HP27 Restoration (Lots of Corrosion)

[teenix]

It might be worth replacing the tantalum capacitors in the power supply. Also, there is a 2u2 power up reset capacitor hanging off pin 18 of the ACT chip. If this is faulty, it may not reset properly.

cheers

Tony

[calc-calcs]

(04-08-2022 10:38 PM)teenix Wrote:  It might be worth replacing the tantalum capacitors in the power supply. Also, there is a 2u2 power up reset capacitor hanging off pin 18 of the ACT chip. If this is faulty, it may not reset properly.

cheers

Tony

I compared the ripple readings of the power supply rails to the ones in the HP21 and they seem very similar, so I would expect that what I saw on the HP27 power rails is normal and the caps should be fine. Besides Tantalum being so stable rarely fails due to ESR or degradation, they usually just like to pop or burn up if voltage stressed or with reverse polarity.

But since the last repair session the HP27 seems to have changed its mind and now upon power on only rarely wants to show any activity on the display at all, perhaps just one digit to the right now and then, but mostly just a blank display, I decided to follow your suggestion about taking a closer look at that reset cap.

The cap reads 2.19uF and 1.7 ohms ESR. I compared that to a new 2.2/16V Tant and the readings where almost identical, 2.12uF and 3.1 ohm ESR, so actually a bit worse than the original. Even so just for the heck of it substituted the old cap with the new one but no change in the calculator, other than I managed to break one of the very fine legs of the original cap but managed to reattach it, green beads and all, so the little guy is back in place.

That not withstanding, the voltage on that 2.2uF reset cap (and pin 18 of U3) is strange for a reset circuit. On power up and with the DMM connected it starts at zero and the voltage very slowly then starts to raise until it reaches about 0.6 volts at which point it only seems to keep climbing very little, like the DMM is actually placing a load (10M) on the charge current. All in all it takes the cap around 45-50 seconds to get to to around 0.6V. If I then disconnect the DMM and wait a couple of seconds and reconnect it the voltage will have risen to around 0.7 - 0.8V and after a couple more measurements will then appear to settle at about 1.2-1.3V. Looks like the cap is very slowly being charged by a minuscule current. I even checked at the actual pin of the IC in case there might be a bad contact with at socket on pin 18, but the voltage was exactly the same ruling that out.

Can someone please confirm the expected behavior of the Reset pin 18 of the HP27 ACT?

Also given the change in behavior of the calculator from one session to the next checked again the system waveforms, but they are still present and look ok, except for the SYNC line (ACT pin 20) which looks odd, but since I had never tested it before no previous reference so not sure if that might be related to the change in behavior. Also checked for SYNC at the other expected nodes on U4 RAM/ROM pin 13, and U2 Anode Driver pin 18 and get the same waveform, so no open traces here.

Will post some updated scope shots including the SYNC line on the next post.

[calc-calcs]

Here are some fresh waveform shots, including the very crucial SYNC line which I had not checked before, but seems very odd. Notice the difference in level of that line as well (had to change that channel from 5V/div to 1V/div to see more detail), aside from the spiky shape.

Can anyone please confirm what is to be expected at the SYNC line of the HP27 ACT pin 20? Or if what is shown on the screenshots bellow is correct for that line?
Since the HP21 does not use a SYNC line could not make a direct comparison.

I also seem to have mixed up the ISA line in my previous scope shots, in these ones is the correct waveform which seems to make more sense.

The last shot also includes the DATA line, as seen while the display was blank. Tried to cycle power to get a reading with something on the display, but gave up after numerous tries ending with same blank screen.

[teenix]

Attached is scope shots of a HP-25

Reset - 50mS/div, a bit noisy, but my setup probably not ideal.
Sync - 500us/div
ISA - 50uS/div

The sync is as is because of the code it is running during the key wait loop. Sync stays LO after an IF instruction in the code. This tells the ACT that the code value itself is a ROM address.

displ5: 0 -> s3 <<< sync HI
if s3 = 0 <<< sync HI
then go to dsst4 <<< sync LO, goto ignored here as S3 = 1
if s15 = 0 <<< sync HI
then go to displ5 <<< sync LO, got executed as S15 = 0 (no keypress)

The ISA bus is showing the code executing for the above.


For the HP-27 it will be executing...

if s15 = 0 <<< sync = HI
then go to $08A <<< sync = LO, got executed as S15 = 0 (no keypress)

so the Sync line will be alternating HI and LO for each instruction.

cheers

Tony

[calc-calcs]

Well my ACT seems to have died, it will no longer boot as it was doing quite often before, and according to Bernhard this is a typical scenario when damaged by overvoltage it can work for a while and then just quit. He also explained that the HP27 uses RAM for the operations that are not working in my machine (Trig, Financial, 1/x, etc) so the RAM memory inside the RAM/ROM chip is definitively bad and of course it is an unobtainable part. I already had a suspicion of the RAM problem due to the storage registers and the E+ function not working and always returning zero.

Something similar happen to me with the RAM chip of the first HP25 that I restored, it had some corrosion on the bottom of the chips which I cleaned, but it seems it can leach into the chips (probably trough the pins) and cause internal damage. In that machine RAM was initially working but after a few days sometimes would work sometimes it would not, and then finally died. Of course never was able to get a replacement RAM chip either, so that HP25 ended up with the Panamatik ACT.

Unfortunately I think this is the end of the line for the HP27 attempt to make it work with its original hardware, so I placed an order for another ACT that will be specifically configured to interface with the HP27 motherboard, and that should take care of all the remaining issues. If for any reason that ends up not working due to problems with the motherboard, I still would have the option to reconfigure the ACT to work on one of my HP21 motherboards, and use it to make the HP27 operational. Bernhard explained that I would only need to add pin 12 to the ACT, and add a wire connection and of course reload the proper firmware. It would be a bit of a cheating restoration, but I do not know of any other alternative if the original motherboard keeps giving me grief.

Thank you everyone that made suggestions and provided information. Stay tuned and I will be updating this thread once the newly ordered ACT arrives in 10-15 days.

[calc-calcs]

Thank you for taking the time to gather this information Tony. Unfortunately I had to decide to stop trying the restoration with the original ACT and the RAM problems so I ordered a Panamatik ACT and hopefully that will take care of all the issues and bring back the machine to fully operational. Hopefully this information that you provided might be useful to someone else in the future.

(04-09-2022 11:05 PM)teenix Wrote:  Attached is scope shots of a HP-25

Reset - 50mS/div, a bit noisy, but my setup probably not ideal.
Sync - 500us/div
ISA - 50uS/div

The sync is as is because of the code it is running during the key wait loop. Sync stays LO after an IF instruction in the code. This tells the ACT that the code value itself is a ROM address.

displ5: 0 -> s3 <<< sync HI
if s3 = 0 <<< sync HI
then go to dsst4 <<< sync LO, goto ignored here as S3 = 1
if s15 = 0 <<< sync HI
then go to displ5 <<< sync LO, got executed as S15 = 0 (no keypress)

The ISA bus is showing the code executing for the above.


For the HP-27 it will be executing...

if s15 = 0 <<< sync = HI
then go to $08A <<< sync = LO, got executed as S15 = 0 (no keypress)

so the Sync line will be alternating HI and LO for each instruction.

cheers

Tony

[calc-calcs]

It lives again!

Received the new Panamatik ACT today after almost a full month of waiting for it to finally arrive, and I am happy to report that it worked on the first try. So the original ACT and/or the RAM chip where definitively toast as suspected.

On first test, all keyboard keys seem to respond and work perfectly without even a hint of bounce and all have the right click feel. Then tested all basic and most higher functions and all seem to be working as expected.

Guess now all is left is to break open the battery case and put in two new Eneloop AA cells. Have done this a couple of other times in the past, but on this occasion seem to be having a bit of difficulty fully separating the two halves of this stubborn battery case. Its been a while since I did this the last time but can't remember it being so difficult.

First used a sharp Xacto knife and went around and slightly into the whole seam. Then carefully using a very small jeweler hammer tapped the battery case from all lateral sides around the seam while wrapped in a protective cloth, but so far made very little progress in completely breaking the seam. There are some small areas of the seam that are starting to separate but there is about 80% still stuck together. I also seem to have hit a plateau with the tapping and for a while now have not made any further progress.

I wonder if carefully heating the plastic just a bit around the seam with a heat gun would help to facilitate the separation the seam with some more careful taps? I have even tried wrapping the battery case in the cloth and whacking it into the carpet, which always seems to work quite well to open sealed "wallwart" power supplies, but in this case that did not help either.

Anyone have any suggestions on this?

Attached a few pictures of the ACT installation and test.

[mfleming]

(05-12-2022 04:49 AM)calc-calcs Wrote:  Guess now all is left is to break open the battery case and put in two new Eneloop AA cells. Have done this a couple of other times in the past, but on this occasion seem to be having a bit of difficulty fully separating the two halves of this stubborn battery case. Its been a while since I did this the last time but can't remember it being so difficult.

Congratulations on the restoration. Always nice to get one of these little wonders working again. Although not an exact answer for your question, I've used an alternative method for restoring the battery case that I read about here. Simply remove the dividing strip that runs down the center of the underside of the case. You can then extract the old NiCd's and replace with Eneloops. Add your own springs and be sure to mark the polarity of the batteries inside the case, or on the bottom of the case. If you're a stickler for maintaining an original look for the case then you can simply cut one end of the strip at the bottom (near the contacts) and bend it upward to insert or remove batteries.

Give it a try, it works well and doesn't require taping the sides of the divided case to keep the two halves together.

~Mark

[calc-calcs]

I've seen it done to some battery packs and I'll admit is a practical way of dealing with it. But personally I like to avoid cutting off that middle section and keep it as original as possible. Once the batteries are inside and thanks to the spring tension, its very easy to charge them individually with a set of solid #22 wires shoved in front of the battery contacts and with a regulated power supply. I don't like to charge both batteries in series at the same time.

My method might be too much for others but basically I just maintain-charge the batteries in my calculators every 6 months or so, so its not a big deal.

Thanks for the tip anyhow.

[calc-calcs]

Keyboard problems.

I have been using the calculator quite a bit to put it through its paces with the manual in hand, and started noticing that the [2] key was sometimes behaving weirdly. Most times it would put a 2 in the display, but sometimes it would be a 3! There was clearly still something amiss so I disassembled the calculator and took another look at this key. By putting the keyboard side onto a strong light source I could see there are some blue-green areas inside the keyboard area shimmering though a bit but that are completely inaccessible. Unless the keyboard is taken apart, which I understand is something that needs to be avoided unless there is really no other alternative, as the end results of the reassembly are not always optimum. So instead decided to put some Deoxit red into the small hole above the key. I had done this on one of my HP25 that had quite a few bouncy or non fully responsive keys with excellent results which brought the whole keypad back to like "new" operation.

So while I was doing this for the apparently faulty [2] key in the HP27, why not put a bit of Deoxit into every key as a preventive/proactive manner? Well what happened next told me that this was NOT a good idea. The whole keyboard became inoperative. I confirmed this by swapping in the keyboard from my test HP21, and sure enough it was working fine. Swap back the HP27 keyboard and nothing, not even a single key. Furthermore I found that the resistance between most rows and columns was way bellow 1M ohm, sometimes just 200-300K. The Deoxit must have reacted with the corrosion and had created conductive paths that completely obliterated the keyboard scanning lines. As a reference, on the untouched HP21 keyboard there are around 3-5 mega ohms of resistance between most columns and rows, although one or two are actually just above 1 mega ohm.

So now what, disassemble and clean the mess? Before the "nuclear" option I decided to give it a try by connecting the ohm meter between one row and column to monitor the resistance of one key, and then putting a bit of IPA into the hole of the key and then twist a small piece of fine multi-strand wire in there to see if it would clear up the problem. But it did not, zero progress. So now one step closer to having to take the whole thing apart. Ugh.

One last try, use a can of compressed air and blow some into each key hole, what the heck. And that did the trick. Now the keyboard is working again, and even a few tiny bits of blue stuff came flying out the side of the keyboard but most is still in there. So I guess time will tell if this will hold or a keyboard disassemble will still be needed if the problem returns, but I am aware this keyboard has very likely a finite lifetime before it needs to go into major surgery.

Calculator Battery Contacts.

One of the battery contacts on the board is really bad (- side as usual). I did of course fully clean off all the corrosion but that did not help too much. While testing the calculator with another battery pack, it was still not making a reliable contact and often the display would flash some random digits, or just turn off and on midway into calculations, etc. Sometimes the flash was rather bright, so needed to take care of this before I damage any LED segments or even worse one of the driver chips.

There is no way to restore the round metal head of the negative contact, it was too corroded and completely lost its plating. So for now as a stop gap solution soldered a bare wire (leftover from a resistor tail) on one side of the contact where it attaches to the board, bent it over and in front of the metal head of the contact and soldered it onto the other side of the board as well. This should provide a much better new mating surface with the battery. Did the same for the other contact as well, so both would be even. The battery is now a bit tighter to insert, but the springs inside the battery pack had enough travel to take up the additional required clearance, and actually helped as contacts are now being held much tighter.

In any event battery contact problems now seem to be a thing of the past. Probably not the best long term solution, but for now it seems to work quite well.

Battery pack restoration.

Well finally managed to open up the very stubborn battery pack. Had to brute force it a bit and paid a price with an unwanted crack, but there was simply no other way. I ended up having to fully cut through all the seams where if was possible but even so there are some inaccessible areas. The side of the contact spring that goes towards the negative battery side was a bit tarnished, but otherwise ok, so I just cleaned it up with the Dremel and a small rotating brass metal brush until it was shiny again. Two new Eneloop cells and temporarily close the pack up with a couple of strips of magic tape on the sides. So far it seems to be working just fine, no contact issues with the springs so far.

Return some of the gold inside.

Lastly, I decided to reinstall the RAM/ROM chip, which is no good outside the calculator anyhow. And out of curiosity also wanted to test it again, and check its checksum. The Panamatik ACT easily allows for this, and to switch between using its internal ROM or the original code on the external ROM chip.

Turns out that with the external ROM all 6 trigonometric functions still stop working just as they did at some point when the original ACT was still operable. Also PI just returns zero as it also did before. Needless to say the checksum between both ROMs are wildly different. Out of curiosity I could also have gone over each line of code and directly compare between both ROMs line by line by switching back and forth to find the defective blocks. But given that there are just too many lines of code, it might end up being an unnecessary heavy burned on the two keys used to scroll up and down. And in the end there is nothing further I could have done to correct the problem anyway.

So far so good.

[mfleming]

(05-16-2022 05:52 PM)calc-calcs Wrote:  Two new Eneloop cells and temporarily close the pack up with a couple of strips of magic tape on the sides. So far it seems to be working just fine, no contact issues with the springs so far.

Just an idea to consider. I'm in the process of replacing the Eneloop batteries in my Woodstock battery holders with 4/5 AA NiMH cells. Their capacity is lower, 1300 vs. 1900 mAh, but full size AA cells tend to squeeze the holder spings flat and make the holder difficult to remove. Due to the shorter cell length you need to shim the springs forward a bit so the battery surfaces at the other end make firm contact with the PCB terminals. The batteries are flat tops so there's plenty of surface making contact with springs and terminals. The battery switchover puts less stress on calculator and holder.

~Mark

[calc-calcs]

For what its worth, in my experience when using 2000mAh Eneloop AA cells they do not really seem to stress the pack much, although it certainly feels a bit tight on removal, probably a bit more than normal. I have never removed a battery pack with original NiCads in it to have a point of reference of how much force is normally needed, but it seems its never too easy. The added length of the positive tab on the Eneloops would certainly ad a tiny bit, perhaps 1/16" or less which I would think the springs would still able to take up.

I have several Woodstock calculators all equipped with Eneloop cells, and from time to time I remove the cells to recharge them individually on an external charger. But I have so far never noticed that the contact springs where getting too flat or compressed to make proper contact.

One thing I have also done sometimes is to hook a short piece of desoldering braid in between both springs and wrap it around each spring once at the point where it contacts the battery to ensure more reliable contact between both batteries.