No matter how many times I see someone changing computer parts while the power is still connected, if I'm doing it I always unplug the power then pressing the power button to ensure it's unpowered
You won't get hurt working on a live computer unless your power supply is dangerously engineered (stick to well known named brands) or wide open (don't open it up unless it's been unplugged and sitting on a shelf for months) for the same reasons pointed out elsewhere in these comments; the voltage simply isn't high enough. It won't even induce a perceptible amount of current (unless you do something stupid like stick your tongue on it), much less a harmful amount.
You'd be surprised at all the stuff you can get away with on a live computer. For example, did you know that AHCI (SATA) and PCIe (and thus NVMe) are both hot-pluggable and hot-swappable? ... and that you can hotswap RAM (with a cooperating motherboard and OS) and even CPUs (on multi-socket boards)?
Sata is hotswappable. IDE was not. And I believe m.2 would be if the connector was designed a little different, because that’s all u.2 is. Compact Flash Express is, but again, that’s NVMe and splitting hairs between u.2.
Typically on hot swappable devices it’s been common to make the ground pins a little longer than the rest, ensuring that those will make contact first. The issue with hot swappable devices aside from a sharp initial draw is “grounding” the circuit to an IO pin before real ground is made. Most default low potential IO pins won’t enjoy taking the full load of the device while the ground plane is missing connection.
Oddly to me, many new PC BIOS ship with SATA hot swapping disabled. IDK why. You can plug the drive in it’ll spin up, but no data transfer will happen until you reboot. Safe, just ineffective. I recently had to learn this the hard way.
I think this is something the OS can override. If I remember correctly I could discover a hotplugged SATA device in Windows by using "Scan for new hardware" in the Device Manager.
More specifically, the system's firmware AHCI hotplugging configuration only applies to the system firmware.
That is to say, if hotplugging is disabled on your SATA controller, then the firmware won't see a new SATA drive if you plug it in while the machine is running. So, for example, you won't get a UEFI boot menu entry for it if you go to the boot override section that presents all of your storage devices.
Once the operating system takes over, that's all a moot point. Whether hot-plugging was disabled in your firmware settings or not, if the OS supports it, it will work. Both Microsoft's AHCI driver (msahci.sys) and the Linux "ahci" kernel module (or built-in) support hot-plugging.
On my Windows desktop with an Intel SATA chipset, hot-plugging a drive did cause Windows to briefly drop all other drives on that controller before re-enumerating all of them, which lead to a BSoD because one of those drives was the drive with Windows itself on it. However, this was solved by installing Intel's chipset-specific driver (Intel Rapid Storage Technology, iaStorV.sys) and tweaking the registry to have that one start at early boot instead of msahci.
The idea is also that pressing the power button will also help discharge some capacitors as the power-supply tries to supply power. Not sure to which degree this works to make things safer, but trying it won't hurt.
(Note this will _not_ make a powersupply safe to open up)
On a dumb device where the power button is part of the actual circuit, let’s say an old radio this will work.
On a PC, the power button isn’t hooked to anything but the chipset or some micro or an upstream circuit that tries to start a micro. Which means to have any effect, that processor needs to be executing code, or needs to have enough power to start the chipset itself. Both of these are unlikely to have any effect on CPU capacitors or caps in the power supply.
I’m fairly certain this is placebo at best on a modern PC.
It definitely kicks the power light on for a bit under a second on my 2010 era desktop. The (high end, low wattage) power supply is delivering enough voltage to keep the motherboard power controller up.
I have no idea if it discharges all the different voltage rails. I don't like having the fans try to spin up while I'm mucking about, and it stops that from happening.
Of course, from an equipment damage perspective, grounding yourself to the case ground is much more important than discharing the PSU capacitors.
I usually physically unplug the computer, push the power button, then open / touch the case (I'm in a humid climate).
I've never had an issue (ignorging fires, but those happened when the computers were intentionally energized).
You need about 24 volts to overcome the skins resistance. Most PCs use some combination of 3, 5, and 12 volt rails so you should be fine, as long as you avoid licking anything.
Edit: Some PSUs have -12 +12 rails, I suppose if you were really unlucky and grabbed both at once the 24v differential could give you a zap.
The -12v rails are typically low current though. The real danger is a poorly built supply or an electrolytic capacitor exploding in your face (had this happen once).
Incidentally, does anyone know what the -12v rails are used for? Some kind of reference?
RS-232 drivers like the old MC1488 were run off plus and minus 12 volt supplies. DRAMS like the MK4116 also required a low current -5V bias supply. And there were other chips that possibly needed a -5V supply. Since the -5V was low current you could generate it off the -12V with a simple linear regulator.
You can wet your fingers and get way lower, but I’ve never had anything much at 24V even when trying. At that low a voltage it may come down to frequency of the source.
Obviously your tongue gets way lower, as is 9V lickers know already.
