PSU Myths?

[jubbie]

Just saw this article at Antec. Truth or marketing sham? Just sparking off discussions.

http://www.antec.com/Believe_it/PSU/index.php

 

[Dangman]

Myth 3 has generated a good discussion over at JonnyGURU:
http://www.jonnyguru.com/forums/showthread.php?t=6625

 

[TIGR]

Multiple 12v rails/OCPs are undeniably safer, at least to some degree; I found it amusing to read in the thread at jonnyGuru about the Antec PSUs that don't have OCP. The way the Antec rep is handling himself over there does not impress me.

At any rate, I still generally prefer PSUs with fewer rails. Nice marketing though, Antec.

 

[25E5V]

Interesting.

 

[JoeBob]

Would be kind of ironic after all these years of PCP&C harping on the "multiple rail PSUs are bad, mmkay!" if people suddenly started downright slamming single rails because of that. =D

 

[HOOfan_1]

Just saw this article at Antec. Truth or marketing sham? Just sparking off discussions.

http://www.antec.com/Believe_it/PSU/index.php

There was an Antec Rep at another forum insisting that multiple 12V rails provided more stable power. Hopefully he/she was just a marketing rep who really didn't know what he/she was talking about.

I just find it funny that certain people used to be huge PC P&C homers and spread Doug Dodson's "single rail is the only way to go" FUD on the internet....now some of those people are spreading Antec's "use multi-rail or else you are in big danger" FUD.

 

[Zero82z]

I just find it funny that certain people used to be huge PC P&C homers and spread Doug Dodson's "single rail is the only way to go" FUD on the internet....now some of those people are spreading Antec's "use multi-rail or else you are in big danger" FUD.

As always, the real truth is somewhere in between...

 

[Siorus]

There was an Antec Rep at another forum insisting that multiple 12V rails provided more stable power. Hopefully he/she was just a marketing rep who really didn't know what he/she was talking about.

I just find it funny that certain people used to be huge PC P&C homers and spread Doug Dodson's "single rail is the only way to go" FUD on the internet....now some of those people are spreading Antec's "use multi-rail or else you are in big danger" FUD.

If you're talking about whatshisface in the thread @ JonnyGuru, from what I gathered from that forum thread, he used to write Anandtech's power supply reviews.

Personally, I think the main point of Antec's article is mostly BS.

There is, however, one big advantage of multi-rail versus single -rail power supplies: Safety. Modern multi-rail power supplies, like the ones that Antec makes, are engineered so that there is no way that you can possibly overload a +12V rail (provided it has the right maximum output wattage). Antec power supplies have separate +12V rails, which means that each rail is limited by OCP. This provides high power that still comes with the assurance of having secure rails that won’t destroy PC components.

You can't overload a single rail power supply's +12v rail, either, unless you're a complete dumbass and try to run a pair of GTX480s and an overclocked i7 off of a 650w PS or something.

The guy in the JG thread that claimed to be from Enermax made an interesting point about OCP taking longer to kick in at higher amperages, thus increasing the risk of smoking components before it kills the party, but I'm afraid I just don't see how, in the real world, if you're not being an idiot with your hardware selections, it makes any difference.

Almost all of this single vs. multiple rail thing is, like HOOfan said, FUD and marketing BS. Personally I prefer having a single +12v rail, but single vs. multiple rails in and of itself is hardly a good reason to make a purchase decision one way or the other.

 

[Tsumi]

It really comes down to how well the unit is built. Multiple rails don't matter too much, except you have to make sure you don't exceed the rail limits.

Well built units will have thicker wires that can handle higher loads from a single output. That eliminates the need for OCP for each rail, and would only need a single OCP for the main 12v rail (99.9% power supplies 1000 watts and below really only have 1 12v rail, the rail ratings are really how much amperage the wires are designed to carry from the 12v rail to the output).

 

[TIGR]

Well built units will have thicker wires that can handle higher loads from a single output. That eliminates the need for OCP for each rail, and would only need a single OCP for the main 12v rail (99.9% power supplies 1000 watts and below really only have 1 12v rail, the rail ratings are really how much amperage the wires are designed to carry from the 12v rail to the output).

OCP may not be needed to protect thicker PSU wires, but what about the components attached to them (referring to both temperature stressing the PSU caps on one end and dangers to the loads on the other)?

 

[keenan]

You can't overload a single rail power supply's +12v rail, either, unless you're a complete dumbass and try to run a pair of GTX480s and an overclocked i7 off of a 650w PS or something.

Well, obviously you're not going to overload it intentionally, the point of OCP is to protect thing in case of failure. Their argument does have some merit. Today's high-end power supplies are capable of providing 50A (or more) on the +12V rails. Their wording is a bit sensationalist, but the fact remains that this is an unsafe amount of current to be flowing in a fault condition.

Say a hard drive fails in such a way that +12V is shorted, with a single-rail PSU, the available current will be flowing through a single dinky 18AWG wirel. The inherent resistance might be enough to limit the current to under the (huge) overcurrent limit, allowing a fault current of tens of amps to flow without tripping any protection. In a wire that can only safely carry maybe 15A. If this happens, you'll get melted insulation and probably other damage, and it'd be easy for such a fault to start a fire.

Which topology is 'better' depends on a lot of factors, but a multi-rail PSU with lower limits on each rail should be safer in a fault situation.

 

[Siorus]

Which topology is 'better' depends on a lot of factors, but a multi-rail PSU with lower limits on each rail should be safer in a fault situation.

Ok, that's a fair point.

However.

Realistically speaking, what are the chances of an actual fire starting in a computer case? There's not really a lot of fuel there. I mean at one point I took a butane fueled plumber's torch to the backside of a Soyo KT266 board because I wanted the NB as a keychain. Even with the flame touching the board, it didn't burst into flame. It bubbled and smoked, sure, but it absolutely refused to catch fire. Metal components obviously aren't going to burn in this scenario, which doesn't leave you with a lot left... I suppose you could get the plastic on a fan or a card slot or the wire insulation burning, but that's about it and I find that unlikely-and even if you do, like I said, there's not much fuel there. Maybe you could get a metal portion of the case hot enough to set fire to something near the PC, but you'd think that someone would notice the smell before that happened, no?

So I mean technically speaking yes, a multiple 12v rail system may be safer in terms of fire risk... but I just don't see that it's a realistic concern.

 

[keenan]

Well, PCBs are made from fiberglass, so yeah, they're not going to burn. And most of the plastic is probably treated with fire-retardants so that a flame won't sustain itself there either. Still, such a situation will make a mess and I'm sure release a lot of toxic crap. Mostly the safety risk, as you point out, is all the heat being dumped where it's not supposed to go. As you say, it's probably not a serious concern, and their wording is definitely sensationalist, but if you're going to talk safety, they do have a point, it's just a more minor one than they'd like you to believe.

And really most PSUs are guilty of this flaw anyway, it's just that these giant single-rail ones are worse. Even inexpensive low-end PSUs can easily overload an 18AWG cable; really each output lead should have its own overcurrent protection, but nobody does that.