67 is flakey

[mbielman]

Hello!

I pulled out my 67's last week and one of them didn't work. Not dead, just acting a bit crazy. (Here we go again) I was troubleshooting w/my scope and after about an hour, it was working! After a while, I put it back together and a bit later, same failure again. After working on it yet again, I discovered that if I leave it on for an hour or more, it works. Let it sit (OFF) for a short time and it's bad again. This is very repeatable. Thinking it might be temp related, I heated it up but still bad.

I think I cleaned this one up from a leaking battery but not sure. I had to do that on more than one, if memory serves. Someone mentioned that if I had to do that, there might be battery junk under the ICs. What do you think about that? (I have a desoldering station now so I can pull chips if I have to)

Thanks!

[mbielman]

Update: I pulled the ACT and the ROMs (not ROM0) and saw no sign of battery leakage. (whew!)
But, decided on a guess to check traces w/ohmeter, set on max R. Found high impedance between traces that go to nothing. huh? Cleaned w/IPA (and dried) and no more high Z found.

An aside: I think I "cleaned" this board w/vinegar years ago. I pulled out my vinegar and checked a cap full...conductive! It might be good for battery mess, but clean w/IPA.

Anyway, thought that might be it, but NO, it was the same issue. (all chips soldered back in)
Replaced the cap (330pf) in the osc LC (a suggestion) but no change at all.

Stumped!

[teenix]

I think if corrosion got under the chips you would have seen it surrounding them as well. (but you never know)

Replace all the tantalum capacitors, that might help. They are getting to an age where some are beginning to fail. Having it turned on for a time, may have rejuvenated them temporarily.

cheers

Tony

[John Garza (3665)]

Yep, Teenix is right.

* Visual inspection with a magnifier will spot corrosion. 5-10X will do.

* Caps die eventually, so replace them all.

* Though rare, chip plastic and ceramic can crack over the ages, leading to a path into the central area with the silicon chip and fine wires, where atmospheric humidity and chip lead corrosion will take it's toll. Or normal aging can occur without cracks. If you have another working machine, try swapping chips to determine if one (like the ACT) is the trouble.

* On the Classics (like the 67), it's pretty rare to get corrosion spreading from the battery compartment to the CPU card, as the battery area is well sealed and electrically connected with lengths of wire. Unlike Woodstocks, where the PC board fingers reach out to touch the battery contacts, inviting corrosion to immediately spread to the traces.

* One of the known failure modes of the ACT chip matches your symptoms - the need to 'warm up' before operating properly. If the ACT is bad, it will have to be replaced either with an ACT from a donor machine or a Panamatik replacement unit. Sometimes a 20K resistor across pins 11 and 12 will help stabilize it. But it's no guarantee if the unit is too far gone.

Hope this helps!

-J

[mbielman]

I'll have to purchase some more caps, especially the tiny 2.2uF ones.

I have a working 67 but I hate to pull chips from that unit in fear that I will have two dead ones!
One thing I noticed is that the flaky one no longer works at all. I left it on overnight and remains
in failing state. So I'm going backwards! Funny thing is that when I first soldered the ROMs back
in, it worked for about 15 seconds and went bad again, and has never recovered.

[John Garza (3665)]

Well... if Teenix had more CPU cards.... (wink wink nudge nudge!)

Seriously, I have one of his CPU cards and they are great. Plug in replacement, no soldering.
But last I heard supply chain issues shut down production.

So given that, looks like you may need a donor machine. The soldering heat may have been enough to push a marginal chip over the edge. Not really a loss as it was already unreliable and headed for it's demise.

But before that it may be worthwhile to recap the card (will eventually need that anyway even if that's not the fault), and try the 20K resistor.

I know it's tough to make that jump to cannibalizing machines. But like lots of collectors here, you will accumulate 'parts units' over time. And you will be less anxious handling your operational machines with parts units on hand. Getting a really banged up machine that looks bad but still works makes it a bit easier.



-J

[mbielman]

Let's look at this from a functional point of view. Maybe someone with knowledge of the ROM code
could help.

Power on and display shows 0. with exponent 00 (as if you pressed SCI 0).
Press a number key (5) and the display shows -5.000000000 00
Press ENTER and it displays 5. with exponent 00

ALL functions appear to work. That is, math functions like +, -, x, SQR etc.
Roll works, STO and RCL work, PI, things like that. Display modes (DSP, FIX, etc) do NOT work. (illegal operations like divide by zero show "Error" as it should.)

Cannot enter anything in PGM mode. Just displays 000

So maybe there's a ROM which handles display functions or does this point to the ACT?

FYI - all power supplies are spot on and no ripple, clocks and other signals look good, and tried a new PWO cap (no change). Card reader "tries" to work but I need to replace the roller (o-rings).

Also, ROM0 and ROM3 are AMI; the others are Mostek. The ACT is also AMI. (if that matters)

[mbielman]

Hi John,

1) I thought those from Teenix were blank boards?
2) I saw 2 67 "parts" units on epay this last week, but when bids reached $200 I said "WHOA!"

Cheers!

[teenix]

(10-11-2023 04:14 PM)mbielman Wrote:  Hi John,

1) I thought those from Teenix were blank boards?
2) I saw 2 67 "parts" units on epay this last week, but when bids reached $200 I said "WHOA!"

Cheers!

The blank boards are copies of the original HP circuit boards and the included gerber files can be sent to a PCB manufacturer to make a replacement.

Th others are fully working replacement CPU boards for 55 65 67 97 and Spice. Hopefully 19C soon.

Parts are available again so I've been able to restock since the pandemic.

The described error could be a faulty RAM register at address 62 ($3E).

Great care should be undertaken when trying to remove the ICs. Apart from unintended static damage, the plastic/ceramic cases may be cracked or brittle with age and can break easily in some cases. This will definitely cause a grimace or two, it did for me on a HP-92 :-(


cheers

Tony

[mbielman]

"The described error could be a faulty RAM register at address 62 ($3E)."

Tony - any idea what IC that might be? (ACT?) I have no clue.

I'm also interested in one of your 67 drop-ins and also perhaps a 25 if you ever make one.

I worked at HP Corvallis and have years of experience removing ICs so no worries (so far!).

thx

Mark

[teenix]

(10-11-2023 08:04 PM)mbielman Wrote:  Tony - any idea what IC that might be? (ACT?) I have no clue.

I'm also interested in one of your 67 drop-ins and also perhaps a 25 if you ever make one.

It will be inside a ROM IC, however the ROM ICs I tried to read from reported fault RAM so I could not determine which address ranges were in them.

I have 67's available, the Woodstock prototype PCB's have been designed and should be here soon, but that project will come after the 19C, unless I can clone myself ;-)

cheers

Tony

[John Garza (3665)]

(10-11-2023 04:08 PM)mbielman Wrote:  Power on and display shows 0. with exponent 00 (as if you pressed SCI 0).
Press a number key (5) and the display shows -5.000000000 00
Press ENTER and it displays 5. with exponent 00

ALL functions appear to work. That is, math functions like +, -, x, SQR etc.
Roll works, STO and RCL work, PI, things like that. Display modes (DSP, FIX, etc) do NOT work. (illegal operations like divide by zero show "Error" as it should.)

This is a very different situation. I was under the impression it did not work (at all).
Thanks for giving a more detailed explanation. It does point to a RAM/ROM issue.

I am wondering if the vinegar residue being conductive caused problems when power was applied.
Shorting out traces to the chips could have damaged them.

Either way, looking at these posts, if Teenix has some CPU cards ready, that would be a cost effective approach. Plus you get some new functionality as well. Lots of space to store all the application pacs internally, and bluetooth connectivity to save/load programs to your PC among others.

-J

[mbielman]

John - I agree. I sent Tony an email requesting one of his boards. Time will tell.

I think I mentioned earlier that even if I find the bad ROM, good luck finding one. So this route gives me a super-duper 67 as well as a stock HP one. (just have to fix the card readers now!)

Cheers!

Mark

[mbielman]

Have I given up? Yes and no. Replacing the logic board with the updated one. But I really want to know which ROM has failed on the bad board. That's on hold.

Cheers!

[teenix]

(10-13-2023 01:49 AM)mbielman Wrote:  Have I given up? Yes and no. Replacing the logic board with the updated one. But I really want to know which ROM has failed on the bad board. That's on hold.

Cheers!

If you have a storage oscilloscope, you can trace the code execution and find out where it fails. It is a bit time consuming, but it works, I have done it plenty of times.

You can use a 67 emulator (like mine) to get a snapshot of the start-up code and then compare it to what you see on the scope.

It would be nice to have the ability to store the scope data to a file and be able to interrogate that file and read the program codes and data line based on the logic flow, a big time saver. I don't think I have it available, hmmmm..., silly mind is ticking over again.

cheers

Tony

[mbielman]

I'm sure you could write code to do that.

So the challenge is: how to record it. If the scope can't do it, build a device that emits audio tones based on the logic level, divided by X so it's slow enough. Record the audio to an 8 track machine. Then you have (up to) eight "traces" as audio that a program could decipher.

Then again...

[mbielman]

Mystery solved! ROM3 (1818-0231) bad. Interestingly, I powered up the unit w/o ROM3 installed and it behaved EXACTLY the same as before. (see previous post about failure mode and observations)

So I still have one "bad" logic board (needs ROM3) and one good one.

But now I can sleep.

Mark

[mbielman]

OH NO!! NOW THEY'RE BOTH BAD!

The one that WAS working (previously bad and swapped in another ROM3) was OK for about an hour. Turned it off and replaced the gummy wheel on the other one. Was preparing to do the card reader on the working one and turned it on... BAD!!! The EXACT SAME THING! How can that be???

Do I have 2 units w/a bad ROM? The SAME ROM? Remember, it's the same exact failure mode when run with ROM3 removed! Both are AMI (other ROMs are Mostek) and date codes are close but not identical.

Back to sleepness nights, more or less. :-(

FYI - Tried the rebuilt cardreader (gummy wheel) and no motor action, displays a "3". They both do this.

[teenix]

(10-14-2023 08:31 PM)mbielman Wrote:  Mystery solved! ROM3 (1818-0231) bad. Interestingly, I powered up the unit w/o ROM3 installed and it behaved EXACTLY the same as before. (see previous post about failure mode and observations)

So I still have one "bad" logic board (needs ROM3) and one good one.

But now I can sleep.

Mark

The basic operation of the 67 doesn't execute code from 1818-023 that I can see, it appears to be generally for entering or running a program.

cheers

Tony

[mbielman]

The unit that WAS working and then failed... I left it ON for a while and it started working again.
A real head scratcher other than to say I suspect these ROMs are a problem.
(and I probably found these in the trash)

And thanks to Tony's o-rings, the card reader works now. ;-)