▼
Posts: 18
Threads: 2
Joined: Mar 2013
I've been recently given an HP 9830A (plus printer, now inoperable, and plotter). The calculator itself operated normally when powered on last time, some five years ago, but naturally I've been afraid of switching it on again. Seems it had been operated less than occasionally in last 15 years -- perhaps a dozen hours altogether. Of course I am aware of certain risk, but wouldn't wish to go against known hazards, such as depolarized capacitors.
Will you please tell me what maintenance and checking would you recommend before applying power?
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
At the very least :
Remove the top cover, PCB hold-down (1 screw) and keyboard (4 screws underneath).
Then remove all the PCBs and modules. The display board has 1 screw in the centre, the PSU board (rearmost in the machine) has 2 countersunk screws on the heatsink bracket. The tape drive is held down by 2 screws, the ROM module cage by 4 screws.
Inspect all parts for burnt/damaged components, etc. Check the setting of the voltage selector switches at the back, and the rating of the 2 fuses.
Put the PSU board back in place and fit the screws. Connect the keyboard mains switch cable (the 6 pin mate-n-lock), but do not connect the logic cable (paddleboard to backplane edge connector) yet.
Connect mains, and check the voltages at the marked testpoints on the PSU PCB (this PSU will run fine with no load).
If everything checks out, unplug it from the mains, put the boards back in place, connect the keyboard logic cable and refit the keyboard. Power up and see if it works. If not, you have a nice bit of logic troubleshooting to do.
There's the official service manual ('boardswapper guide') and 'my' schematics over on the Australian site.
▼
Posts: 18
Threads: 2
Joined: Mar 2013
Many thanks for this rapid and detailed reaction. So far I've just removed the top cover, being surprised how little dust is inside.
Do the PCB's with MOS circuits require any special cautiousness (higher than contemporary boards), as regards ESD ?
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
The HP service manual does have big warnings about ESD damage to the ROM and RAM boards, I suspect that's mostly because at the time many service engineers would be unfamiliar with MOS circuitry.
However, the ROM chips are (of course) HP custom and the 1103 DRAMs are very hard to get now. It makes sense to treat these boards with care. Handle them by the edges, put them on an anti-static surface, etc. Wear a grounding wrist strap if you have one.
I've worked on many 9810s and 9830s and never damaged a MOS chip.
▼
Posts: 18
Threads: 2
Joined: Mar 2013
I just heard that the first generation(s) of MOS IC's didn't have any ESD protection on chip terminals; this was why I asked.
Posts: 18
Threads: 2
Joined: Mar 2013
Quote:
Connect mains, and check the voltages at the marked testpoints on the PSU PCB...
Strange, I can't see a +5V testpoint (only +16, +19.5, and -12V ones).
▼
Posts: 18
Threads: 2
Joined: Mar 2013
On the three testpoints, the voltages are 18.6V, 22.7V, and -12.2V, the first two being substantially higher than the text on PCB reads. I think this is quite explicable, as the PSU is now totally unloaded. The +5V rail on the backplane is accurate. I think I might go ahead?
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
There may well not be a +5V testpoint, I forget (there isn't one in the HP9810/9820 supply for some odd reason). But if it's OK at the backplane, go ahead.
I wouldn't worry about the other voltages being a bit high. As you said the supply is unloaded. And they're not wildly out. I'd add the logic boards and see what happens.
▼
Posts: 18
Threads: 2
Joined: Mar 2013
Great! It powers on, shows the prompt, even calculates. As I have no keycaps on at the moment, it would be too clumsy to check something more complex.
Not all LED's are operable, though, and it seems the inoperable ones form columns (mostly, if not always). If I'm not mistaken, this concerns the 4th dot column of characters 4, 12, 20, and 28 (0 being the leftmost). Also, some 5th dot columns flicker temporarily, e.g. in chars 6, 10, 26, and 30. I heard of that vaguely before the machine came here, and thought that particular LED's were defective. This looks like a driver fault, therefore possibly reparable.
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
You've got a schematic of the display board, right?
The first thing to realise is that while it appears physically as a 7 row by 160 column matrix, electrically it's wired as 28 rows by 40 columns (this reduces the total number of driver circuits needed).
The second thing to realise is that the data is sent to the display driver board 2 characters at a time (12 bits). They correspond to 2 display positions 16 characters apart (so 0 and 16, 1 and 17, and so on).
Entire missing columns, particularly with the pattern you've described sounds like a driver fault, probably on the display PCB itself. This shouldn't be too hard to troubleshoot.
▼
Posts: 18
Threads: 2
Joined: Mar 2013
Sorry for the delay. Would you please verify my findings:
Referring to your schematic (Display PCB Sheet 3), the missing column is related to the signal path denoted as DispCol24, and to U39 pin 5. I suppose the U39 simply means the bottom row, 9th IC from the left. Then its pin 5 is led to the tenth Motorola 1858-0014 in the row above, pin 9. I can't find its internal structure anywhere, but assume it's a 4-transistor array, the 7 pins in each pin row being C - B - E - nc - E - B - C. The same structure I suppose in its NPN counterpart, one row above again.
Now a rather silly question: If I grounded U39/5 or shorted the U28c transistor (thus increasing the LED column's duty cycle by a factor of 40) for a second or so, is there a substantial risk of damaging the dot matrix (or anything else)? (I think I could also try to put a resistor between ColSel4 and the PNP transistor's base, which would be 8 times less violent.) Unfortunately I don't know how could I find the dot matrix datasheet, although I have some later HP ones piled somewhere.
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
Quote:
Sorry for the delay. Would you please verify my findings:
Referring to your schematic (Display PCB Sheet 3), the missing column is related to the signal path denoted as DispCol24, and to U39 pin 5. I suppose the U39 simply means the bottom row, 9th IC from the left. Then its pin 5 is led to the tenth Motorola 1858-
I numbered the ICs (U1, U2, etc) starting at the top left corner of the board and going by rows. I am not sure where U39 is, therefore.
Quote:
0014 in the row above, pin 9. I can't find its internal structure anywhere, but assume it's a 4-transistor array, the 7 pins in each pin row being C - B - E - nc - E - B - C. The same structure I suppose in its NPN counterpart, one row above again.
Yes. Simple NPN and PNP transistor arrays. If all else fails (and given that the connections are so simple) you could probably fit 4 TO92-packaged transistors in there -- I can't believe the charactertistics are that cricital
Quote:
Now a rather silly question: If I grounded U39/5 or shorted the U28c transistor (thus increasing the LED column's duty cycle by a factor of 40) for a second or so, is there a substantial risk of damaging the dot matrix (or anything else)? (I think I could also
I certainly would not try that. There is a risk of damaging the LED matrix, and those displays are essentially unobtainable now (unless you have a 'spare' 9820 or 9830 to raid for parts.
Is there any way you could borrow an oscilloscope? I think it would be quite easy to trace the signal from the decoder circuit through the transistors. Or maybe just test the transistors -- even checking the junctions as diodes using a multimeter would pick up open-circuit transistors.
Quote:
try to put a resistor between ColSel4 and the PNP transistor's base, which would be 8 times less violent.) Unfortunately I don't know how could I find the dot matrix datasheet, although I have some later HP ones piled somewhere.
Well, you have the pinout of the display, but alas no characteristics (peak current, etc) for the LEDs. But I would 'tread carefully' -- you really don't want to burn it out!
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
Sorry for following up my own post...
I've looked at the schematic. What I'd do, I think (if I didn't have test equipment) is to desolder the end of the 221 Ohm resistor that connects to the 74L42 output corresponding to the faulty column, and also that end of one other resistor. Then use wires to 'swap them round'. Put the display board back in the machine, power up and type some characters.
If column 24 now works and the other column is dead, it points to a problem in the 74L42 chip. If column 24 is still dead (and its pixels appear in the other column), then it would probably indicate one of the transistors has failed.
▼
Posts: 18
Threads: 2
Joined: Mar 2013
That's a very smart idea, indeed.
Personally, I'm not much happy with any soldering on this fine 4-layer PCB, perhaps the neatest one of the whole machine. (However, I am slowly getting reconciled with the probable result of the whole affair: a single TO92 soldered onto the bottom side of the board in parallel to a faulty transistor I surmise in one of these transistor arrays.) Possibly I could just interconnect two PNP-transistor bases (i.e. the ends of the resistors you mention), which could be accomplished even without soldering, using a mere clip of wire or something. This would OR two LED columns together -- I think they should survive the doubled duty cycle for a while. (I suppose the '42 wouldn't be able to pull the center of these resistors up too strongly in the situation when one end of that resistor pair is driven low and the other high.)
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
I wouldn't worry about soldering on the display board. The HP boards of that period were good quality, and provided you use a fine-tipped temperature-controlled iron you'll have no problems
(I desoldered all the connections on an HP9100B ROM board -- which is an older 14 layer board and where the layout of the tracks is critical so if you damage a via you can't repair it with a wire on the surface. Yes, it was a bit stressful (to me), but the ROM worked when I'd put it all back together).
I woudl be more careful about paralleling transistor junctions. It's possible one will take all the current and the other almost nothing. This won't do any damage here, but it will produce misleading results. And it's possible one of the defective transistors is shorted, not open, I think.
Posts: 18
Threads: 2
Joined: Mar 2013
Quote:
Is there any way you could borrow an oscilloscope?
I believe I can -- it might be more difficult to acquire probes small enough to be placed under the keyboard, which I would need to activate the particular LED's. Of course I could solder wires there to extract the signals, but I think then it would be simpler to try soldering a parallel transistor which may even solve the problem (if it is where I guess).
Quote:
Or maybe just test the transistors -- even checking the junctions as diodes using a multimeter would pick up open-circuit transistors.
I already attempted to do so last week (and compared them with their neighbours), but with no clear result. Seems the 22 Ohm resistor complicates that. I could try to use another 'meter, though.
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
The way to get better access to the display board is to partially dismantle the keyboard (this is not in the HP manuals, but it's not difficult).
Start by removing the keyboard assembly (and unplug the 'paddleboard' from the main backplane in front of the display and the mains switch connector). Turn the keyboard over and rest it on the keys with the display section hanging over the edge of the bench. The plastic bezel is fragile now, which is why I suggest placing it as described
Now carefully ease off the interconnecting PCB (the one that plugs into both the keyboard and the encoder PCB). Undo the screws and take off the encoder.
Now undo the screws that hold the keyboard to the bezel (some are reached through holes in the keyboard PCB). Put the bezel carefully aside.
Desolder the wires to the power indicator bulb from the keyboard PCB. Undo the screws and take off the front strip and put that aside too. Then take off the power switch (2 screws).
Put the encoder board back on with a couple of screws and plug in the interconnecting PCB.
What I recomend doing then is connecting up the keyboard (just plug the paddleboard into the backplane) and the switch. Turn the switch on (and make sure it can't short to anything -- there is mains on the switch terminals. I put it inside a little plastic bag). Turn the machine on using the switch on the wall socket (if you don't have switched wall sockets, I recomend making up a mains lead with a switch in it).
You can now easily probe connections on the display board.
Posts: 1,788
Threads: 36
Joined: Aug 2007
You can usually find the problem with a multimeter (hopefully one with a diode check function). A transistor looks like two diodes connected together back to back at at the base junction. One way (B-C or B-E) they read around 0.6-1.0V, the other way they appear open. Remember to check from the collector to emitter also for a short. In a display circuit you have lots of identically connected transistors to compare to. But, as you noted, sometimes the external circuitry makes diode checking rather iffy.
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
I use a go/no-go, in-circuit transistor tester when trying to find problems on old calculators. The B&K 510 is small and comes with tiny probes and you can often find the older version on ebay for not that much money (e.g., B&K 510, I bought one for $30). The current version B&K 510A is pretty expensive though.
It works great for in-circuit checks in almost all cases.
Posts: 18
Threads: 2
Joined: Mar 2013
Now I feel a bit embarrassed, as I initiated this thread a few weeks ago but am having difficulties to continue the repair during the Xmas.
A little off-topic question before I proceed with the real hardware: Studying Tony's schematic, I've been forced to admit I don't understand the way HP designed the column drivers. On one hand, they supply all the PNP emitters by a voltage denoted Col+ (some 2.8 V -- four diode junctions). This was unavoidable, as the logic H wouldn't have otherwise been high enough to keep the PNP transistors open. However, once the PNP is switched on, its collector current seems to be as high as (2.8 - 0.7) V / 15 Ohm = 140 mA, which is far more than necessary to drive the NPN transistor (many types can be even damaged by such current). Why? HP could have used a higher resistance than 15 Ohm, having also exploited the fact that no more than one column is ever active at a time. Moreover, it seems to me that then they could have saved 40 pcs of the 22 Ohm resistors. I don't see why did they use them anyway, as now it's some 32 mA flowing thru the one in the driver being actuated. (Noise immunity would certainly be a nonsense here.) I can't explain that -- possibly I'm disregarding something. After all, aren't the Motorola's something more complex than plain transistor arrays?
Posts: 1,162
Threads: 26
Joined: Aug 2005
Something I forgot to ask :
What's the problem with the printer? If it's a 9866A, there's the service manual and schematics over on the Australian site. If it simply needs paper, then a thermal fax roll works fine.
▼
Posts: 18
Threads: 2
Joined: Mar 2013
At the moment, I have no idea. I've been only told that paper didn't advance.
I've also obtained several rolls of original paper -- unfortunately original only in the sense it had been used back then. In fact, it's a sort of substitute, more abrasive, also leaving some remains thermally isolating printer head resistors, thus probably having caused some more frequently used printer columns damage.
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
If it is a 9866, then mechanically it's very simple. There's a big stepper motor on the RHS of the mechanism that directly drives the platten. And a belt drive from that to the paper loading rollers.
Electronically, the motor driver board is the rightmost one in the machine (green ejector handles). It's just the driver transistor circuits, though. It takes logic-level signals from the control logic board (second from the left, red ejector handles). There's a schematic of the former both in the service manual and 'my' schematics, the later is in 'my' diagrams only.
What I would do is remove the covers, and see if anything turns when you press the 'feed' button on the front. If not, then look at the logic level inputs to the driver PCB with a 'scope. If they're not toggling when you press 'feed', you have to troubleshoot the control logic. If they are, then check the motor winding resistance (the edge connector socket on the top of the driver board carries the motor wires), and then the motor driver transistors.
▼
Posts: 18
Threads: 2
Joined: Mar 2013
I forgot to reply: Yes, it's a 9866A.
Quote:
What I would do...
I understand I can do all that with the mains plugged only, without any interconnection with the computer(?).
▼
Posts: 1,162
Threads: 26
Joined: Aug 2005
Yes. The paper feed button will work fine without the printer being connected to the computer.
Posts: 1,477
Threads: 71
Joined: Jan 2005
On the 9810A that I have the printer platen had turned to goo. I built a new one using a lucite rod wrapped with silicon sealer tape. I also had to replace the drive belt with a large o-ring. If these are you're problems let me know and I can describe what I did in more detail.
-Katie
▼
Posts: 1,788
Threads: 36
Joined: Aug 2007
I posted an piece of fixing the platen on a HP9120 printer using 3M Cold Shrink. Works better than perfectly. Same stuff may be applicable to the 9830 printer.
Posts: 1,162
Threads: 26
Joined: Aug 2005
I've got a 9820 coming onto the bench soon (well, as soon as I finish working on the HP9817...)
Anyway, apart from a logic fault somewhere in the processor/memory section (no display) and the obvious card reader roller problem (which I've repaired several times in 9810s, so I know what to do, have the puller tool, etc), the platten in the printer seems to have turned to goo. I was thinking of making a mould and using something like one of the Devcon synthetic rubbers to make a replacement.
▼
Posts: 1,788
Threads: 36
Joined: Aug 2007
If you can find it, I HIGHLY recommend the 3M Cold Shrink. It would be way easier to do than making a mold. DigiKey sells (sold?) the stuff. It is also listed on Ebay quite a bit, but finding that particular size may be a problem.
|