these switches because they're readily available (see: used on Apple //e keyboards), as well as being easy to disassemble for cleaning the actual metal switch contacts inside.

Tony Duell · 02-28-2014, 07:32 AM

(02-28-2014 05:25 AM)dramsey Wrote: So the 9830 actually used Cherry key switches? Makes me wonder if modern Cherry MX switches would fit.

Alas they're not even close. The HP9830 keyswitch is much taller. It was a standard keyswitch about 40 years ago, I got a surplus keyboard off some word processing thing that used them. It got re-wired for another of my computers (not HP) and I had a few swtiches left over, a couple of which ended up in my HP9830.

(02-28-2014 05:25 AM)dramsey Wrote: The 9826 I have uses Maxiswitch key switches...in fact it uses the exact same switch used on the Apple //e keyboards. I like these switches because they're readily available (see: used on Apple //e keyboards), as well as being easy to disassemble for cleaning the actual metal switch contacts inside.

That's useful to know. The 9836 will be the same, it's the same keyboard PCB.