What is going on with this sleep button?
August 27, 2010 5:28 AM   Subscribe

Trying to repair a laptop, strangest behavior ever. Many more details after the fold.

I do laptop repair often enough, usually just resoldering a broken power connector, changing a display ribbon, that sort of thing. Right now I've got one that's driving me nuts.

One of my regular customers has an older (It's a Pentium M) Fujitsu Lifebook T-series, that's the tablet. Recently, it stopped turning on. I couldn't get to him fast enough, so he took it to another local repair shop. They replaced the cable but said it wasn't the issue, that in fact it was the nub that gets pressed with the screen opens/closes to sleep the machine, and that they couldn't fix it.

So what the heck, I offered to have a look.

Here is the exact behavior:
You lift the screen, it spins up, and after about maybe 5 seconds it just turns off.

So I thought maybe the sleep-nub switch was bad, so I opened the case to have a look. It's a teeny, tiny spring loaded nub. Here's the weird part.

If I press it once and let it go, the machine comes on. If I don't hold it in, it shuts back off. If I hold it in, it stays on and is running, but the backlight doesn't come on. If I let it go for up to about 1.5 seconds, the light will come on, keys work to enter BIOS, boot the machine, etc, but within about 4-5 seconds of letting go, it shuts off.

Also, if the machine is off, and I press the nub but do NOT let it go, it does not power on. No powering on until it's released.

Also, the power button does nothing in any state, regardless of the status of the sleep nub.

So I had a look at the power button, unfortunately it's hardly mechanical but seems to be fine.

So I'm assuming there's maybe a short in the sleep switch? My original thought was to simply remove it and bypass it to a contant-closed circuit, but like I said, the only way to turn the machine on is to close then open it.

I'm totally open for suggestions here. Even crazy ideas. Unfortunately the switch is micro soldered, so that makes things even harder.

I've considered a power supply issue, I checked for capacitor bleed, etc. Right now I'm flummoxed about the behavior and about why the backlight only comes on when the button is let go.

Hacks are welcome as well. :) I'm not even sure if it's possible to tell the system to ignore the sleep button, perhaps in the BIOS? I haven't had enough hands to keep pressing and releasing the button to stay inside the BIOS long enough to actually look at settings.
posted by TomMelee to Computers & Internet (4 answers total) 1 user marked this as a favorite
 
Don't know the specific computer, but I do a lot of troubleshooting.

Odds are that the switch you are dealing with is a Normally Open (NO) momentary contact. Because of miniaturization criteria, it is unlikely to have NC contacts or anything more complex than a single set of contacts.

(Usual failure mode of such a beast is either catastrophic open or excessive resistance. )


If I'm right, then the contacts bridge two points somewhere... probably grounding a resistor that pulls up a processor or reset chip input. I'd look for those points, solder on two little pieces of wire wrap wire, and place a parallel NO switch outside the computer. Make it a toggle switch, so it stays put. Voila, you can test your theory about the switch, you can actually measure the contact resistance, and you can make it do what you want with one set of hands. Because of the proprietary nature of this little switch and the low probability that it's at fault, I'd resist the urge to mess with it directly, and leave it in place. If you break it, you'll probably play hell finding another one.

The most important thing in this troubleshoot seems to be first eliminating the switch as a probable cause.


My money is on something else. Probably a supervisor chip.
posted by FauxScot at 6:05 AM on August 27, 2010


Response by poster: Hmm. Interesting hypothesis.

So then, do you think it makes more sense to just swap out the mobo for ~$40 than bother messing with isolating a component?
posted by TomMelee at 9:01 AM on August 27, 2010


If you can get a MB for $40, I think it would be the best solution.

Nonetheless, if you actually have a source for a $40 motherboard, then 1/2 hour doing this would be fun and potentially save you $40. Personally, I like to explain why I am replacing things versus just replacing them, but just to myself. Makes me feel better about the fix.

If this beast uses a supervisor chip (and it probably does), it should be close to the switch and fairly small. Usually, they provide 'Power supply good' signals and Reset signal in the proper polarity. (They usually have internal comparators and use external resistors to divide down the input voltage, and also usually have a cap for delay. A laptop has a premium on power consumption, so small dedicated circuits like this optimized for low power are likely to be found in them. Not sure where I'd point you to look, and I am risking over-generalizing on a box of which I admit ignorance.)

One thing I DO know, and that is that the switch pins go somewhere and can be found with an Ohmmeter and some persistence. If you can find these points, you are halfway to ruling out the switch as a cause. If it's NOT the cause, then it's motherboard time.

Feel free to memail. (A digipix of the area clear enough to see the chips and designations would helpful if you do.)

Good luck.
posted by FauxScot at 12:00 PM on August 27, 2010


Response by poster: I will probably do that. Expect memail this weekend, just what I know you wanted. :)
posted by TomMelee at 12:37 PM on August 27, 2010


« Older Visiting Toronto next weekend (labor day weekend)...   |   Mortgage Fun Newer »
This thread is closed to new comments.