ATX power supply issue

[GeorgeTsak]

Hello my name is GeorgeTsak and this is my first topin in EEVblog. Before some months I got an old ATX power supply from an old PC and it was working perfectly. However in the last 5 days I was pushing the power supply to it's limits.(I was pulling out 24A at 5V when it is rated at 26A 5V). Anyway the problem is that it won't turn it on, when I hit the switch It will provide power for milliseconds and it will turn off again.(The exact same behavior when sorting cables, e.g. the 5V cable with the COM cable.). I opened it up to see if there is any physical damage but I could not see anything wrong with the board. Accidentally I discovered that one of the two big heatsinks was live at 170VDC, which seemed strange to me. I did some research and I rode that many devices use live heatsinks in order to have common base in the transistors. However when looking at the transistors I can see 3 pins going to the board and this is my question: If indeed the transistors share the same base shouldn't be two pins going to the board?(the Emitter and the collector). So the reason my power supply is turning off is because there is a leakage? Thanks in advance!

If it matters here is the power supply brand/model: Enhance ATX-1125BTA

[Aodhan145]

If you can get a new one they are so cheap. Scrap that one take the heatsinks, etc.

[max666]

Yes, be careful. Heatsinks in SMPS can be live, not only by accident, but by design. And keep in mind that the main filter cap can store hundreds of volts for some time after you disconnected the supply, discharge it before you touch anything on the pcb.
In an ATX power supply and the majority of other SMPS's you will find MOSFET's as switching transistors not bipolar transistors and MOSFET's usually have source and body terminals connected together internally (sometimes drain and body):

Although the MOSFET is a four-terminal device with source (S), gate (G), drain (D), and body (B) terminals,[1] the body (or substrate) of the MOSFET is often connected to the source terminal, making it a three-terminal device like other field-effect transistors. Because these two terminals are normally connected to each other (short-circuited) internally, only three terminals appear in electrical diagrams.

Now why would you ever need that third pin then? Well, often you can't have a live heatsink, maybe for safety reasons, maybe you have other devices mounted on the same heatsink that have a different potential on their tab. In that case you have to isolate the tab from the heatsink, by using a sil pad for example, or you can buy transistors with isolated tab, but now you obviously need a third pin to connect to the source.

And for why your power supply isn't working any more, I'm no expert, but your power supply may have several MOSFET's paralleled up to increase the current handling capability and one of those transistors might have died when you pushed it, because cheap ass ATX supply probably don't implement paralleling very well. Or your electrolytic cap's have gone bad. At least those two things come to my mind.

[amyk]

It is probably a standard half-bridge TL494 type PSU, see schematics here:

http://danyk.cz/s_atx_en.html

The heatsinks are live because it reduces the need for insulation and you're not supposed to touch them anyway.

[wraper]

Those supplies aren't really meant to deliver rated power for extended periods of time. I'm sure if you can find the data sheet it will specify intermittent and countinous ratings both will likely be overly optimistic. They put peak power handling on the label so it looks impressive. If you actually want a supply to deliver countinous power buy a purpose built one which clearly states its ratings or try another PC SMPS of higher quality rated for a higher power then you plan on using it.

This one actually should be quality stuff holding the rated load because this is good manufacturer. Most likely dual diode on the 5V rail became short. Actually measure resistance from 5V rail to GND, if it is less than 10 ohm, the fault should be there.

[GeorgeTsak]

Those supplies aren't really meant to deliver rated power for extended periods of time. I'm sure if you can find the data sheet it will specify intermittent and countinous ratings both will likely be overly optimistic. They put peak power handling on the label so it looks impressive. If you actually want a supply to deliver countinous power buy a purpose built one which clearly states its ratings or try another PC SMPS of higher quality rated for a higher power then you plan on using it.

This one actually should be quality stuff holding the rated load because this is good manufacturer. Most likely dual diode on the 5V rail became short. Actually measure resistance from 5V rail to GND, if it is less than 10 ohm, the fault should be there.

I measured the resistance and I got around 1?. So it seems I burned the diodes  . Thanks everyone for their response!!!