Why does pulling the power cord then pressing the power button fix a non-booting PC?

Question

I've been working at this institution for about 6 years. One thing thing that I've always found curious is that sometimes—especially after a power outage—we find a PC that won't boot when the power button is pressed. Usually, the fans will spin up, but it won't POST. Our solution is to pull the power cord, press the power button with the computer unplugged, then plug it in and turn it on. It seems more common with Gateway brand PCs than the Dells or HPs that we have around.

Does anybody know what pressing the power button does when the computer is unplugged? I have some vague notion that closing the power button circuit allows some capacitors to discharge or something, but I'd like a firmer answer to offer my users when they ask me what I'm doing.

My best guess as to why fans can spin but it can't POST is that the BIOS is in some non-functional state. I don't know how BIOS stores state, but my best guess is that there is some residual garbage in its registers or something, like the stack pointer isn't starting at 0 maybe?

Answer 1

The computer is supposed to eliminate the residual static charge, but sometimes in such abnormal conditions it doesn't happen. Due to this the motherboard protects itself by turning itself off. This accounts for the symptoms you describe : fans spinning up but no POST, meaning that power is coming in but the motherboard does not function.

I suppose that if you left the computer alone for some time, the problem will clear up by itself, even via the LED lights. However, pressing the power button will instantly discharge any power stored in PSU and get rid of the static charge.

The following thread expresses this well :

On an old AT type power supply, your procedure doesn't mean diddley. Nobody (well ok, very few people) use AT power supplies any more. Now we use ATX power supplies, which have some smart circuitry in them which is always on, even when the computer is off. Your TV is the same way. Most of the computer (or TV, or stereo) is actually powered off, but there are small circuits in there which are active which are used to sense when you want the device turned on, and they turn it on. Your procedure insures that these little circuits are completely powered off, and goes to the extra trouble of discharging all of the capacitors in the power supply to make sure that they aren't powering the circuits inside your computer too. Capacitors store energy, kind of like a battery. They are used in computers to smooth out variations in the power supply and keep things stable. Even when you turn something off, the capacitors inside it are still charged, and may have enough energy in them to power the really low power circuits inside for quite some time.

Answer 2

I have noticed this problem occur more often with computers that are not plugged into surge protection (UPS battery backup units usually provide excellent surge protection too). Another common anomaly I've encountered a few times is that the system will boot up but certain peripherals like the network card won't function.

I suspect that what's happening is that the power surge is overcharging various components in your system, and that this prevents the system from starting up correctly. Un-plugging the system then pressing the power button causes all this excess charge to be discharged, and then when you plug back in and power up you get a clean start.

The amount of power stored (normally in capacitors; some ICs may contain capacitors too) varies depending on the design of each motherboard as well as the power supply (and any other components that are involved in this), so you'll probably notice varying results with different machines.

Answer 3

I have had theories and I don't know if any are based in any firm foundation though...

I recently helped a friend test a new mobo and CPU before he had his RAM and PSU. I simply unhooked my cables, and reached them to his hardware, powered it up, and all was well... When I went to reattach my cable to my hardware, I realized I had left the 6-pin auxiliary GPU power connector attached to my graphics card. When I tried to power up my system, the fans briefly spun, and it immediately powered off. Removing the GPU was the only way for the PC to boot. I thankfully had integrated graphics. On a whim, I let the card sit for some time, and tried again, and was able to get it to boot.

My theory, is that there are capacitors for the auxiliary and the primary (mother board PCIe slot supplied) power rails on the GPU. There are likely MOSFETS that are used int he power regulation circuitry for the GPU hardware, and I am wondering if the anomalous charge conditions (power int he aux side but not the PCIe side) in the power supply caused some manner of non standard voltage potential to be applied to the regulation circuits. If the Gate of a MOSFET were hit by a charge, when it normally would start up in a non charged state, it would inhibit flow between the Source and Drain of the MOSFET. Normally, regulation occurs by pulsing the gate to enable and inhibit flow of electrons, to carefully adjust the flow. If a stray charge were being held there, due to the anomalous charge state of the circuit, it would prevent any flow at all.

If the motherboard or PSU detect an error state at power up... Well, to protect the hardware, the safest thing to do is shut down.

That's my running theory.

Answer 4

I don't know why it won't turn on, hopefully someone can provide that particular answer. My best guess has been a failsafe in the motherboard to protect it.

You're spot on with the capacitors though. As a test, you could unplug the machine and let it sit there for awhile and eventually the capacitors will have discharged enough that the effect would be equivalent to pressing the power button after unplugging it.

In my case, there was enough charge left in the capacitor(s) to spin up fans, begin spinning up hard drives, and some other effects were usually noticed such as monitors flickering, keyboard LEDs turning on, and speakers receiving some amount of static (though I suppose that could be induced current instead of a signal coming from the sound card).

For reference, the computer I saw that issue with was custom built by me. It's probably a feature (or flaw?) in the motherboard. I tended to go with Asus branded motherboards back when I built it.

Answer 5

When you unplug a computer and press the power button, it discharges any power stored in the capacitors, like you say. I don't know why that would fix anything, though.