HP27 Restoration (Lots of Corrosion)

[calc-calcs]

Just got this HP27 Woodstock off ebay. From the pictures on the listing there was obvious corrosion present on the battery contacts which would probably (almost certainly) mean there would also be internal corrosion. Typical calculator with its battery pack left installed, give or take 30-40 years ago until someone found it and put it on ebay. Oh and on top of it all the original charger was also (unfortunately!) in the drawer, so the seller must have decided why not just try to make it work... and probably roasted one or more ICs. It was listed as having an intermittent display. Hmmm, were did I hear that before...

After having restored two HP25 (one 25 needed a Panamatik ACT) and three HP21 calculators, last two mostly purchased for parts but ended up getting them all to work, and having dealt several times with moderate internal corrosion I felt confident (or hoped) I could also handle this one without too much work. And maybe, just maybe the ACT might still be alive. Otherwise there is always Panamatik to the rescue.

The calculator really looks quite nice on the outside, no scratches or blemishes, label is fine, and all keys still have a good feel. But after opening it up I realized this one is bad inside. Really BAD. It will need a lot of cleaning, and might even need to rebuild some traces as a few have really been eaten away by the blue stuff. At least the corrosion seems to be only very heavy in the ACT and ROM chips area. There is also slight corrosion around the power supply components, and around the charge limiting resistor on the opposite side. At least the display and driver chips part of the board seems to be clean. Fortunately the keypad traces are mostly (golden) clean, and only a few will need attention, hope that the contacts of the switches are ok. If not there is always a drop of Deoxit that might help.

Oh well, lots of work to do. First will be removing both the ACT and RAM chip (or is it a ROM chip?) from the board. Experience has taught me that corrosion likes to hide under those chips as well. Going to clean around the pins before attempting to desolder them.

Will be updating this thread as I move forward. Any comments or suggestions welcomed.

[calc-calcs]

Two more pictures of the corrosion at the ACT and ROM (or RAM) chips.

[calc-calcs]

OK both chips are out, and as anticipated, there was indeed LOTS of corrosion on the traces bellow them. Much more than was initially visible. I wonder what effect this corrosion had on the chips when the seller tried to test the calculator with the charger. Still hoping for a miracle and the ACT might still be good (saved by corrosion?).

And now the clean-up process will be next. After that will have to verify continuity of ALL traces in that area, and also the vias on all the IC pins and contact fingers. This will easily be so far my most challenging calculator restoration for sure. But given the model at hand, well worth the effort.

To be able to properly heat up the solder joints, washed down the IC pins in the the affected area before removing the chips. First used some alcohol and Qtips, then hand soap and a soft toothbrush and gently brushed some of the stuff away and let the board dry for an hour. Some of the harder spots just used a Xacto knife to very carefully flick/scrape them off and then another wash with soap. I tried to be careful not to touch the top of the ICs, unfortunately it seems that during the process a bit of the labels came off just by getting them wet. But at least I have them documented in previous pictures, not that the labels where completely intact to begin with. Very fragile ink after all these years, or perhaps the weakening was also related to exposure to the corrosion chemicals.

I'm thinking to remove as well the two transistors of the power supply and the yellow and the small black tantalum cap with the green beads to be able to clean under both of them. Might even need to remove most components of the power supply as well to properly reach all the traces and remove all the CuSO4. At least its way less corrosion in those areas, mostly just some spots here and there, but not all of it is directly accessible.

Given the state of the traces at the ACT/ROM, does anybody have the schematic of the HP27, or perhaps the PCB layout? Probably a long shot, but that would be a big help. I assume the PCB layout might be similar to the HP25? or at the very least it might be a good general guide. As said before, I anticipate quite a few traces will be gone or close to that and will need to be repaired after I'm though cleaning off all the blue stuff, which really takes a bit of watchmaker precision and a good magnifier to get to the very last bits as they can really be almost microscopic. No point leaving behind any of it.

Going to get some rest and continue with this tomorrow.

[calc-calcs]

Could not help it to carry on and made some cleaning progress. Some of the traces are really fragile.

Feeling a bit tired and have a busy work day tomorrow, so will stop for now.
More updates to come soon.

[Jonas Sandstedt]

Hello,

Good work, it seems like a real challenge and it is interesting to follow.

Perhaps you already know that, but you can find schematics for HP-27 at
http://www.teenix.org
and a service manual for HP-29c at
https://literature.hpcalc.org/

It might help.

best regards
Jonas

[calc-calcs]

(04-06-2022 08:41 AM)Jonas Sandstedt Wrote:  Hello,
Good work, it seems like a real challenge and it is interesting to follow.
Perhaps you already know that, but you can find schematics for HP-27 at
http://www.teenix.org
and a service manual for HP-29c at
https://literature.hpcalc.org/

It might help.

best regards
Jonas

Hello Jonas,

Thank you for the information. However I looked at the teenix.com website that you indicated for the HP-27 schematic, but the only link there related to this model seems to be for an emulator. Not sure if I missed something? If so could you possibly provide a direct link to the download?

I already had the HP-29C service manual for some time, and looked at it yesterday to compare, but the board layout in the ACT area is quite different. And at least in my copy of this manual there is not even a PCB trace layout, only parts layout on a blank board, and the schematic, which may be of some help.

Also compared the traces under the ACT with pictures of one of my HP-25 back from when I was working on them and had removed the ACT and RAM chips, and traces seem also mostly different. Seems I am going to have to create my own PCB trace layout, at least I can still see it quite clearly when shining a light on the opposite side of the board. At least the solder side traces look quite healthy and clean. Some holes of the chip pins appear to be clogged in the pictures bellow, but its just that not enough light was shining into them. At this point all thru-holes of both chips are reasonably clean. Will go over them again before installing the low profile socket strips.

Thanks again.

[teenix]

(04-06-2022 11:44 AM)calc-calcs Wrote:  Thank you for the information. However I looked at the teenix.com website that you indicated for the HP-27 schematic, but the only link there related to this model seems to be for an emulator. Not sure if I missed something? If so could you possibly provide a direct link to the download?

I already had the HP-29C service manual for some time, and looked at it yesterday to compare, but the board layout in the ACT area is quite different. And at least in my copy of this manual there is not even a PCB trace layout, only parts layout on a blank board, and the schematic, which may be of some help.

Also compared the traces under the ACT with pictures of one of my HP-25 back from when I was working on them and had removed the ACT and RAM chips, and traces seem also mostly different. Seems I am going to have to create my own PCB trace layout, at least I can still see it quite clearly when shining a light on the opposite side of the board. At least the solder side traces look quite healthy and clean. Some holes of the chip pins appear to be clogged in the pictures bellow, but its just that not enough light was shining into them. At this point all thru-holes of both chips are clean.

Thanks again.

There is a bit of difference in the 29C and 27, mainly ROM/RAM chip count. The 27 has less and the power on reset setup is a bit different. The RAM chips have continuous memory support in the 29C also. The ACT is mostly wired the same.

Tony Duell's circuit for the 27 is here:
http://www.hpcc.org/cdroms/schematics5.0...k/hp27.pdf

Teenix schematics
http://www.teenix.org/classSchm.zip

If you end up stripping the circuit board down to clean it, and if you have time, perhaps you could post or send me some pictures of the bare board from both sides with a light source behind as in you other pics. I might be able to re-engineer the board design and add it to the ones on teenix.org

cheers

Tony

[Jonas Sandstedt]

Hello,

What a pity that the HP-29c service manual could not offer better guidance for the HP-27,

The schematics provided at teenix.org are almost at the bottom:

"Reprinted schematics for Classics, Woodstock, HP-10, and HP-67. Zipped PDF files."

Link:

http://www.teenix.org/classSchm.zip

Hope these help better, let me know if you do not find them.

Best regards
Jonas

[calc-calcs]

Got both files, thank you everyone.

[GreyUser]

Also noteworthy is that the 27 was built exclusively with one-of-a-kind HP nmos process technology, it has no interchangeable ic’s with any other hp model. Note the very non-standard (for hp calcs) voltages shown on Tony’s schematic.

White vinegar in an ultrasonic cleaner is your friend for removing that level of corrosion. Yes, some of those traces will be open afterwards but that will happen regardless of the method used.

The traces can always be repaired. I’d be more concerned with the level of corrosion that has found its way into the keyboard flex strips. That’s the real challenge.

[Duane Hess]

Hi:

Interesting. Certainly a lot of tracings look like they may be incomplete. Should the chips be OK and you can re-lay/repair the tracings I wanted to point out one specific trace. As if it is incomplete, an otherwise functional unit behaves in a manner similar to the "fried ACT".

This might not apply to you, but might help others:

I make comments due to my own experience and not because I even begin to be knowledgeable in the matter. After obtaining a super cheap, claimed non-functional 27, I checked the insides. Obviously nearly every component was unsoldered and resoldered. Evidently the previous owner was trying to diagnose the unit.

The behavior of the unit was as follows:
- when powered on, the display was ALL zeros, but shimmied, VERY rapidly
- closer examination showed the zeros blinked numerous times, then changed
- the change was, steadily spaced, all zeros counted then up/down (forget) 0 thru 9
== steadily paced means after 'x' flickers of all zeros, the display blinked once all 1s
== then 'x' blinks all zeros, and one flash of all 2's; 'x' all zeros & one flash all 3's
(0, 1, 2, 3...etc. or 0, 9, 8...etc. recall it counting up)
- it was really hard to see, as it was so fast; needed to get somewhere dark

The previous owner was unsuccessful. By accident I got it to work 100%. Which involved shunting one defective trace. As so much was resoldered, performed many continuity checks. Had screwy resistance values on one trace. Found the problem when a section of the trace simply fell off.

Your 4/6 12:57 post with the before and after pictures shows the trace I dealt with. As does your 4/6 06:44 back-lit pictures.

Looking at the 'after' picture from 12:57:
you see the PCB cut-out for the battery pack. Look immediately left of the cut-out to the 1st trace you see. This trace goes towards the top of the calc. It also goes towards the bottom of the calc. Connecting to the 3rd hole from the left for the top chip (16 pin). Amungst other connections, following the tracings on both sides of the PCB, it connects to a hole on the bottom edge of the PCB. The 1st hole immediately left of the sticking-up connector pins at the bottom of the PCB. The perspective is as you look at the 'after' picture as shown.

The trace goes upwards to a chip under the display panel. Examining the trace, as a whole, shows directly connecting a wire from the top-edge chip pin directly to the bottom edge PCB hole you do not by-pass any functionality of the tracing. But, instead, complete the tracing.

That is how I tested it. Which was a bit tedious, as for testing I didn't want to solder anything. Tightly wrapped a wire around the pin of the top chip and inserted the other end into the bottom hole. Placing the components in the case knocked the wire loose, so held it all in my hand. Attached the keyboard, held the battery in place and after some dexterity was able to hold it all steady. The unit worked flawlessly.

Looks like you have your work cut out for you. May have no pertinance to you. But after you're done, if you have the display behavior, you might want to look towards that trace. The 'after' picture looks like part of the tracing is missing just above the yellow blob (capacitor?).

Overall, that methodology is how I managed to fix the problem. Don't recall every detail, but assuming I examined the pictures correctly, it all should be correct. If not, you'll figure it out. As the 1st trace to the left of the battery cut-out on the surface of the PCB facing the camera of the 'after' picture is definately the tracing I'm referring to. And I remember connection a pin from a top-edge chip to some hole on the bottom of the PCB.

Whatever its worth.

[calc-calcs]

Some progress made. This evening I set out to finish cleaning the board and test the power supply by itself. That shouldn't be too hard.

So pulled out both transistors of the power supply and pulled one leg out of the first string of adjacent components, the zener, the 680R resistor and 3x 1N914 to properly clean under them. Also the yellow 47uF cap and the small 2.2uF with the green beads on its legs, both located to the left of the ACT.

Tested those components and after cleaning as best I could resoldered them. Without the ACT and ROM chip, or display or keyboard attached powered it up and power supply seems to be working fine. These are the voltages (Spec value from schematic) and actual measured values:

Buss Spec / Measured
-------------------------------
Vcc 2.5 / 2.53
Vdd 6.4 / 6.44
Vgg 11.8 / 12.65
-V -2.8 / -3.06
Vdisp 4.9 / 4.91

Power consumption was about 40mA.

Then of course wanted more. I also did a bit of reality check and decided I am not going to go overboard with the cleaning, at least not yet, as it will be way too much work without even knowing if there is even any chance for the calculator to work. So just gave the traces a couple of additional rubdowns with a Q-tip and alcohol until there was no longer any blue residue left on the Q-tip, and also removed any obvious visible residues.

Then made a hand drawing of all the component side traces and started checking them for continuity and marking them down according to the test results. After that repaired all the broken traces with #30 Kynar wire which I had ordered a roll some weeks ago "just in case" so it came in very handy. The trace repairs also included one trace that takes ground up on the right side of the board (along the charge resistor and diode) and reaches all the way into the display drivers. Also had to repair two other traces that go up on the left side of the board and snake through the power supply and also reach all the way up into the display driver chips. These last ones where ISA and "01". Once this was done checked continuity of the repaired traces and all looked good.

Next installed low profile socket strips for both the ACT and the RAM/ROM chip. All good so far, but first the chips needed a bit of further cleaning off flux on the underside and straitening of the pins so that took another while. The ROM chip still even had a bit of crusty blue stuff on the inside back side of one pin. So in they went into their sockets. Then attached the display and keyboard. Time to power this up up and see what happens!

When I first turned it on nothing was shown on the display, current was about 45mA. But after a few seconds suddenly a combination of sometimes the right most digit flashing, sometimes the left side one. Sometimes all zeros would show. All this without touching anything yet. Then I tried the keyboard but no response to it. The flashing digits continued. Hmm looks like Panamatik is going to be selling me another ACT.

At this point decided to pull out my test HP21 and put the HP27 keyboard on it to make sure it was even working. The keyboard turned out to be just fine, except for the first 3 right side keys from the top down which did not work ([g], y^x, and CLX). All other keys worked just fine, no bounce or any other problems.

Next I put the HP27 keyboard back on its own board, and got a few flashes as before, but after about 10 seconds of leaving it alone, lo and behold suddenly "0.00" came up on the display!! And all keys started working as well, that is except again for the same 3 that did not work before while on the HP21 board. I looked at the schematic and actually those 3 keys are exclusive to one row, so since the keyboard had the exact same problem on both calculator boards it has to be a problem in the keyboard so will need to check those mating connectors as they had some blue stuff inside in a few of the gold plated holes although I managed to clean them ok or so I thought. Or perhaps is just the keys that need a bit of Deoxit, but it would be almost too great a coincidence that only these 3 keys of a specific row would have contact problems. Could be, but I think its rather related to one of the keyboard push in mating jacks. Will look into that tomorrow.

There must also be still something loose somewhere, I can sometimes make the display go haywire by moving or pushing the main board around a bit. Sometimes the display will momentarily go blank and a second later will at times come back on as "0.000" as if I had done a [f] FIX 3, and everything still worked. Another test I remembered doing when working on the HP25 is to hit the E+ key and display should advance by one each time. If not the RAM chip might be bad. Unless there is something else going on, this unfortunately also seems to be the case here. Not sure yet, will definitively have take a closer look at this.

But in any event I think I made a bit of progress today. OK its time to hit the sack. And of course added some pictures. Posts are not as much fun without them. But since there is a 5 picture limit per post, might upload a few more tomorrow on a follow up. BTW let me know if you want to see anything specific up close, etc.

Good night, or good morning....

[teenix]

There could be corrosion inside the holes that mate to the interconnection pins.

The pins themselves may not have much tension in the holes and are not connecting properly.

These pins are brittle and break easily if stressed.

cheers

Tony

[Maximilian Hohmann]

Hello!

(04-07-2022 09:24 AM)teenix Wrote:  There could be corrosion inside the holes that mate to the interconnection pins.

These pins are brittle and break easily if stressed.

Yes, I've seen that too (both: the corrosion inside the holes and broken pins). Regarding the green and blue corrosion, outside and inside the holes, I came to the conclusion that the best way to cure it is bathing the PCB in vinegar. The same type that I put on my salad, undiluted. No need to desolder anything, unless broken PCB traces are suspected under the integrated circuits, because the components are not harmed by this rough treatment. Rinse with plenty of water and dry it in a warm place before applying power again.

Regards
Max

[calc-calcs]

The keypad problem of the 3 non functional keys is solved. There was a very small but obvious gap in one of the traces that I had overlooked where corrosion did its thing. Fixing this gap solved the problem.

However still having lots of display flashing issues, pushing or moving the top area of the board where the display and drivers are located seems quite sensitive to trigger these issues. Will have to check around there to see if there is a bad contact somewhere in the mating pins, or even an overlooked faulty trace. I am concerned that sometimes when only one or a few separate digits flash they seem overly bright. Hope this will not burn out segments or digits so just in case I am going to swap the display module with the one in my test HP21 that is already missing about half the A segment of the rightmost digit to lessen stress on the HP27 one which seems to be fully operational.

Also noticed that aside from the storage registers always returning zero (saving anything to any location seems to return Ok), and the E+ count also staying on zero after each key push. Now also noticed that doing a 10 [1/X] operation always returns "Error" instead of the expected 0.10 result. I suspect that RAM memory might be toast, or somehow the chip is not being addressed properly. Perhaps I missed some broken around the connections to U3/U4? One of my HP25 had almost identical problems due to an intermittently working RAM chip. That one ended up with the Panamatik ACT just to solve that problem as it has its own internal RAM and ROM.

[teenix]

(04-07-2022 03:30 PM)calc-calcs Wrote:  However still having lots of display flashing issues, pushing or moving the top area of the board where the display and drivers are located seems quite sensitive to trigger these issues. Will have to check around there to see if there is a bad contact somewhere in the mating pins, or even an overlooked faulty trace. I am concerned that sometimes when only one or a few separate digits flash they seem overly bright. Hope this will not burn out segments or digits so just in case I am going to swap the display module with the one in my test HP21 that is already missing about half the A segment of the rightmost digit to lessen stress on the HP27 one which seems to be fully operational.

It maybe the interconnect pins that supply power to the CPU board from the keyboard via the power switch. Maybe corroded pins, PCB mating holes, and or poor tensile contact from the pin.

This rapid power on/off cycling may be upsetting the ACT and/or ROM0 chip (5061-0430) causing them to stop working. This can be damaging to the LED segments, and maybe the drivers, as they will be getting continuous high currents flowing through them which is the brightness that you see. This high current normally flows but only for around 300uS through each segment during the display refresh process.

If you have a soldering iron, you could try re-soldering the IC connections. They can be corroded also, resulting in dry joints and poor connection. You have to be careful as the new solder might not take and you will be tempted to keep applying the iron causing overheat damage.

It is better to unsolder them and then clean things up and re-solder.

As you mention the RAM chip seems to be inoperative. (5061-0459) If so, then only a replacement will fix things.

These chips are sensitive to static electricity damage so proper handling of them and the circuit board should be observed.

cheers

Tony

[calc-calcs]

Checked continuity between all inter-chip connection and only one was partially failing, all other where solid OK. All power, data, and ground connections where checked between all ICs.

So found another somewhat intermittent trace that carries the Ø1 clock signal from U3 ACT pin 17 to U2 ROM chip pin 15. The connection would measure zero ohms most of the time, but when moving the board a bit it went anywhere between 135 and 350 ohms. The jumper pictured bellow took care of that.

Next I pulled both chips out of their sockets, cleaned all pins again, and reinserted them. Just in case. Also resoldering both the ACT clock inductor and capacitor which appeared to help a bit as well.

Calculator now seems a bit more stable and almost half the time powers up to 0.00, but the other half it still misbehaves, flashes digits, or just a blank display, all zeros, etc.

Also as mentioned before, when the calculator comes up normally to 0.00, all the basic algebraic functions work, but some higher functions don't seem to work, neither the storage registers or financial functions which use storage registers. The non working functions are: 10[1/X] returns "Error" instead of showing expected 0.10 result. PHI key returns zero. The Trig SIN, COS and TAN also always return zero, and the inverse ARC functions blank the display. I did not test the statistical functions yet, but assume they will not work.

The working functions: LN/Log, x^2, 10^x, Y^x, SQRT, %, H.MS/H, X!, E^x, and of course + - x / are fine.

Also cleaned again the receptacles and the keyboard connection pins, and found that a few keys needed the Deoxit treatment. Among them where the 1, 2 and 6. The 2 and 6 started working well immediately after a drop of Deoxit was put into the back opening of the key, but the 1 key stopped working altogether. Had to snake a bit of braided wire inside the switch and a bit more Deoxit and now it is finally starting to work again but still not entirely reliable so will need a bit more work.

I tested all DC supply busses and found quite a bit of ripple, but I assume this would be normal figures for this simple switcher with such small filter capacitors. I was going to set the scope for 20MHz BW limit (100MHz scope) as a lot of the ripple seems to be high frequency content, and re-read the ripple figures but forgot. Will redo it if these ripple readings appear way too high to anyone that might know the usual expected levels.

Rail Volt Ripple P-P
-----------------------------
Vdd 6.45 290mV
Vgg 12.66 205mV
V- -3,72 200mV
Vdp 4.76 185mV

Even the Vcc (2.5V battery rail) had about 440mV of ripple on it.

I am going to hook up my scope to all the data lines and take a picture and post it next.

[calc-calcs]

Here are some scope screen shots of the data lines: clocks Ø1 and Ø2, ISA and DATA. Also the signal at the timing inductor of the ACT, frequency seems to be 707KHz, which kind of rings a bell from when I was working on the HP25, so it should be correct. ISA buss is basically at the same frequency of the LC oscillator, only higher in amplitude, but exactly the same waveform shape.

I did verify that all data was present on all IC pins where it should be, and previously also tested continuity between all those points.

As can be seen there is a 1.4uS shift between the Ø1 and Ø2 which I assume to be on spec.

I did not see much if any activity on the SYNC or RCD lines. Should there be continuous activity on these lines?

Anything else I should be looking at?

[Ren]

{Viewing photos of corrosion on HP-27]

Please! Don't show such photos just before bedtime!
I hope I don't have nightmares!

B^)

[calc-calcs]

You knew it and came looking for corrosion, I put a warning in the subject!

You're welcomed