What is KILLING my PSUs?

[Gondi]

*tears hair out*

I'm typing on a PC that ever so often will pop up an error msg "The driver detected a controller error on \Device\Harddisk2\DR2. " I have 2 hard drives that are sitting unplugged in my tower. And a power supply that seems like it's on its way out.

This is #3.

First two were Antec TruPower 430W. After the 2nd one I figured I wouldn't mess around anymore and went with a Sparkle 550W with 36A on the 12V. Why 36? for the 6 HDDs, 2 optical drives, 9500, overclocked CPU and everything else.

This is starting out the same as the other ones.... The drives clicking and resetting, the PC freezing randomly. Each time it was solved instantly by a new PSU, but 6 months later it would start all over again.

I run the PC 24/7 and the drives are almost constantly being accessed, but is this really such a load for a quality 550W PSU to handle? Can there be something else that is causing my PSUs to die so readily?

Also it seems like I'll need a new PSU soon enough too.... I'll probably be going with PC Power and Cooling, but it would suck royally to drop $200 on a PSU that might be plauged by the same problems if the drain is coming from some place else. Has anyone had any similar issues?

 

[OldM3ta]

maybe your motherboard isn't handling the voltage probably...

but usually power flows from PSU to motherboard so think what might be flowing into the PSU... I would put a FAT UPS before your PSU or make sure to get one with PFC.

Or both... plus.. go through your house and make sure all your plugs are properly placed. Unpolarized connections elsewhere in the house can cause surges and spikes that will kill electrical components like PSUs.

Ideas at ya

 

[Ice Czar]

first thing that comes to mind (same as MeTaSpARKs )
Power Conditioning 101
with the additional caution of the Dangers of MOV surge protectors
which can often seem to be functioning while in fact they have failed
a UPS is in any event an investment well made

and till you rule out the AC source power as the cause, youll always be guessing
PSUs are all about the worse case senerio
a supply too hot (dropping its capacity) with too high a draw on it, when all of a sudden there is a varience in the AC source

A Technical PS: Output Voltage Regulation and AC Voltage @ silentPCreview by Mike Chin

Voltage regulation with most PSUs examined by SPCR has been virtually unnecessary. Perhaps because only reputable brand samples are submitted for our stringent testing, almost every PSU has passed voltage regulation checks easily, often with flying colors. The Seasonic Super series were among the best in this regard, and the new Rev.03 models are no different, with tolerance better than +/-2% in every case.

A test instrument mentioned in the article SPCR's Revised PSU Testing System was used briefly with the new Rev. 03 samples. The California Instruments 801RP Variable AC Power Supply (CI-801RP) allows AC voltage to be varied in order to consider the effects of brownouts and other real-world conditions on PSU performance.

As the voltage is reduced, the PSU has to draw more current (Amps) in order to maintain the output voltages. The actual power drawn by the PSU rises when the input voltage is reduced. This is why, for example, that a power supply rated for 120VAC, 5A might be rated for 240VAC, 3A. Maintaining steady output voltage lines becomes much harder with lower or varying AC voltage. This is how PSU voltage regulation can be really tested.

The PSU was placed in the thermal test rig, with 120VAC power delivered by the CI-801RP. Each PSU was warmed up for a few minutes at 120W, then set to deliver full power. Internal test box temps were ~37°C. AC measurements were made with the Kill-a-Watt AC power meter. Voltmeters were used to continuously monitor the 12V, 5V and 3.3V output lines. The VAC was then reduced from the standard 120VAC down to 80VAC. This is a very demanding test, as the Seasonic PSUs are rated for full power output with 100-240VAC.

<CHART>

Note that the output voltage remains constant through these dramatic drops in AC input voltage — and the associated increases in the current and power drawn by the PSU.

The same test was conducted at 300W output on two other SPCR-reviewed PSUs that will remain unnamed at this time: One rated at 350W sparked and shut down within 15 seconds of running at 90VAC. It appears to be dead. The latter, rated at 380W, auto-shutdown safely within seconds of 100VAC operation. It appears to consider operation at any AC voltage under ~103W to be unsafe.

It is a measure of the Super series' quality design. It means that sags and brownouts in AC power, experienced by many people in many places in the world, should have no effect on a computer powered by one of the Seasonic Super series PSUs. A table of the Super Silencer 400 could have been posted but it would have been redundant because the voltage accuracy and output remained constant down to 80VAC there as well.

We'll look at the VR / VAC relationship in more detail, and examine this aspect of performance in other PSUs in the near future.

a high quality PSU provides you with a larger operating range, but even the best supply will shutdown (hopefully protecting the mobo) or fry the PSU at some point, after all the PSU is but a link in the chain, the source power and the mobos Voltage regulation being the other ends

in short, the three potential contributing factors
AC source
temperature
load (overloading one of the rails) run the numbers to approximately know what safety margin you have

 

[larrymoencurly]

What are your voltages (taken with a digital meter) and temperatures?

Could some connections be corroded or loose? I've had HD power connectors where i had to squeeze the round metal tubes tighter.

One person with a fast system (maybe a dualie) and a bunch of HDs, including a couple of 10Ks, said that it never drew more than 400W from the AC outlet, which translates to roughly 300W PSU output.

What happens if you put back one of the PSUs that was in the system when it acted up? If the computer works OK for a while, I think that would indicate a connector problem because the act of plugging in the connectors wipes them off a little.

What wears out in PSUs in normal use, other than fans and capacitors?

 

[Gondi]

Thanks for the info guys, time to do some research...

BTW, love the locale, IC, Tom Waits is god.