HP-80 display problem.

[jlw]

I very recently acquired an HP-80 calculator that works fine except for a display problem/anomaly. It lights segment c (the one that's off only in a 2) for all digits except the space where a dot would not be when indicating low battery with the decimal points.

If I power this externally and lower the voltage to about 3.5V, those extra segments go away and all digits look correct. If I lower it a tiny bit more, the low battery indication comes on. So there's a tiny voltage window for proper operation. Alas, the battery pack is about 3.9V, well outside this tiny window. The AC adapter is even higher.

If I install the HP-80 logic board in an HP-45, it works and displays just fine. I have three HP-45 keyboard/displays, in various stages of fixing their own display problems (single segments wrong), but one is fully restored. For that matter, an HP-45 logic board on the HP-80 display board exhibits the same problem, starting at 4.2V.

I fetched and perused the hand-drawn schematics for both machines (from http://www.hpcc.org/cdroms/schematics5.0/index.html) looking for differences. One thing I noted was that the value of the resistor on the Rset pin (10) of the anode driver chip U1 was 220k on the 80 and 330k on the 45. (It's 300k on the 35, 220k on the 70 and 65, 120k on the 55. The 67 is a different beast).

Per these schematics, part numbers for the keyboard/display board are 00035-60004 for the 70, 80, and the 45, 00035-80004 for the 35. All of mine are 60004.

Examining the boards, two of them have 220k resistors installed, one has a 330k, and the 80 has a 390k. It looks like it had been replaced once. I found a 220k (measured 216k) to replace it (Wow, those resistors are tiny! Not quite SMD, but getting there.). The 390k that came out measured 520k or so. However, this change did not affect the display anomaly.

The anomaly is observable on a scope at the c pin (13) of the anode driver as a whole bunch of extra pulses. They go away when the voltage is reduced.

I was hoping that passive components could fix this problem. I don't particularly want to replace the anode driver (I'm down to one uncommitted good one, and they're not much fun to remove, and this might not be the problem either.). There aren't very many components on the display board beyond these. I also don't want to cobble up a 3.5V regulator, or a voltage divider, either.

Suggestions (besides give up, or only use it on my lab supply)?

[teenix]

The resistor is to set an internal reference for the battery low indication (all decimal points being lit) I don't know why the resistor is different for the Classic models as the anode driver chip has the same part number. Maybe different battery loads ??

The anode driver is wired so that the decimal point for all digits should light in a low battery condition. If it is not lit on the actual decimal point digit under low battery, then that is strange, even stranger if the c segment does not light on that digit.

In normal operation, is the C segment lit regardless of what is on the display and does it also light where the decimal point is?

The fault is not on the CPU board, so check pin 1 of the anode driver and see if it is Logic 1 all the time under normal operating conditions. If it is this will be decoded inside the anode driver to display segment C all the time.

It should stay low all the time if only 2's are on the display.

If you can clearly see that this pin is changing when changing the display digits, then the fault could either be a faulty display - maybe an internal short, a short on the display board pcb tracks, (corrosion etc) or the anode driver has failed and is not decoding the incoming data properly.

Under certain conditions, an anode driver output can be destroyed by inductive voltages created by the display coils. It could be internally shorted for the C segment.

More display info can be found in my Classic Notes PDF file if you are interested.

cheers

Tony

[jlw]

Perhaps I should put the original resistor back.
Under low battery, all decimal points light, except the one following the decimal point.
Under high battery, the c segment lights in all positions except the decimal point.
Pin 1 often has a 3kHz pulse train on it, corresponding to which digits have that segment lit.
With only 2s and a decimal point in the display, the pin 1 trace is flat, but it's not ground. It rises as the battery voltage does, and probably crosses the logic threshold and lights the c segments. (45 probably doesn't do this as much, if at all) Missing or faulty pull-down resistor? (cure-by-symptom) I don't know what logic family this is. 45 or 80 logic board on another working (for our purposes) display board, pin 1 stays at ground as the power supply is varied about a whole volt.
It seems unlikely to be a faulty display as everything works fine (=larger top end before the Cs appear) with the HP-45 logic board installed. Pin 1 still floats, but not as much.

[teenix]

Those connections are direct chip to chip so won't be an external issue unless there is corrosion or a bad connection at the pin header between the boards.

As the CPU board works in another calculator, and assuming there are no board faults, it would seem the anode driver has gone faulty. Pin #1 on the anode driver has a direct internal connection to an input on the NAND gate that drives segment C, maybe this is the culprit.

Also, the anode driver power is sourced directly from the battery supply. If the calculator has had a higher than normal voltage applied from different batteries, an external supply, a spike, or accidental reverse voltage then it might have partially failed. If it has a fault, it might feel warm to the tough after being turned on for awhile.

I believe the CPU circuitry is PMOS, not sure about the anode driver, could be the same or TTL maybe.

cheers

Tony

[jlw]

A pull-down resistor on pin 1 seems to solve (or at least mitigate) the problem. Now all I have to do is route it so that the calculator can still be put back together. The fit is a little tight.

The negative battery terminal has finally broken. It was severely corroded and hanging by a thread. I don't quite feel like stealing one from the 2nd 45 I'm fixing, so a kludge may ensue. Until then, AC adapter ho!

FWIW, the AC adapter with this thing was putting out half-wave rectified DC for the calculator not to run on. Opened it up and replaced the largest capacitor with a much newer one from my stock. Works fine after that.

Following my hunches seems to be working these days. Must be time to buy a lottery ticket, except that luck is conserved.

Reassembled calculator. Keys need cleaning; looks dirty next to the 45.