HP Forums

Full Version: HP97 crashing with trig and logs
You're currently viewing a stripped down version of our content. View the full version with proper formatting.

Just got a HP97 which is in pretty good condition but has one unusual problem - using trigs and logs (and a small number of non-related functions) causes the machine to crash and lock up.

If you have the full trace going on the printer, the key press will be printed but as soon as it does that, the calc hangs.

So far, the functions appear to be all trigs, all logs, pi and label buttons including lbl. I havn't compiled an exhaustive list yet because I don't like switching it off and on repeatidly.

There is a broken key on the left hand side key block which doesn't return (it appears the white disk underneath the key mechanism has broken in half - another topic for how to fix that) but would that cause the calc to crash on the other buttons?

Or, and I'm guessing, could it be a corrupt ROM?

Anyway, has anyone ever come across this problem before? Any ideas please?

Thanks :)

Mark

Perhaps if you have not already done, you could try running the general function test found in the HP 97 Service manual available from the Museum. I have found this test to be pretty clever way of finding malfunctions.

Good idea! I haven't got the museum files yet as I haven't needed them until now. I've got the standard 97 manuals but not the service one. I'll have to order it.

In the meantime so I can do some testing right away, would someone be ultra kind and mind telling me the sequence for a ROM and RAM check please (assuming these can be triggered from the keyboard)? That would be hugely appreciated as it will take a week for the museum DVDs to arrive here in the UK.

Sincere thanks :)

Mark

Hi Mark,

Which key is broken and doesn't return on your 97? Perhaps this key is staying engaged and causing the problem. I'll try to dup the problem on my 97 by holding down your bad key while executing trig functions.

Concerning the self test, if you have (or can aquire) a standard pack, one of those pre-recorded cards has a BIT check routine on it.

Best regards, Hal

Thanks Hal. It is the y^x key but to be honest, I doubt it is the cause because there wouldn't be enough pressure in a dead key to make the contacts. However, if you could try please that would be appreciated.

Meantime, this has been very frustrating. My attempts to fix the keyboard have all failed. There is a little white plastic washer or disc that clips to the key shaft and when pressed down, makes contact for the key. However, the white pastic disc is badly designed - a distinct weakness of the keyboard design. Instead of being a single mould, it is split into two half-discs and stuck together which means that over time and usage, the join gradually opens up or cracks completely (as happened with my one). When that happens, the key top and spring are completely loose and can ping out any time so make sure you don't lose the spring!

The fourth attempt at repairing the broken disc and getting the key shaft fixed is drying at the moment. Hopefully, I can put the machine back together again soon and try out some of the testing ideas people have kindly given me.

Mark

Hi again Mark,

I tried holding down the y^x key, and that seems to lock up the calculator completely, with no other keys working while it's being held down. I would agree with you, that's probably not what the cause is.


Good luck with your efforts...keep us posted

Best regards, Hal

Thanks for trying Hal. There is still a remote possibility there is some link with the broken button and the functions not working. However, I examined the track layout of the keyboard PCB hoping that some of the dead functions might be on the same track as the y^x key but none were. Whether that would have been valid anyway, I just don't know.

I'll probably leave the glue to set overnight to really make sure the fix is good so updates tomorrow I expect...

:)

Mark

Quote:
However, the white pastic disc is badly designed - a distinct weakness of the keyboard design. Instead of being a single mould, it is split into two half-discs and stuck together which means that over time and usage, the join gradually opens up or cracks completely (as happened with my one).

I'm not sure what is going on with your particular machine but FWIW two pieces wasn't what left the factory... both the large and smaller diameter discs are all one piece, attached to the rod that that is part of the keytop. Perhaps the previous owner(s) had their way with it? Overall, the 90 series were some of the best keyboard designs ever done by HP.

Randy - I'll admit this is my first experience of this keyboard type but this is what I've got:

The main key block on the RHS of the calc is fine. Nice chunky connections underneath the keybed. No sign of damage or any weakness.

However, the second key block, the function block on the LHS is a mess. There are 30 keys in this block and 17 of them on my machine are showing signs of the white-disk cracking or splitting. One of them has clearly been repaired before.

The white disk looks like it is a different material than the main keys as well and of course, it is smaller.

Now, I've got no previous basis for judging how bad the decay is but it really doesn't look good here. Maybe a previous owner has been a bit rough and used a pneumatic drill on it?

This 97 is a Singapore, November 1976 model. Could different production lines or time makes a difference?

Anyway, I doubt the keys are going to break again now I've done the repair but that leaves the problem of the rest of the machine!

Mark

OK, finally had time to get on with this and test the keyboard fix results.

And they are .... no change. Damn.

Basically, any function related to logs, trig and pi causes the machine to hang. The rest of it works fine so I've got a nice "HP97-lite".

Not happy of course.

My hardware debugging skills are almost zero so I don't know where to look next.

My theory is that there is a corrupt ROM or maybe RAM. The basis for this being that the code for the dead functions exists in a section of the ROM which is bad or they require a section of RAM that is bad.

Unfortunately, I can't run the diagnostic program because I can't type it in without the calc locking up.

So... somewhat up a creek without a paddle. I can use the calc as an addition machine with stats but I am not happy to leave it at that.

I don't know if anyone has any brainwaves for things I can try both external and internal but I'll be really grateful for any suggestions. I want to get this working and I'm really not prepared to leave it neutered as it is now.

Thank you! :)

Mark

OK, this continues to get bizzare.

I've left the machine on a bench and everytime I walk past, I switch it on and punch a few keys.

On this particular occasion, I switched it on and typed in 45 SIN then after a short pause, it returned 0.71 !! Disbelieving, I tried again and the same result. I typed in 45 TAN and it came back 1. Then I tried 1 LN and it froze and I haven't been able to get the trigs/log back since.

So, the ROM is there and not corrupt. Could still be intermittent memory problem I suppose or maybe there is a bad connection I didn't notice when I examined the insides.

Mark

Mark,

This is somewhat of a good news, because it shows that the chips are still functional -- but not at all times. You may want to look at the small 8-pin DIP chips with a magnifying glass, look for cracked soldered joints or hairline breaks in the PC board traces.
You may also try to melt each of the soldered joints, which would correct any soldering problems.
Another idea: Try running the failing functions while applying pressure on various points of the PC board, close to the 8-pin chips. This will make or break the continuity of traces between the chips.

And please keep us posted!

Quote:
look for cracked soldered joints or hairline breaks in the PC board traces

This is very important if the unit has been repaired by someone other than HP. There was a large scale HP dealer here in the US that did repairs in the mid nineties after HP stopped doing so. The quality of their work was to say the least, horrible. As a result, I've repaired many 97's that had just this problem, where the plate-though connections had been destroyed or significantly damaged in the desoldering/component removal process.

Based on your earlier description of the key discs, I'd say looking over the soldering would be time well spent. You should be able to discern what has been reworked and what is factory original, take note of the reflow up the chip lead sides on the top of the board. Uneven and missing reflow is a sure sign of rework.

Thanks Joel and Randy :) I will give this a go this evening. I do recall when I opened it last time that a few areas looked suspect.

If I see something that looks bad, I'll try to link an image for a second opinion if possible.

Regards,

Mark