Unusual tactile dome fault 74HC165D (Ohaus Explorer Pro)

[Rjc987]

Hi guys, hoping somebody can help me figure this one out. I just purchased a Ohaus explorer pro That was listed as non-working, for parts, received it and turns out the power adapter contacts (female) were stretched and not making contact so I pinched them back together a little bit and the unit powered on perfectly. Everything was working great, including all the buttons, however after 10 minutes or so, the button click sound started going repeatedly, and when I looked over It was cycling into the menu, exiting, entering, etc over and over and the buttons were unresponsive. So I unplugged the unit and plugged it back in and this time it immediately started making the button press sounds before the screen powered on and the buttons were again unresponsive. So I unplugged the unit and took out the I/O board to inspect it. Everything looked good and I did a little work fixing the nice big crack that FEDEX put in the outer shell, and after about an hour or two, I reseated the tactile dome membrane, reconnected everything and plugged it back in.

Again it worked great. This time for about five minutes and then started cycling again. So I unplugged it and tried disconnecting the membrane entirely and powering on by directly shorting the power button pin to its common. It powered up and worked perfectly. Left it for at least 20 minutes with no cycling and perfect operation. Reconnected the membrane and powered it back on and the cycling came back. Tried disconnecting the membrane while it was cycling through menus and the cycling immediately stops and the buttons work fine again by directly shorting them to common.

There’s a definite correlation between how long the unit has been powered off and how long it works before cycling. That leads me to believe there has to be a capacitor somewhere that’s playing a roll, no?

The tactile dome membrane has 12 pins servicing 10 buttons, with 2 commons handling 5 buttons each. The commons are commoned together on the board and come in through a ferrite and a 33 ohm resister to ground. Each button goes through its own ferrite and branches off to its designated input of a 74HC165D and to a 47kohm resistor. All the resistors are commoned and connected to the Vcc pin (16) of the 74 before passing through a single smd cap to ground. I tried to measure the cap in situ and it slowly cycles through the ranges until eventually showing OL - I tried adding a 0.5nf cap but the problem remains.

The menu button is the only one being “pushed” (I assume it goes stuck closed). I measured the resistance of the buttons and the menu button is right in the middle at about 100 ohms, and it’s wide open when not pushed even when flexing and moving things. Also measured the capacitance between the buttons and the common and they’re all around 2nf. I tried cleaning up the connector contacts too but no joy. The fact that everything works perfectly until I plug in the buttons really gives the impression that the fault is there - after all they’re just non-matrixed open contact buttons so how could the fault be anywhere else, but then again, the fact that it works perfectly after being powered off for a while and only appears after some time doesn’t make sense for a purely mechanical issue.

I’ve never seen anything like this so before I break out the logic analyzer, any ideas on what to try??

[Andy Watson]

Is there any possibility of liquid (water, coffee etc) ingress into the keypad, or the remains (metal salts) of a previous ingress ?

[Rjc987]

So the whole keyboard is laminated between 2 thick layers of plastic so there doesn’t look like any way for liquid to get it. Also looking at the domes, they all look clean and the resistance is the same for all of them. 

[nigelwright7557]

The old Sinclair Spectrum membranes used to wear out and start shorting out.
Maybe same problem here ?

[wraper]

Such switches do fail, water ingress is not even necessary as conductive dendrites between contacts can form even without presence of liquid, humid or corrosive atmosphere is enough. There is probably leakage in tens of Kohm somewhere. Heat the thing to loosen adhesive (you can use kitchen oven if it can do ~70^oC), carefully remove the dome sheet and clean flex PCB and metal domes (if needed) while trying to not touch adhesive as otherwise you'll contaminate it and it will no longer stick. Then put dome sheet back starting from one side to not entrap any air. I've done this with some laboratory balances, IIRC some Mettler and Sartorius.

[wraper]

Just be careful to not pull apart wrong layers. There could be a docorative top film painted on other side and laminated on other film which actually sits on metal domes on sticky adhesive layer. You don't want to pull these  films apart and there likely won't be any sticky adhesive between them. Not to say you'll damage paint and it will look ugly. You want to pull both of these films and adhesive together.