Disturbing discovery on Aopen PSU

Gandalf · Post by **Gandalf** » Sun Dec 28, 2003 12:46 pm

I have an energy-meter laying around and I was curious to see how much some of systems consumed. One of them is a Duron 950mhz with a simple 3gig hd, 64mb SDRAM and one fan (the CPU fan). It used an Aopen 300W PSU. Odd enough it consumed a whopping 90Watts while *IDLING* under OpenBSD.
I swapped the PSU with a "codegen" (never heard of) 300W psu and suddenly the meter indicated under 30Watts whilst doing the same idling.

How on earth is it possible for an aopen PSU to waste over 50watts of energy? Both units have a passive PFC (as pfc's are required by law in europe) and have pretty much the same specs ..
Any thoughts?

larrymoencurly · Post by **larrymoencurly** » Sun Dec 28, 2003 3:13 pm

Do you mean you have an energy meter that reads out in joules or kilowatt-hours? Does it just measure current and assume that watts is simply amps X average volts, or does it also measure volts continuously to correct for power factor?

The passive power factor correction coil in my 300W Fortron/Sparkle has 2.0 ohms of resistance

Codegen is one of the brands from Deer, like Allied, Foxconn, L&C, etc., and isn't considered very good at all.

Gandalf · Post by **Gandalf** » Sun Dec 28, 2003 3:33 pm

It reads out in KWh, as well as volts and amps seperately. The reading is pretty accurate imho as the device is certified by my power company (so I'm assuming the results are comparable to the ones used to base my bills on).

It's still quite odd and makes very little sense. Maybe I should try another AOpen PSU ... *goes off to find one*.

halcyon · Post by **halcyon** » Sun Dec 28, 2003 11:50 pm

Gandalf,

stupid question: are you sure (as in sure-sure) that the system conditions were exactly the same under both testing conditions?

i.e. no system intensive automatic cron processes starting up when you tried with the AOpen?

I'm not trying to dispute your finding and they are most likely correct, just checking

Gandalf · Post by **Gandalf** » Mon Dec 29, 2003 3:11 am

Haha. Yes yes, I am quite certain of it. It was a simple boot process followed by 10 minutes of inactivity. The difference was pretty much the same during the boot process.

And even if it was a PFC issue my line would have to be serious shit. I mean 60W difference ...... that'd mean I'd have to have surges of what, 350volts? My line never goes under 220 and never above 225v. So it's relatively stable.
Makes no sense

larrymoencurly · Post by **larrymoencurly** » Mon Dec 29, 2003 12:32 pm

If the measurements are accurate and the computer drew the same amount of power in each test, then assuming 70% efficiency for the Codegen, the Codegen dissipated 9W while the Aopen/Fortron dissipated 69W (23% efficiency). Did the Aopen/Fortron feel a lot hotter? Because when I loaded 300W PSUs to 380W, they got very hot on the outside, and I could even feel heat on top of the computer case.

Gandalf · Post by **Gandalf** » Mon Dec 29, 2003 2:55 pm

The air blowing out of the AOpen was definitely warmer than the Codegen yes.

larrymoencurly · Post by **larrymoencurly** » Tue Dec 30, 2003 1:41 pm

But that could just be from the Aopen's fan running slower, as is the case with my Antec SmartPower. OTOH a PSU dissipating 69W will make the top of the computer case really warm.

Inexplicable · Post by **Inexplicable** » Tue Dec 30, 2003 2:17 pm

Gandalf -- How do they behave under higher load? What do the current readings show?

What else... Did you try banging the meter and each PSU against the wall a few times?