Page 1 of 1

Cherry G80-3000 typing itself

Posted: 25 Jan 2014, 10:59
by 16bitboy
Hi!

I own this keyboard since a year and a week ago it started to type random letters itself. I've checked and it's not a virus nor a software problem not happening with other spare keyboard I have. It happened at random times but now its constant. I've unmounted the case and cleaned the board of dust and dirt but I can't see anything that could seem wrong at first glance.

Could someone give me some pointers to what should I look for?

Thanks!

Posted: 25 Jan 2014, 11:22
by Muirium
Tried another computer?

If the keyboard has problems, it could be leaky capacitors inside (which just need replaced) or something else shorting out.

Posted: 25 Jan 2014, 11:36
by 16bitboy
Yes, it fails wherever I connect it :/ I'll look for that, thanks!

Re: Cherry G80-3000 typing itself

Posted: 25 Jan 2014, 13:56
by Broadmonkey
I don't think it's the capacitors. How old is it? And is it what characters does it output, is there a pattern to it?

Posted: 25 Jan 2014, 15:04
by adhoc

Posted: 25 Jan 2014, 16:04
by Findecanor
Is it the same letters? Does it happen as you type, or when the keyboard is idle?
"Chattering" is a common problem with Cherry MX switches - it is when a key that is pressed once emits two or more presses, but it shouldn't happen when idle.

My first thought was that a friend of yours would have plugged in a novelty dongle device into another USB port that emits phantom key presses, but if the symptoms occur when you plug the keyboard into another computer that you have control over, then that can't be the case.

Posted: 25 Jan 2014, 18:32
by 16bitboy
It happens both when is idle and when I type, just connecting it starts with the key presses.
I'll open it again and see if there is some capacitor I can see in bad state. I've run the software adhoc mentioned and I've observed the keys that are causing most key presses are those on the middle row of letters (asdf...) and sometimes the numbers 1,2 and 3 of the numpad and also the intro key. This is some log about that:

Code: Select all

21:22.0758 F (0x46, BIOS 0x21) DOWN
21:22.0763 D (0x44, BIOS 0x20) DOWN
21:22.0765 A (0x41, BIOS 0x1E) DOWN
21:22.0768 ; (0xC0, BIOS 0x27) DOWN
21:22.0770 K (0x4B, BIOS 0x25) DOWN
21:22.0772  (0xE9, BIOS 0x71) UP -> 35016ms
21:23.0005  (0xE9, BIOS 0x71) UP -> 234ms
21:23.0038  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0071  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0104  (0xE9, BIOS 0x71) UP -> 32ms
21:23.0137  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0170  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0203  (0xE9, BIOS 0x71) UP -> 34ms
21:23.0236  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0269  (0xE9, BIOS 0x71) UP -> 32ms
21:23.0302  (0xE9, BIOS 0x71) UP -> 34ms
21:23.0335  (0xE9, BIOS 0x71) UP -> 32ms
21:23.0368  (0xE9, BIOS 0x71) UP -> 34ms
21:23.0401  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0434  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0467  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0500  (0xE9, BIOS 0x71) UP -> 33ms
21:23.0533  (0xE9, BIOS 0x71) UP -> 33ms
....(It continues like this)

Posted: 25 Jan 2014, 19:46
by adhoc
Yikes! Some of response times are huge there!

Can't really help any further, but I'm sure otherwise will be able to make more of this!

Posted: 25 Jan 2014, 20:34
by IvanIvanovich
I got a used one once that did this. It turned out a bunch of switch contact leafs were deformed inside, and were extremely soft so would just make contact with itself in random fashion. I guess the keyboard was heavily used and possible exceed the switch lifetime. After I replaced those switch that would type themselves the keyboard was perfectly fine again.

Posted: 25 Jan 2014, 22:48
by terrpn
sounds like you maybe got a keylogger virus.......

run whatever spyware, antiviris, trojan-root removal software you got

Posted: 28 Jan 2014, 10:51
by 16bitboy
No, I've discarded that possibility already. I've tested in other computers that same keyboard and connected a cheap one. Yesterday I opened it to look for capacitors and resistors. There is no one of them in the board that I can spot. If they are on the board, may be they are very small or integrated somewhere, I'm not very fond in electronics :/ But something tells me it could be the little board that gets the cable connected to it. I mean they say this keyboards are so durable because of their switches. I've used it for only a year, so I don't think it could be the switches. All feel like when I bought it.

EDIT: Also, in the logs you can see, the spammed key is an UP event of 0xE9 (Win Code) which looking for this key on google I've got its an OEM specific keycode, not directly mapped to a switch.

I'm also thinking in buying another keyboard if I can't repair it. Currently I'm using a spare and cheap membrane logitech keyboard (a 9€ one) and well its doing its job but I'm not very comfortable typing on it.

Posted: 28 Jan 2014, 13:52
by kint
If it behaves the same-same on different computers it is the keyboard, it's obvious.
You're saying daughter board, that implies the modern revision of the G80-3k, that has the whole controlling done in that tiny board and the switch PCB is just the matrix. Capacitors and resistors are solely SMD then. I just looked at the matrix of such a board I own, and the whole home row (ASDF) lies on one PCB trace starting with A, jumping the Return, cursor block, and then splitting and heading for controller and NUM 1, 2, and Enter. Num 3 shouldn't be affected much as of my boards revision. Not saying your board has the same routing and that this trace is causing this to happen but I'ld investigate that. May be a broken trace or some conductive dirt laying on the PCB or jumper wire. it certainly is informative that this dedicated row is affected.
Unmounting the switch PCB by deplugging the case/controller and removing the caps should be the obvious as for that.
I can't use the above log much, to me it's more interesting what keys fail exactly, and also what layout you're on.

Posted: 28 Jan 2014, 15:58
by 16bitboy
I'm using the spanish layout. I've tried to unmount the switch pcb from the controller one, just to know if the controller still sends events without the switches connected, but they are connected by a (strip? not sure how to call it in english, it's a set of wires wrapped with some kind of cloth or plastic in a flat shape) that is soldered in both ends to the boards so I can't disconnect them without having tools for de/soldering small pieces. If those keys in the log arent the ones causing trouble, I don't know which ones are, because the keyboard doesn't respond to any key press, I think due to the spamming of that event (UP event of 0xE9 Win code). Sometimes it stops doing the spamming, and I can type, but is rare and still misbehaves when responding to my keypresses by sending may be two keypresses or more of random keys.
I've also removed all the caps of the alphanumerical part of the keyboard but I haven't seen anything out of place :/

Posted: 28 Jan 2014, 16:16
by kint
oh, didn't know that, my board has Female Pin Headers like these on the controller side, and the wire bands can easily be pulled out of them.
I just took a look at the spanish layout, http://en.wikipedia.org/wiki/File:KB_Spanish.svg to justify my idea that it's just one trace being affected, but unfortunately the ; ain't on the same row on a spanish keyboard.
Take the PCB out, remove the caps, check the traces and PCB for a crack and clean everything again. You can also try to open a simple editor and whilst the keyboard is connected flex and twist the PCB lightly to see whether it affects the outcome.

Posted: 28 Jan 2014, 16:24
by 16bitboy
Yes, ";" its not in the same row, but i can tell that software that I used to get the log is referring to the "ñ" key as ";" so it's really the ñ key and that would be the same row, I think you are right on that one.