Texas TI-58: Repairing the keyboard bounce and miss to register keystrokes

[jebem]

I have this Texas TI-58 calculator missing a few parts used to repair other Texas machines.

One of the missing parts was the keyboard, used to repair one of my TI-59.
The original TI-59 keyboard had a few troubles, including battery acid leaking light corrosion and it failed to register a few keys and sometimes several keys were registered from a single keystroke - a classic case of bouncing, I would say.

So as I had nothing to lose, I decided to disassemble the TI-59 keyboard piece by piece, fix the issues and put it back on operation on my TI-58.

Looking to the keyboard white plastic frame, we can see a number of small holes in front of the crimping/welding points. I think Texas made a less than perfect job when welding these joints, resulting in a lot of failing keyboards.

In my case, these mechanical/electrical joints were broken in several positions, and I believe this was causing both bouncing and missing keys as well.
Pressing a key can either cause multiple contacts spaced in time in a way that the debouncing software routine can't handle, or a total failure to register.

After reading so much reports from others here and there, I have noted that some people can fix some of these keyboard issues by just manipulating the original foam that sits between the plastic keys and the domes.
This makes sense to me, as the foam makes some pressure to the domes forcing them to make a better contact with the keyboard frame wires.
A thicker foam should give better results as well. I did it on one TI-57 some time ago with some success.

To test my hypothesis, I decided on soldering all the 9 dome's connection points where originally it were crimped/welded at factory.


Separating the 9 dome rows from the keyboard assembly.

Once the keyboard is desoldered from the main board, the starting point is to remove the plastic cover from the keyboard base, where the 9 dome rows comes attached to it.
Some of the joints of the 9 rows were lose, creating the issues. Other joints were good, so some force is required to pull all the set apart from the keyboard base.
The plastic cover and the metal domes are resistant, though, so no damage was caused by this procedure.

Next step involved the separation of each one of the 9 dome rows from the plastic cover.



Preparing the keyboard base 9 wire rows dome's joint points by tinning them.
This is a delicate operation. Needs to be done fast and precisely. The base plastic is very soft and melts easily if applying too much heat for to much time. One to two seconds maximum, using a small but powerful solder gun.



Prepare the 9 metal dome rows by tinning the joint points on the both sides.
Very delicate operation, as one needs to apply the least amount of solder as possible on the underneath side to tin just a small area. Otherwise it can short circuit the 5 columns wires as well.
Tinning the underneath side first:



Tinning the upper side of the dome row allows to do a fast and precise soldering operation when putting it back into the keyboard base. Before tinning it, it is necessary to remove a small portion of the strong glue layer that covers the entire dome row.
Tinning the upper side next:

[jebem]

Reassemble the 9 dome rows

Put back the 9 dome rows by soldering them to the keyboard base at the 9 rows joint points.
I used tooth ticks as dome's alignment guides by inserting them into the two existing small holes in each one of the 9 rows.




Re-apply the thin plastic cover onto the 9 sticky dome rows
Because the dome upper sides are covered with a very strong adhesive layer, it was enough to fix the plastic cover back in place.

Keyboard base populated with the 9 dome rows:


Plastic sheet cover back in its original place.

[jebem]

Rendez-vous time and Testing

After soldering back the keyboard to the calculator main board, it was time for testing.

Power ON, and CLR and typed the sequence "123456789" without looking to the display.
Result: No duplicated entries and no missing keys.

EDIT: I did miss one key entry above. So, bounce is gone but typing too fast can still miss one key register.

It is too soon to say that the bounce issues are gone for good.
But at least it didn't fail so far.



P.S. - One small note for those who want to try this procedure.
The 9 metal dome rows are asymmetric. By mistake, I got the metal rows installed 180 degrees left to right, meaning that I had to solder them at different points from the correct original crimping points. It doesn't affect the normal key registration, though.

[Marcus von Cube]

Thanks for sharing your findings. I hope I can apply a similar fix to some of my TIs.

[jebem]

(12-06-2014 07:58 PM)Marcus von Cube Wrote:  Thanks for sharing your findings. I hope I can apply a similar fix to some of my TIs.

You are welcome, Marcus.

But I did miss one key to register entry above.
So, bounce is gone but typing too fast can still miss key entries.
I have yet to assemble it and do the final test in the proper way.

[Thomas Radtke]

(12-06-2014 06:52 PM)jebem Wrote:  After reading so much reports from others here and there, I have noted that some people can fix some of these keyboard issues by just manipulating the original foam that sits between the plastic keys and the domes.
This makes sense to me, as the foam makes some pressure to the domes forcing them to make a better contact with the keyboard frame wires.
A thicker foam should give better results as well. I did it on one TI-57 some time ago with some success.

I was one of those writing about it. Actually, the foam in my 59 had dissolved nearly completely, so I removed what was left and cleaned the PCB. Unil today, I operate it w/o the foam and it doesn't show any bouncing or missing keystrokes.

OTOH, the same procedure unfortunately failed on another TI (a car computer sold exclusively in Germany).

Anyway, the message would be that this piece of foam isn't required.

[jebem]

(12-07-2014 11:31 AM)Thomas Radtke Wrote:  Anyway, the message would be that this piece of foam isn't required.

Hi Thomas!
Thanks for sharing your experience here.

Agree, the foam is not needed when the keyboard is not defective.
That is why I test the keyboard having the machine main board out of the case when you can not use the foam.

However, the foam helps to reduce the plastic keys rocking/jingle movements in the slots, giving a more firm key presses.
Also, the foam helps to maintain the domes rows against the 9 row wires when the domes rows are lose (defective) from the wires (that is the main issue with these keyboards as far as I can tell), and this may reduce the key registration issues.

[circuityrboy]

This might be of interest:
I acquired an LED TI Programmer on eBay. In fair condition, it ran when rechargeable power pack was replaced with a PP3.
However, it proved almost useless owing to frequent keyboard bounce/non-entry. This was worst in Hex mode down the F to 3 column of keys. Gentle 'key exercising' didn't help.
At that time I had not seen anything online about dismantling so I inspected inside the battery compartment to see how things went together.
With keys down and supported at edges on my fingertips, I pressed with thumbs on the white back of the keyboard to see if there was any movement. Mainly at the edges, but also flexing the centre of the white panel.
I didn't learn enough to attempt dismantling, but I then found that the keyboard was restored to clean key action.
Obviously, I recommend trying this before undertaking any TI surgery.

[magore]

I have a few TI-59 calculators with keybounce. I can confirm that repairing the 9 row solder joints resolved my keyboard bounce porblems. However my membrane tore - and I was being very careful. You can replace the membrane with Polyimide or Polyester tape that uses Silicone Adhesive. I soldered the 9 row wires and it worked perfectly afterwords. I looked up what type of material membranes were often made of to find a replacement. I ended up getting both types of tape from Amazon in 4" wide rolls. I used Polyimide for my first test and it worked great. If I were to try this again I would likely just try cutting the membrane from the top exposing just the 9 rectangular row wire junctions holes and soldering without removing the membrane. A sharp pointed exacto whould do the trick to cut them out. You need a fine tip and steady hand on the soldering iron but this works well.