Weird problem in HP 20S

[Marcovecchio]

I'm facing a very (VERY) weird problem in a HP 20S I'm trying to repair.

I bought it defective, some keys not working. I must confess that I didn't test it thoroughly before opening the case, I was very confident it was the aged rubber pressing keyboard contacts. So I opened it right away.

It was very dirty, pcb solder was very oxidized, so used a very delicate soldering iron to redo all soldering pads of the processor (I'm experienced on SMD soldering), and I'm sure there are no shorts or any other problems. Then I thoroughly washed the pcb with isopropyl alchool. It looks like a new PCB now, and looks perfect after detailed examination in the microscope.

Well, after cleaning the case, LCD, zebra strips, and replacing the keyboard rubber, I reinstalled the LCD and PCB, pressing it down with some clamps (to avoid stress on the twist tabs for now), and now the calculator works for a few keystrokes, and then the keyboard dies completely. If I short the battery contacts, it resets and works for a few more keystrokes, and stops again. Then I tried again, but waiting 10 seconds before hitting any key, same result, so it's not the keystrokes that freezes it, it's time after power up. I have no idea of what to do!

Some more information:
- I can't confirm if the calculator freezes, or only the keyboard stops working. So it may be a keyboard, PCB or processor issue.
- If I use this same PCB in another Pioneer shell, the same problem happens, so I can conclude the keyboard is ok.
- If I use the 20S keyboard with a 32SII PCB, it works perfectly, so I can be very sure the keyboard is not the problem
- This 20S is an older model, serial 3031B92619
- I tried with another HP 20S LCD and zebra strips, and the problem remains unchanged.

One interesting point: if I power up the PCB with no keyboard and no Pioneer sheel (pressing the LCD with my fingers), using an external power supply, I can use a metal wire to short keyboard pads, to simulate keystrokes, and this way the calculator remains working for several minutes with no problems. The "freezing after some seconds" goes completely away!

I have a theory, that something in the shell may be shorting something in the PCB, that may be seen by the processor as a stuck key, or something similar. However, I examined very carefully how the PCB touches the shell, and I can't see how a short could possibly happen.

Another similar theory is the creation of a capacitance or something like that, between two very close metal parts and/or PCB tracks, creating the same stuck key effect.

These theories relate to the fact the PCB works ok outside the Pioneer shell!

Well, this is the problem, I never seen anything like that, and any idea on why this is happening is very welcome.

Thank you all!

Eduardo