deskthority

Suppose you spill water inside a laptop, and the laptop no longer boots. The laptop functions perfectly if the keyboard is disconnected from the motherboard. If the keyboard is connected, the LED beside the DC jack illuminates (and blinks slowly), and that's it. The computer does not otherwise function — the fan on the motherboard (visible when the keyboard is removed) does not even spin, nor is there a picture or any beeps. Remove the keyboard and it functions normally.

From what I can tell, the keyboard contains no logic at all: the ribbon cable is just the matrix wiring, to a controller on the laptop motherboard.

How does the laptop know whether the keyboard is connected?

I don't know whether there's a pair of traces used for keyboard detection (such an idea would be news to me), or whether conductive contamination on the membrane is shorting together two row or two column traces in such a way that the controller gets completely confused and locks up. Shorting rows against columns should just trigger a stuck key warning in the BIOS, unless the BIOS is badly broken and hangs instead (this is a Lenovo).

It's very odd, and it just seems odd that a water spill would affect only the keyboard controller, although it's conceivable that water entered the chassis through the hole where the keyboard cable passes through.

Any ideas?

(PS it's not my laptop ;-)

The latter sounds about right. Any hole guarantees water will have gone in there too. But as disconnecting the keyboard removes the problem, it seems likely that the only damage is to the keyboard / controller. A membrane keyboard is a far larger target than the tiny crevices between pins and such elsewhere in a machine.

I'd try swapping the keyboard out for another. If manufacturers are so kind as to let you source them these days…

Muirium wrote:… it seems likely that the only damage is to the keyboard / controller …

Which? The controller is on the motherboard, which would mean motherboard replacement if it's blown. I don't have a spare keyboard module.

Water trapped between the membranes would just give you a stuck key error from the BIOS, unless the BIOS has a really horrible bug that causes it to crash before it even displays a picture if there's a key held. The motherboard fan doesn't even spin up if the keyboard is connected.

My guess is that the water's effects are entirely in the membrane keyboard. The controller is then being overrun with massive amounts of rollover — I've seen a spilled keyboard showing half of all its keys pressed simultaneously, months after the spill, via PS/2 — and the BIOS is crashed the very first moment it polls the thing. Again and again and again. BIOSes are born to be flaky. The fact the fan and all else is working when the keyboard isn't hooked up points the finger right at such a bug.

Another cause could be a giant short that's throwing off system power. I had that with a motherboard long ago. (The culprit was the metal plate it was mounted on; which was designed by some vile moron genius to be just enough out of shape that I was in big trouble. The computer would just play dead until I figured that one out. PSU fans and all. I could hear the capacitors though.) But, again, if it's not happening with the keyboard removed: it's the keyboard.

The best fix then is to pull out the keyboard and provide a Siig Minitouch, Poker or similar!

I suppose there's no reason why the controller can't be PS/2, and wired to understand the matrix sufficiently well as to allow more than the basic two keys plus modifiers. Would be surprising that anyone would care enough about rollover to have the controller not simply block pretty quickly. No idea who makes the controller; the keyboard itself is a Lite-On (as in Silitek), so maybe they do care. I don't know.

I'm probably not fully done with my MiniTouch (I took some photos of the switch contacts last night for the wiki), but it's too bad even for my enemies. (Even if I could get the switches back together.) The MiniTouch is another deep mystery. Pretty sure they're NOS rebadged as SIIG.

are you sure that the kyeboard is completely dry? I mean completely.

water may just short the keyboard which in turn shorts something else in the PC and the mobo is smart enough not to turn on to avoid more damage.

Is the keyboard completely dry? I don't know. So far as I know it's nothing more the switch matrix — the ribbon cable from the keyboard to the motherboard just appears to carry the matrix trace lines.

It's supposed to short out the matrix lines, otherwise it wouldn't register any keys. The worst you should ever get with a short inside a bare membrane assembly is a load of stuck keys. It would appear that BIOS is so horrendously broken that a few stuck keys are crashing it before any power gets applied to any part of the computer, which seems rather odd.

that seems impossible to me. there must be some circuitry that is shorted out.

I grant you, membranes are a bit odd. This is from a Logitech "Scarlet" (Dell KB-522 multimedia keyboard) — firstly, the controller board connection:
Both membranes overlaid:
The membranes separated:
Notice how numerous connections to the controller are shared between the two membrane layers? I never noticed that before. I would have to dig that keyboard back out to see if I could follow what they've done there. Electronics never was my strong suit.

Modern laptop keyboards are sealed and cannot be disassembled for inspection, so I don't know for sure where any of the traces go.

It's possible that water has actually shorted rows together, or columns together. What effect this would have on a controller, I have no idea at all. It would have to be a connection between pairs of traces that would represent something other than keys being held, otherwise the controller would simply read stuck keys. After all, it's a grid of switches! That's what switches do. Bypassing a switch with a short just has the same effect as pressing it.

Imo the odd thing here is the fan ain't running at all. If there is just some short in the keyboard like a stuck switch it should spin, albeit idling before/at POST. Keyboard comes after CPU and RAM -normally the fan should spin at that time already. Maybe there's some odd wiring inside the Laptop where a shortened keyboard prevents other components getting power. Just replace it then, the cause seems clear if not the reason. Either open the keyboard if the Laptop ain't worth a spare, or bath it in Isopropanol, which may cause it to break even more, be warned.

I've selectively masked off parts of the matrix until I got to the stage where the computer would boot and where any keys actually worked (as opposed to booting but having zero responsive keys). I've now got alternate columns working. Typing 6 and j inputs 6nj, while n works normally.

It looks like there's one or more stuck keys that do indeed stop the BIOS from functioning at all. After pressing power, there's a delay of around a second before the fan starts, and then another four seconds before the screen backlight comes on. (Total time to backlight on was 5 seconds according to my stopwatch.)

The keyboard blocks the BIOS very early on!