Power Supplies

[casualsax3]

Hey all,

Last Friday I had to test some power supplies to see if it's worth spending $70 on a power supply vs the crappy stock PSU that comes with a lot of cases you can find on NewEgg.

I tested 4 machines with 5 power supplies in 0, 1, 2, 3, and 4 drive configurations. I also took a reading of how much power the systems drew when I powered them on at 4 drives, which shows how efficient the power supplies become under serious load (it takes a good chunk of power to spin up 4 drives)

The machines were all tested with the same 1x1GB PC5300 RAM, and the same four Western Digital SATA drives. The Intel systems were LGA775 chips on an Asus, and the AMD's were AM2 - also using an Asus motherboard.

Here are the results.

Power supplies tested:
Silverstone
Antec
Seasonic
Winsis

In a lot of cases the stock power supply uses almost twice as much power.

In New York City I pay $.19c/kwh, so 1 amp of power can cost around $20 a month - ((volts * amps) / 1000 ) * time (in hours).
This means pretty plainly, that the stock PSU here would cost me another $15 per month on my one desktop that I always have on.

Now if my office switches all of our workstations to one of the three 80% efficient power supplies, we stand to save a few hundred per month. Add to that the fact that these power supplies are going to be less susceptible to surges and spikes and last a lot longer, and it's really a no brainier.

[funkyass]

Add to that the fact that these power supplies are going to be less susceptible to surges and spikes and last a lot longer, and it's really a no brainier.

Are you investing in some UPSs, because all Power Supplies are equally susceptible to surges and drops without them.

[Oskar_Hanberg]

Add to that the fact that these power supplies are going to be less susceptible to surges and spikes and last a lot longer, and it's really a no brainier.

Are you investing in some UPSs, because all Power Supplies are equally susceptible to surges and drops without them.

True, what makes the difference between a good and crappy Power supply ()beside the efficiency) is the fact that good power supplies deliver more stable voltages to the components in the pc, thus prolonging their lifetime. I have a Seasonic 430W and I'm very happy with it.

[casualsax3]

Add to that the fact that these power supplies are going to be less susceptible to surges and spikes and last a lot longer, and it's really a no brainier.

Are you investing in some UPSs, because all Power Supplies are equally susceptible to surges and drops without them.

My explanation was off there, I meant that the rails are not going to fluctuate as much.

[Aaron]

Very nice comparisons. Do you do this for a living?

[snkcube]

Interesting stuff. Seems like Antec delivers some good results.

[Nightcrawler]

Antec makes some very efficient supplies. Last year, I bought a 430W Antec in their NeoHE high efficiency line.

http://www.newegg.com/Product/Product.a ... 6817103939

It's been great. It outperforms the economy branded 500W supply that came with the case and it's rated at 70W less. Efficiency is really something to consider when dealing with power supplies these days.

[Deathlike2]

I like Enermax better. However, the Antec I got most recently is pretty damn good.

[casualsax3]

Very nice comparisons. Do you do this for a living?

More or less - I'm a sysadmin.

[Nightcrawler]

I should mention one other thing.

The efficiency rating is also directly related to how much heat your Power Supply gives off. All of the wasted un-transferred power is given off as heat. Electronics always love to run cooler. It extends their life.

Now, as far as how many actual degrees 85% efficiency versus 75% actually makes, I don't have the knowledge to say. It may not even be very significant, however it's still one piece in the puzzle.

[kevman]

It's not hard to do the math. If the PSU is 500Watt and is using 500w (which would require a Pentium D or multi-CPU sytem and at least 5 HDs to sustain that kind of draw), the additional 10% power consumption would lead to another 50w of heat. Contrast that to the 50-60W a Core 2 CPU uses.

Quite a bit of heat. Remember, though, that a PSU that runs constantly at anywhere near its limit is going to fail, becuase that means that the peak load will be way beyond its capacity.

[whicker]

I'm curious to know what device you used to measure, an ammeter or something like kill-a-watt?

Just because the engineer in me is kicking in, a lower amperage for AC current doesn't always mean less power used, because rarely are loads simply resistive. There is another concept called reactive power (essentially power shifted back and forth between producer and consumer).

Not that there's anything wrong with your measurements as the basic design of all the psu's are probably the same.

[casualsax3]

Hey Whichker, I used a Kill a Watt type device, as well as readings from an APC 7932.

[Joe Camacho]

Now that we are on the topic of power supplies: I need replace the power supply for my computer (It refuses to turn on after a series of random forced reboots and hearing "clicking" noise coming from said PSU) but it's been years since I bought this computer and I can't find the receipt with the description of the parts so..

How can I identify what kind of PSU I need to buy? I know it's an ATX kind of power supply, but after browsing computer part sites for a while, I can't identify the diferences between them.

So?

[kevman]

Then buy an ATX.

Unless your system is really strange (like a Dell or a server or something) that's all you need to know.

To be sure, look at how many different cables plug into the motherboard. There should be a big rectangular one, and (maybe) a smaller 4-pin square one with two yellow and two black wires. That's ATX12v, which all PSUs you buy new are. If you don't have the square one, don't worry, you can just not use it in the new PSU. ATX12v supplies work as ATX supplies.

If there's a third rectangular one with only one row of pins, that's the AUX, and while almost never used, still ships with most PSUs, so you don't have to worry about that either.

I doubt you have any Sata drives, so buy one with more Molexes than Sata power connectors.

If its a Dell, or in some cases, HP, you might be fucked, or you might not be fucked. Depends. Dell sometimes uses BTX, and HP and Dell both uses whatever the hell they feel like on a lot of machines. The case might be shaped strange.