HP Forums

Full Version: Heat staked keyboards with sticky, bouncy or corrosion present keyboard PCAs
You're currently viewing a stripped down version of our content. View the full version with proper formatting.

Thought I would post the following which is found embedded in a previous topic soon to be lost. I hope this helps the community at large. It was in response to the question:

What did you mean by "washing" the PCA's?

When I restore a calculator with a heat staked keyboard, especially one with corrosion due to either battery leakage or out gassing I use a wash on the PCA.

a: To remove any vertigris (green or white precipitate) I rinse
(wash) the PCA with a mild vinegar solution followed by a
distilled water rinse and then 95% anhydrous ethanol (in Canada
thats known generically as 95% rubbing alcohol although I have
seen it labelled isopropyl alcohol).
b: For stickey keys, a very dilute soap and water wash with water
rinse and anhydrous ethanol rinse
c: If there is no corrosion or stickey keys I use the water wash
followed by the anhydrous ethanol rinse.

Of course, sticky keys with corrosion would get both treatment a and b.

Heat staked keyboards with bouncy keys or even intermittant keys respond well to this and I have resurrected bad keyboards on:

Woodstocks
Coconuts
Voyagers
Pioneers
Stretch Pioneers

Of course the Classics and Spice keyboards can be diassembled and cleaned mechanically.

Cheers, Geoff

Fill the glass to a level below the LCD or LED if they are soldered in place. In the case of the woodstocks the LED module can be removed.


Edited: 11 Aug 2009, 4:30 a.m.

Thanks Geoff. Very interesting. I do have one keyboard/display PCA from an HP-45 with a bit of that verdigris. A lovely aqua really. Never thought of immersing it. But you do not submerge the IC's or LED chips?

Dan,

I try to stay away from the ICs but if there is corrosion/vertigris present, you can be sure it is active, especially in humid environments. When present it will be a certainty that the corrosion will enter the chip via the connectors.

The LED block however has wire bonded (extremely thin) connections. These too can suffer from the outgassing or battery leakage. It is a trade off. They will also deteriorate. You have to make the decision whether to hault the corrosion and risk harming the LED block or leave the vertigris and keep the calculator dry (silica gel).

I do stay away from the LED block if corrosion is not present as the water may stain or leave a dry film (depends on the hardness of tap water, another reason to use distilled) under the block where it is seen but cannot be removed. This applies to LCD panels on heat staked calculator also.

So the thinking by me is:

1  include the IC's if corrosion is present around the connectors.
the vinegar wash may be applied with a brush to limit exposure
2 include the LED if you are concerned enough with the
possible corrosion.
3 try to keep the water and other washes clear of the LED block
and LCD panels.

I had a bottle in the lab with a spout which originated in the bottom of the bottle (name?). If you can get a hold of one of these you can direct the wash at specific parts of the PCA.

Hope this helps,

Geoff.

P.S. I just completed two HP-41CX halfnuts. Soaked both keyboards with logic PCA attached in a glass of soap/water mix overnight. Followed with the rinse and a two day dry box. Instead of the dry box prop the keyboard PCA vertically and place in front of a fan for a couple of days. Both keyboards with intermittent and faulty keyboards as well as rocker switches came back to PERFECT function for both feel and logic response.

Edited: 12 Aug 2009, 3:57 p.m.

Well Geoff, I'll give it a try. Do you let it soak in vinegar overnight, or just go at it with an old toothbrush?

Also, dumb question, what is a "heat staked" keyboard? (The only keyboard PCA's I am familiar with are the classics).

I have a candidate in mind...

Ah, a picture is worth a thousand words.

Clean the OFF/ON contact with some alcohol to remove the silicon grease remnant.

Using a paint brush, natural bristle or a SOFT tooth brush and vinegar clean around the LED block and the ICs.

Soak overnight to just below the level of the two ICs marked 1820-1061.

If you had key contact problems, cut long strips of construction paper (heavy weight paper) wide enough to fit through the key contacts. Press down on a contact while dragging the strip through. The strips could be damp with vinegar also.

Rinse with water and alcohol.

Woodstocks, Voyagers, Pioneers and later families had the keyboard PCA heat staked (plastic welds) which preclude diassembly like the classics and spice. Removal of the heatstakes is the only way to gain access to the keys and key contacts. Once apart they are difficult to restake. Voyagers have over 40 stakes holding them together.

Of course the "book" will have some methods explained to re-stake the keyboard ;-)

Cheers, Geoff

Here is a Museum link that I wrote about a 45 restoration last year:

HP 45

I subsequently modified this to accept the Quartz crystal and enter key modification to allow easy and accurate timing.

quartz crystal

Cheers,
Geoff

I cleaned the kybd/LED PCA pictured above by soaking in a vinegar/water solution, up to the top row, for about 30 minutes, then took a toothbrush using the same solution to scrub around the chips.

The result is pictured below. The keypad cleaned up nicely. I have tested each key and I get a good positive "HP click" and the calculator does what its supposed to do. Obviously the LED drivers and the LC circuit are all working.

The only bad news is that there is inside damage to the LED chips. This appears to get progressively worse from left to right, and the rightmost chip appears to be working fine. I do not think the damage to the LEDs was a result of the cleaning.

So, two thumbs up for the technique. Hard to believe it is the same PCA as I photographed earlier.

-- Dan

Good job!

Now find a defunct keyboard PCA and swap out the LED blocks. I just swapped an LED block from a spare HP-55 parts machine into an HP-45 and an HP 80. In fact the 45 in my pictures above had one decimal place that was not working. ONE decimal place in the 5th division, amazing, but now it works. It was to be a spare parts machine as it did not work either!

Once and awhile just the keyboard PCAs show up on Ebay or a completely dead machine shows up with no chance of resurrection. I keep these around for LED spares, drivers and etc.

Your correct, if you kept the solution clear of the LED block or if the LED block was in excellent condition and the solution was introduced the LEDs would still be functional. Judging from the amount of corrosion and the diameter of the bonding wire in the LEDs i am not surprised that some of the bonds let goe. Remember, each LED dot has a wire as opposed to the bar LEDs from later periods.

Cheers, Geoff

Edited: 13 Aug 2009, 12:15 p.m.

As it turns out I do have another PCA with bad key bubbles - cracked all the way through. But it has a good display.

I have recently been studying the LED driving technology. Very interesting, using inductors to provide the current. Usually when I put it on a scope it is fairly obvious where the fault is. There is one problem that alludes me. I call it ghosting. It is where one LED has a 3-4 segments that are faintly lit when those same segments are lit on some other digit. Any ideas what causes this?

I have restored a number of Classics. I'll buy dead units from auction and restore them and sometimes sell. Often the problem is as simple as crud build-up on the power switch (I think they over applied the silicon lube). I've learned a lot and it helps finance my collecting habit.

I have one 65 that has a really minor ghost, visible at night in the dark.

I remember someone, somewhere swapped a resistor to limit current to the LED driver or Block, can't remember, and this fixed the problem.

Anyone out there with an idea and specific location for the resistor, that would be helpful.

Congratulations on reserrecting another HP-45! Something I never thought of but convenient; flipping the logic PCA over while applying power, duh, no why didn't I think of that!

;-)

After staring at scope images of cathode and anode signals a while and coming up empty, I decided the best approach was the time honored brute force method. I swapped LED chips. The ghosting followed the chip. Therefore the problem is the LED chip itself.

I suppose it may be possible to add a resistor somewhere to mitigate the problem by dimming the LEDs, but I don't see that it would solve the problem.

I did see a minor difference in scope signals when comparing with a known working calculator.

Also in the before picture note a very slight ghosting of segment e in the digits immediately left and right of the bad digit. This disappeared after the swap. I had not even noticed these segments lit until I saw it in the photo.

I swapped the right and center LED chips. The ghosting was on the left most digit of the right chip.

Before the swap:

After the swap:

Thanks Dan, emails to follow!

Away from home at the moment, but thanks for your email.

Looks like, as you said, a LED Block problem and not the driver. Well that is better news then the driver problem. I have not been able to find the article on the resistor addition so anyone out there who knows, that would be a help.

At least I know that I can swap the HP-65 ghost block out if need be as I do have a spare 65 blocks.

Does the scope still show a difference when compared to the 'normal' calculator. Of course, what is normal in a run of two!

Cheers, Geoff

I think I found the post that referred to a resistor fix.

http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv019.cgi?read=150663

Quote:
I decided that maybe the driver for the left digit needed to be pulled up a little. This was because it just seemed to sink current during the time that the other digits were driven, causing the cross-talk from the other digits.

Indeed the problem vanished with a resistor of 150 Ohm between Vcc and k15. To avoid possible damage and to stay on the conservative side I chose to use 200 Ohm. In the pictures you can see that the display is now much better looking, though a little less bright than the other HP45 that I have.


I am thinking that some ghosting could be due to stray capacitance or a resistance path between traces due to crud. I think what he's done is attenuate the current, sort of masking the problem.

Dan,

I agree with the fault being in the LED block!

I have a 45 with a ghost problem and missing decimal point in the middle of the ghosting block.

Swapped with a spare 55 block and the decimal is back and the ghost gone.

Another salvaged 45!

Cheers, Geoff