In a computer of a friend, an EAGLE 400W power supply was used.

Claims:
- Maximum continuous output 400W
- 12V: up to 17A
- 5V: up to 40A
- 3.3V: up to 28A
- Total on positive rails 380W
- Total on negative rails 11.1W
- Total on standby 10W
- 401.1W (400W max)

Okay, so the numbers add up but can it actually manage 400W?

No way. Said friend was using it in a gaming rig, where a minimum 350W is advised by the graphics card manufacturer. They reported their games would often freeze and the computer often rebooted itself. So he tried everything, reinstalled the OS, changed the hard disk, even replaced the RAM, and it still gave problems. Then, one day he was showing me a game, it froze (like normal) and then suddenly BOOOOM the power supply exploded in a shower of sparks and smoke and the computer turned off. It scared the @#*! out of me! The computer was okay after a new PSU was fitted, thankfully.

So he gives this piece of now fried PSU to me and I find that this power supply was just waiting to fail. On the main board of the power supply it says (for the fuse): 5A for 180W-235W, 6.3A 250W-350W. A 5A fuse was fitted. So this PSU would barely manage 235W, let alone 400W. Luckily the fuse did its job, it blew. (Plug fuse did *not* go.)

What actually went wrong? It seems to be a case of catastrophic failure. I know what caused the complete failure of the PSU. It was the power MOSFETs going pop. They basically became 20 ohm shorts (as tested with my multimeter.) Dumping ~400V from charged caps into 20 ohms is bound to incinerate anything. Both of them simultaneously went bang (or possibly the failure of the first caused the failure of the second) and this is probably what caused the sparks and explosion. But the MOSFETs were not alone in the parts that failed. Several resistors and a capacitor were also popped, but these were all on the secondary side, well away from the fireball from the MOSFETs which charred other parts. It seems to me like something in the feedback loop went crazy and caused the transistors to pop whilst also causing damage to the secondary side components. Also, the output filter caps are bulging; possibly the reason it was often crashing the computer.

It's very much apparent that this power supply ran hot, because the PCB is discoloured in many places around high power components. Also, the Schottky diodes on the output are underrated. 40A for the 5V (just in spec), 20A for the 3.3V (out of spec, requires a minimum of 28A), and 16A for the 12V rail (which is rated to 17A.) And being "5V heavy" (more current on the 5V rail than 12V) means modern graphics cards won't work too well on it...

So does anyone know what would cause this? Seems that these cheap power supplies are very dangerous. This could have easily started a fire.

Tom

 

The bulging caps seem to be a likely culprit, but they should've caused the fuse to blow rather than a meltdown.

 

Its called cheap parts. Minimal air flow because no one cleans them out periodically when the dust builds up.

They are not worth repairing since they are so cheap and the time required for repair isn't feasible. Most of them are a so packed together as a technician I don't want to take them apart.

I do collect them for spare parts of other equipment I repair. They can be fixed but I have a dollar limit on what I repair. If its a fast fix and easy I will do them, but I have a limit of 100.00, if its under order a new one, and when I have time I will look at it and see if I desire to work on it. Over 100.00 yes I work on it if its repairable in a hour or less. 300.00 I repair.

I repair one cpu, 3-5k approximately. Power supply 3-500 each. I don't have time to work on 25 dollar parts when I have much more expensive stuff to fix.

 

Its called cheap parts. Minimal air flow because no one cleans them out periodically when the dust builds up.

They are not worth repairing since they are so cheap and the time required for repair isn't feasible. Most of them are a so packed together as a technician I don't want to take them apart.

I do collect them for spare parts of other equipment I repair. They can be fixed but I have a dollar limit on what I repair. If its a fast fix nd easy I will do them, but I have a limit of 100.00, if its under order a new one, and when I have time I will look at it and see if I desire to work on it. Over 100.00 yes I work on it if its repairable in a hour or less. 300.00 I repair.

I repair one cpu, 3-5k approximately. Power supply 3-500 each. I don't have time to work on 25 dollar parts when I have much more expensive stuff to fix.

I agree; I wouldn't bother repairing this given the failure mode is quite spectacular.

I have a Thermaltake 480W power supply which is okay (it seems to meet its ratings), but it runs HOT! I know why - it's just packed FULL of stuff. There is no more space inside it to fit anything else. There is so little space, there are extra little boards added in places which provide things like active PFC, or a fan controller... The problem is, the ATX case style for power supplies is only really good for maybe 400W maximum units. It needs to be made bigger to allow for more filtering components, and more efficient architectures.

 

How does the 5V stand-by look like? It could be that that this voltage rose too high and killed the pwm IC in the process, which the lead to total destruction.

 

How does the 5V stand-by look like? It could be that that this voltage rose too high and killed the pwm IC in the process, which the lead to total destruction.

I think that the resistors and caps that are burnt are in the area of the 5Vsb, so it's plausible. Dunno why the 5V would fail though.

 

Basically cheap counterfeit even for a part that is known for being cheap. Labels lie.

 

I've done a bit of diagnosing and I think I have figured out what caused it to explode.

First, the freezing: He was playing a game and it just froze about 5 seconds before the bang. So something caused it to freeze; possibly due to increased ripple or a voltage sag on any of the rails. I believe the 5Vsb circuit began to malfunction in some way, as many of the burnt components are in the 5Vsb area. The malfunctioning might have caused the rails to go out of spec. However, I have no idea why it malfunctioned here, possibly a bulging cap or maybe a broken solder joint from heat (some of the soldering on this unit is not good.)

Anyway, a small 1/4W resistor was toasted and this drives the feedback transformer. It seems that the feedback is inverted in some way although I have not figured out entirely how it works so I am guessing a bit here. Maybe when that resistor went pop it made both MOSFETs switch on at the same time which is what destroyed them, as well as blowing the fuse. Or perhaps one switched on while the other was switched on: if done briefly, this isn't a problem (apart from reduced efficiency), but if done many times a second it can cause heating and failure of the FETs. Or perhaps it was even a frequency issue, causing dV/dt transients which killed the FETs.

That is as far as I have got... Any other ideas?

 

Use it for a boat anchor and be done with it.

 

Use it for a boat anchor and be done with it.

I am not planning on repairing it, it's not worth it.

Just curious as to why it blew up. I like understanding failure modes...

 

I've done my best learning that way. Can't complain about you using the same methods I use.