my HP-67 died displays "-0.____0-"

[DaveG]

I'm hoping someone will have seen a similar fault and could suggest where the fault may be.

My HP-67 was powered up today for the first time in about 2 years, I let the battery pack have a good charge before powering on. First sign of trouble was that the prime factors program went off into an infinite loop, so I loaded the diagnostic program (from the standard pack) and ran that. I was seeing intermittent faults, where it would sometimes show 22, 23 or -9.9999 errors. 22 and 23 indicate faulty registers. I have manually stepped through the loaded diagnostic program and confirmed it is the same as in the standard pack manual, so it had loaded from the card ok.

I was repeatedly executing the self test to note the spread of error codes, when the calculator 'crashed' ie the display filled up with a random selection of segments lit and no keys responding.

I powered off/on and now all I get is a display "-0.__________0-" (blank digit = underscore). If I enter '4 ENT 3 +' I see the '4' displayed then the '3' but the result is '6.8203607 4 0-' I see the same answer for '8 ENT 3 *'

I have checked the battery pack and its still got a good charge in it (about 3.9V) and I see the same problems running with the charger plugged in.

Has anyone seen this kind of fault before? I have got a horrible feeling that it won't be a quick fix, but please prove me wrong.

[Tony Duell]

My guess is that it's a chip failure, but it's unwise to jump to conclusions...

I would start by checking the outputs of the internal power converter on the CPU board, both for correct voltage and ripple (See the HP67-notes on this site somewhere for pinouts of the ACT chip, you want to check the Vss and Vgg pins). This doesn't strike me as a PSU problem, but it's easy to check and easy to repair if it is.

Then check the reset capacitor connected to the ACT.

After that, I'd suspect IC problems. ACT, one of the ROM/RAM chips, ROM0 in that order, I think. Replacement ACTs can come from the 67, any topcat, any woodstock apart from the 27, 19C. ROMs have to come from a 67 (a couple of them are also used in the 97 I think).

[DaveG]

Thanks for the useful suggestions. I have checked with a multimeter the voltages and they are +6.14V on pin 1 and -11.34V on pin 2 of the ACT, which I would guess is close enough. I found 182.2KHz on 16 and 17 (clock lines) so they look ok, but there is 0.4V on pin 18 (POR). I will get my scope out tomorrow and take a look round for noise.

Looks like I'll be looking for a donor machine for spares. I took a look on ebay over the weekend and HP67's are going for £80 or more.

Anyone want to sell me a (fairly cheap) HP67 with a good circuit board but with a duff card reader/keyboard/display or broken case??

[David Smith]

Another thing to try is to clean the circuit boards very well with 91%+ isopropyl alcohol. Often times there is (often invisible) solder flux residue that seems to drive old HP calculators batty.

[DaveG]

Here is what I found with a scope on the ACT pins:

1 Vss: 6V with ripple well below 0.1V
2 Vgg: -12V with about +/- 1V ripple. period abt. 10uS
3 F1: to
9 KC5: steady voltages no pulses or ripple
10 ISA: +6V pulses clean
11 Data: rather poor noisy pulses about +0.5 to -0.8V
13 LC2: and
14 LC1: fairly clean oscillation period abt. 1.5uS
15 at 6V no ripple
16 Th2: +7 to -12V clean. period about 5uS
17 Th1: same
18 POR: at 0V no ripple
19 PWO: at 6V no ripple
20 Sync: 0 to 6V pulses clean
21 RCD: +6V with 0V pulses clean period abt. 4.5mS

Unless the ripple on my -12V line is too high, I would say the problem could be the poor pulses on the DATA line.

Should these be 0 to 6V pulses similar to to the ISA line?

I don't have a schematic, is there anything on the DATA line apart from the logic chips? (eg resistors or caps that I can replace)

If one of the chips is shorting the DATA line, I could desolder them one at a time to find the problem, but is it really worth it until I have got hold of some spares?

[Tony Duell]

That DATA line does sound wrong, I am pretty sure you should see decent pulses there all the time the machine is running.

The ACT bus lines are basically as follows. There are 2 clocks, Phi_1 and Phi_2, and sync which synchronises all chips so that they start the (56 bit) cycle at the same time. Note that sync is not output in _every_ bus cycle, so don't worry if it sometimes seems to be missing.

ISA is the instruction and address line to the ROM. The ACT outputs an address, then reads back an instruction, bit serially, of course. Data is the data line to/from RAMs, etc. Normally it's an output from the ACT, during read cycles it becomes an input and the RAM drives the line. The RAM (or other chip) decodes the instruction sent from the ROM over the ISA line to determine when it should do this, and indeed when it should write data from the DATA line to a RAM location.

AFAIK there are no passive components connected to the DATA line. I can check the schematics later if need be (as ever these are on the HPCC schematics CD-ROM...)

Can you detect a measurable short from the DATA line to ground? If so, you could try desoldering the 8 pin ROM/RAM chips to find out which one is shorted.

Do I take it from the fact that you gave the price of a replacement 67 in pounds that you're in the UK? If so, any way you could get to an HPCC meeting in London? I could take a quick look at the machine for you there.

[DaveG]

Hi,
Yes, I should be able to get to HPCC in London, as I live in Luton. I don't know anything about the HPCC, I'm not a member. I guess that would be a good place to look for spares once I know better which part I need.

I measured DATA to GND and its not short, shows more than 20K across those pins. I'll see about pulling the 8pin chips. When I saw the DATA line on the scope I thought it looked like a short too. I have seen that before, last time was on a commodore PET I fixed last year. But in that I was replacing TTL chips which I could buy easily. I need to get some 8pin low profile sockets to stick in their place so I keep the soldering to an absolute minimum.

[Tony Duell]

To answer the last point first, HPCC meetings are the second saturday of every month (usually!) at Imperial College, London. More details on http://www.hpcc.org/

Your problem might also be the driver circuit for the data line in the ACT. Try connecting a resistor (about 10k?) between the data line and Vss (the +6V line) to see if you can pull it up. If so, it's likely to be the ACT.