Re: the gutless, bloated, and fried power supply hall of shame
in order to be a little more realistic I dug an old crappy motherboard (ECS L4VCA2 socket 478 board with a 2.5 ghz Celeron) I didn't care about (since the likelihood of this POS PSU blowing up and killing the board was decent) out of the shed, fired up memtest86 and added an additional 55W 12V load in the form of an automotive headlight to simulate drives and a load on the GPU (the CPU is 61W TDP).
The results are interesting...., while all rails are now in ATX spec, the 12V rail is on the low side and dancing all over the place (note: the 12V pics were taken with my Nikon D7000's 6fps burst mode to give you an idea of how rapidly it is fluctuating, it went from 11.784 to 11.888 to 11.783, to 11.778, 11.781 in the space of about a second, I don't own an oscilloscope but I imagine it would look pretty bad on one):
5V @5.1834V:
3.3V @ 3.3618V:
Its is drawing 167W to power what should be a 110-120W load, so efficiency is still in the 65-70% range:
Originally posted by momaka
View Post
I guess my guess of "65%" wasn't that far off. That's truly terrible, though.
A big part of that comes from the fact that this PSU uses a linear 3.3V rail that is derived from the 5V rail. So with your 12 Amp load on the 3.3V rail, the loses are (4.8383V - 3.3366V) * 12 Amps = 1.5V * 12 Amps = 18.02 Watts of loss just from the 5V-->3.3V rail linear regulator circuit.
So I guess we can give this PSU the benefit of the doubt that with a more realistic PC load (either 5V or 12V heavy instead of 3.3V-heavy), the efficiency should be somewhat better - probably in the low 70% range. That's still laughable, of course.
A big part of that comes from the fact that this PSU uses a linear 3.3V rail that is derived from the 5V rail. So with your 12 Amp load on the 3.3V rail, the loses are (4.8383V - 3.3366V) * 12 Amps = 1.5V * 12 Amps = 18.02 Watts of loss just from the 5V-->3.3V rail linear regulator circuit.
So I guess we can give this PSU the benefit of the doubt that with a more realistic PC load (either 5V or 12V heavy instead of 3.3V-heavy), the efficiency should be somewhat better - probably in the low 70% range. That's still laughable, of course.
The results are interesting...., while all rails are now in ATX spec, the 12V rail is on the low side and dancing all over the place (note: the 12V pics were taken with my Nikon D7000's 6fps burst mode to give you an idea of how rapidly it is fluctuating, it went from 11.784 to 11.888 to 11.783, to 11.778, 11.781 in the space of about a second, I don't own an oscilloscope but I imagine it would look pretty bad on one):
5V @5.1834V:
3.3V @ 3.3618V:
Its is drawing 167W to power what should be a 110-120W load, so efficiency is still in the 65-70% range:
Attached Files
)
Well, that was before I removed 6 Teapo caps from it. Sorry, I forgot to take pictures before that.
Comment