I once came across a dodgy AT PSU where only the live was wired to the power switch, while the neutral was connected directly from the socket to the PCB. So in that case pressing the button only connected the live wire, which allowed the PSU to start.

There's also the case of some AT power supplies (usually from OEMs) which provide a +5VSB line and pressing the power switch just connects two wires (similar to PS_ON and ground in ATX) to start the PSU. In that case there are no dangerous voltages near the switch.

Miphee wrote on 2020-08-26, 06:13:

Mephusto wrote on 2020-08-26, 05:52:

Yeah I will. I usually dont mess around anything that is still connected to an outlet or with large capacitors. Electricity is very respected by me.....I'm scared of it.

You can also bypass the PSU rocker switch with a single switch adapter, that way you don't have to touch the PSU at all. Just switch it on and use the adapter's switch to turn it on/off.

How is that different from using the wall switch?

I strongly advice everyone that are new to AT PSU's, to read up on safety warnings and how they work. It gives solid knowledge on how they are actually working, and thus eliminating potential risk of shock. 110 volt are no fun, imagine 220 to 240 volts.... Ouch. 240volt shock is absolutely no fun (Speaking out of at least 4 times personal experiences to this date)

Vaudane wrote:

What sort of loads would be required on each rail for a decent power tester?

My guess would be switchable banks of resistors, to be able to load the main rails up to their rated current. And maybe some fixed 0.3A loads for the minor rails.
Resistors can be overloaded quite a lot if you provide forced airflow, and for that purpose you can use computer fans supplied from the very PSU being tested.

Oetker wrote:

I've noticed that sometimes AT PSU power buttons have four wires, sometimes two. Isn't a two-wire switch rather unsafe as, depending on how the user's outlet might be wired, it could only be switching neutral?

Haven't seen one with two wires, but I wouldn't rule out their existence. Most newer ATX power supplies have a single pole switch on their backs, and in countries that use non-polarized outlets this means in about 50% of cases you're switching the neutral (yeah, I know, nobody switches the ATX power from the back, but still). Shouldn't be dangerous, unless you start poking the insides while plugged, and that's a bad idea regardless if the switch is single pole or 2-pole.

Vaudane wrote on 2020-08-26, 15:17:

quicknick wrote on 2020-08-24, 01:15:

You'll need to provide a minimum load, else it won't start or the voltages will be all over the place. I keep a couple of dead HDDs (that are still spinning) for this purpose. The voltages might still be a bit off with such a light load, but at least you'll know if it's safe to use it to power a full system before a final voltage check.

This is something I have toyed with for actually building an AT power tester. 50-100W resistors are cheap. What sort of loads would be required on each rail for a decent power tester? Do you just gun it at max power output as this should be specified on the PSU somewhere? Do you use high resistances to measure the voltage accurately, but that gives low load? Obviously overloading with low resistances is a bad idea as once the magic smoke gets out, you can't put it back in...

You can build an electrolyte load bank using a pair of electrodes, a pail of water, and an acid or other electrolyte. Set the lower bound for impedance by placing the electrodes at full depth and adding electrolyte until the desired impedance is achieved. Reduce the depth of the electrodes to increase your resistance.

Oj0 wrote on 2020-08-26, 21:40:

How is that different from using the wall switch?

You don't have to unplug it every time you want to turn the PSU off.