"Multiple 12V Rails" PSUs <--

[flexy]

First i apologize since this issue might be known to anyone into PSUs already.

But i found some very in-depth analyses of recent high-end PSUs on http://www.hardwaresecrets.com where they actually look at the components and PSU layout. I just took a quick glance at so called "multiple 12V rails" reading their tests of the OCZ Powerstream, Corsair, as well as my current Thermaltake Toughpower.

First:

All/Most current high-end PSUs actually do NOT have real independent 12V rails - regardles of what they state.
Eg claims (example OCZ) "independent, separate 12V rails" are 100% so not true!

In fact, ALL of those supplies do *in fact* "only" have ONE 12V rail - (check the circuit-photos on that website).

The separate 12V rails, be it dual, three or four, are all "virtual".....a "separation" of those rails in reality is only in the fact that each of the "rails" (==wires) has a limiter, limiting the power draw per "rail", as required by ATX 2.2 specs, to 18A.

Again: All those PSUs have ONE 12V rail - and then 18A limiting circuits for each of the wires or "virtual rails" preventing each wire to draw more than 18A.

SO..is this a bad thing?

NO! I actually dont think it matters, there is the known user discussion like "pro and against Multi-Rail PSUs"...some say its better to have ONE strong rails, others swear by "separate" rails.

But why i am writing this?

a) because it surprised me, assuming that advertising like "separate 12V rail" would imply a true independent power source, eg. for pciex, CPU, motherboard etc. And it wouldnt matter how much each of the sperate powered componenets would draw - it also would not influence the components connected to the "other" separate rails. Eg. you could overclock and overvolt component #1 - but this not influencing current and wattage on your component #2, say, your CPU. (Which is the usual assumption)

But this is just a fallacy and merely a "PR stunt" to confirm to ATX specs, or, just plainly misleading and making the customer believe something which is not the case. ("Separate and independent 12V rails ---> total nonsense!! This applies for as good as any current high-end and reputable company, be it OCZ, TT o whoever)

*) There is a benefit to the fact that all those PSUs in reality only have one 12V rail - and it is actually that you dont need o bother to "waste" wattage on a dedicated line

Say..if your assumption is, you have a PSU with four rails each 18A (which it would also state in the manual as well as on the package and ad!)...you might think if you draw 2Amps from this line...you wasted 16A since rail #4 is "independent" and therefore this wattage is not available for the rest of the system.
But such thinking doesnt need to apply. It comes all out from ONE rail.

Georg.

 

[flexy]

what then actually is the "disadvantage" of "separate rail PSUs" (which actually are not what they claim, see post #1)?

Some hardcore overclocker state that PSUs with *one* strong, say 40A 12V rails have benefits.

The reason is clear once you do a simple calculation:

Let say, your separate rails are limited to 18A. You can use a calculator and simply 12V x 18A == 216W

Which limites the draw per "rail" to 216W - otherwise PSU would shut off.

Now..it is actually possible to hit 216W+ per rail, through extreme overclocking or similar.

You can check http://www.extreme.outervision.../psucalculatorlite.jsp this very good PSU wattage calculator, and then you can see that overclocking a the CPU alone, Kentsfield Q6600 to 3800Mhz and 1.5V vcore would result in 217Watts - thus already exceeding what the PSU could handle on the wire/rail going to the CPU. (This said, this is an EXTREME overclock

So..knowing about this limiting-circuits, hardcore-overclockers (and i am REALLY talking hardcore) might prefer "one extremely strong 12V rail" and go for the biggest beast with the most amps on this one rail "for headroom"...

For pushing really extreme limits (like in my 3800mhz 1.5V example) look at your PSU also - better, look at it first.
(Sidenote: For such extreme overclocks in conjunction with unhealthy voltages you would have to look at other things also, mainly cooling.)

Edit: Some nonsense removed. the 18A are not dictated by ATX specs, its the separate "rails". (Be i pseudo virtual, or whatever). There are in fact PSUs with "multiple rails" which have more than 18A on one rail.

This simple calculation above also shows, that, even with a 750W PSU, it would literally be impossible to achieve ultra-mega overclocks. So..if you are one of the LN people trying to hit overclocks 3.8+ with monstrous vcores (1.5V+) and beyond...you know what to look for to get you even started.

 

[robertk2012]

What is the point of this?

a) Multiple rails can be better as they increase stability. One strong rail isnt that bad either it really just depends on the situation.

b) If you had taken the time to read some of the posts about PSU you would see that some of brightest people here believe the "trapped" power theory is rubbish. If you add up the max draw from each rail you end up with a number that is much larger than what the PSU can actually output.

 

[flexy]

robert,

you missed the point of my posting. To answer your question where the point is.

The point is there IS no "increased stability" since literally ALL current so called "multi rail" PSUs use ONE rail. The "multi rail" is virtual and its only "multi" since each wire has a limiter, limiting draw per lead/wire to a certain amperage.

Look at the site above, look at the circuits and their anylyses. Be it OCZ Powerstream, Corsair, Thermaltake... all go in ONE common 12V output...so the "multiple rail stability" thing is a very well established MYTH....CPU, Graphics Card, Mobo, Sata, all get their 12V from ONE point in the PSU.

There are NO dedicated, separate circuits in the PSUs delivering 3 or 4 independent 12V rails of 12V.

This applies to 99% of all current so called "multiple rail" PSUs.

 

[jonnyGURU]

Originally posted by: flexy
robert,

you missed the point of my posting. To answer your question where the point is.

He didn't miss the point of the post. He was asking your point at posting it. It looks like you're looking for a sticky, but I only stick correct information....

Let me point out some inconsistancies in your post:

• It is true that most multiple +12V rail PSU's are not TRULY multiple +12V rail power supplies. Typically, you have a single +12V source that's split up into multiple +12V outputs. These "splits" are typically, but not always, capped off by current limiters called "OCPs" (Over Current Protection).

Why do they do this??

• Short circuit protection only works if there's minimal to no resistance in the short (like two wires touching or a hot lead touching a ground like the chassis wall, etc.) If the short occurs on a PCB, in a motor, etc. the resistance in this circuit will typically NOT trip short circuit protection. What does happen is the short essentially creates a load. Without an OCP the load just increases and increases until the wire heats up and the insulation melts off and there's a molten pile of flaming plastic at the bottom of the chassis. This is why rails are split up and "capped off" in most power supplies; there is a safety concern. We talked about this in the other thread but neglected to read any of it before bumping the thread with your mis-information.

• It IS true that marketing folks have been telling us why we need to split up our +12V rails for all the wrong reasons. But that's the marketing folks. It's pretty ignorant to assume that because THEY are wrong that there is no purpose to splitting up +12V rails. Quite frankly, "offers stability and cleaner power" sounds much more palletable than "won't necessarily catch fire". Like I said before, typically there is only one +12V source and there is no additional filtering stage added when the rails are split off that makes the rails any more stable or cleaner than if they weren't split at all. But the fact of the matter is, there's less components and less engineering to make a PSU that DOES NOT have the +12V rail split up, so it actually behooves marketing to convince you that you only need single +12V rails. Be mindful of this if someone starts trying to recite "myths" to you.

Is it true that some PSU's that claim to be multiple +12V rails don't have the +12V rail split at all?

• Yes, this is true. But it's the exception and not the norm. It's typically seen in Seasonic built units like the Corsair HX and Antec True Power Trio. But you can't apply what reviewers like Gabriel (Hardware Secrets) and myself have found on a handful of units to ALL units with multiple +12V outputs. Again: It's cheaper to make a single +12V rail PSU because you forego all of the components used in splitting up and limiting each rail. But some system builders adhere very closely to ATX12V specification for liability reasons, so a company that wants to get that business but also save money and reduce R&D costs will often "fib" and say the PSU has it's +12V split when it does not. Why don't those PSU companies get in trouble? Because Intel actually lifted the split +12V rail requirement from spec, but they didn't actually "announce" it. They just changed the verbiage from "required" to "recommended" leaving system builders a bit confused as to what the specification really is.

So I've pointed out the disadvantage of single +12V rails. Is there a disadvantage to multiple +12V rails?

• Yes, there can be. Early on, ATX12V spec said that the +12V should be split in two and that one of the two rails go only to the CPU and nothing else. The power supplies that did this split only allowed for up to 20A to go to all drives, graphics cards, etc. Not enough power. We typically only see this in power supplies with less than 600W. Once 600W+ power supplies became the norm, engineers started to follow the EPS12V specification which is meant more for servers than it is for gaming machines. What happened is we ended up with the same problem we had with two +12V rails because the EPS12V spec was having manufacturers put PCIe power connectors on the same rail as drives, etc. Since this time, Nvidia put their foot down and said, "Look.. if you want SLI certification you have to put PCIe on it's own +12V rail and you can't put any more than two PCIe connectors on that rail". This is what we commonly see today and why, for the most part, the fact that a PSU has multiple +12V rail outputs is not an issue.

Of course, the same marketing people that told you that you needed multiple +12V rails for "cleaner power" is now telling you you don't want multiple +12V rails because of "trapped power". You don't believe what they said then, why believe what they say now? Do you want to trust marketing or common sense and simple math?

Just to recap:

Originally posted by: flexy

The point is there IS no "increased stability" since literally ALL current so called "multi rail" PSUs use ONE rail.

True, there is no "increased stability", but it's not true that "ALL" so called multi-rail PSU's use one rail. This is a false generalization that you extrapulated from what you mis-read.

Originally posted by: flexy
The "multi rail" is virtual and its only "multi" since each wire has a limiter, limiting draw per lead/wire to a certain amperage.

This is correct.

Originally posted by: flexy
Look at the site above, look at the circuits and their anylyses. Be it OCZ Powerstream, Corsair, Thermaltake... all go in ONE common 12V output...so the "multiple rail stability" thing is a very well established MYTH....CPU, Graphics Card, Mobo, Sata, all get their 12V from ONE point in the PSU.

This is also correct.

Originally posted by: flexy
There are NO dedicated, separate circuits in the PSUs delivering 3 or 4 independent 12V rails of 12V.

There are some examples of two independent +12V rails for you: Topower/Tagan "Dual Engine" units, Enermax Galaxy, Channel Well's 1000W and up units including the Thermaltake Toughpower and Corsair HX1000, Cooler Master's Real Power Pro, Zalman's 850W and 1000W....

Originally posted by: flexy
This applies to 99% of all current so called "multiple rail" PSUs.

Wait a minute.... first you said "ALL" and now you say "99%" Besides, "who cares?" You seem to think that most people have been buying multiple +12V rails for the wrong reason. This may be true, but that doesn't discout their purpose, usefullness or what have you. Lynch the marketing guys. Not the engineers. Otherwise you're tipping at windmills.

 

[flexy]

appreciate your input!

Yes, on one site i read a statement which i interpreted as that actually a big majority use this "virtual rail" design, without being "exact" in my numerology I know now there's a a small amount of "real" split rails PSUs, as you also say mostly 1000W+ units.

>>
You seem to think that most people have been buying multiple +12V rails for the wrong reason.
>>

well actually i dont know why other people buy, or why they would have a preference for split rails over one rail.
I can just deduct from my own (false) assumptions: Reading "split rails" on a label made me think this means, well, truly independent power-circuits for each of the rails. "Stabilty" actually NOT so much in terms that the rails have OCP...(i didnt know about the technical details of implementation until today)...but rather assuming this "stabilty" would come from the fact that a draw on rail #1 wouldnt result in a voltage drop/fluctuation on any of the other rails. (Since they're supposed to be "independent"). (as in: I could put the strongest graphics-cards running SLI on such a PSU and not worry about problems arising from this on the CPU rail.....or vice versa.) At least that's how i interpreted what "split rails" actually means.

 

[jonnyGURU]

Originally posted by: flexy
well actually i dont know why other people buy, or why they would have a preference for split rails over one rail.

Even after I explained it? Wow. Ok... Given the REAL reasons why they split the rails and given the reason why "trapped power" is in itself a myth; tell me why someone would have a preference of single over split +12V rails?

Originally posted by: flexy
I can just deduct from my own (false) assumptions: Reading "split rails" on a label made me think this means, well, truly independent power-circuits for each of the rails. "Stabilty" actually NOT so much in terms that the rails have OCP...(i didnt know about the technical details of implementation until today)...but rather assuming this "stabilty" would come from the fact that a draw on rail #1 wouldnt result in a voltage drop/fluctuation on any of the other rails. (Since they're supposed to be "independent"). (as in: I could put the strongest graphics-cards running SLI on such a PSU and not worry about problems arising from this on the CPU rail.....or vice versa.) At least that's how i interpreted what "split rails" actually means.

Right, it was a false assumption, but it is a common one because of the way split +12V rails have been falsely marketed in the past. But that doesn't negate multiple +12V rails' usefulness. Let's apply this to a timeline and maybe you'll better understand:

Step 1: 240VA limit was implemented into the ATX12V so DC output wires wouldn't burst into flames during a short.

Step 2: Marketing people put a spin on the real reason PSU's had split +12V rails so it sounded like we'd actually get a performance gain instead of explaining the safety concerns.

Step 3: We found out that marketing was lying to us.

Step 4: The same marketing folks are now telling us that we want single +12V rail because of "trapped power".

When we figured things out during "step 3" we should have referenced "step 1" for the facts. Instead, we're buying into more FUD from the same guys that lied to us before. When will the vicious cycle end?

 

[flexy]

jonny, someone could see a single-rail PSU, not care about ATX specs and see "60A" on 12V, giving him more headroom instead of having a PSU capped at 20A per rail since he want to run his Quad at 1.6Vcore 4000mhz?? (He could, however, certainly also get a split-rail with multiple 25A rails....but..i am just guessing the single rail 60A would be cheaper.)

 

[jonnyGURU]

Originally posted by: flexy
jonny, someone could see a single-rail PSU, not care about ATX specs and see "60A" on 12V, giving him more headroom instead of having a PSU capped at 20A per rail since he want to run his Quad at 1.6Vcore 4000mhz??

And my answer to that would be... "Why would it/should it make any difference?" If a person really thinks they need more than 20A for a CPU, never mind 60A, I would be half tempted to tell them to put the screw driver down and step away from the computer. Even an overclocked CPU isn't going to use/need HALF of the power that's available to it.

You have a PSU with 60A on one +12V rail or a PSU with four +12V rails at 20A each. It shouldn't make any difference as far as the overclockabiltiy of the CPU or overhead or anything. Honestly, let's use real world examples here. How many posts can you find on the Internet where someone had a PC, overclocked the CPU and suddenly had the PC shut down during benching? It doesn't happen.

The marketing people have you (among others) brainwashed and scared. They had you brainwashed into thinking that multiple +12V rails would bring you "stability" and now they have you scared into thinking that rails somehow limit your capabilities.

I suggest taking a look at some real world examples. Look at some PSU's with four or more +12V rails, see how those rails are split up, what the limit (officially called OCP. I'm throwing that out there in case I decide to use that term) on each rail is and what connectors are on those rails (Wolf and I try to do this on all of my reviews) and then see what the typical power consumption is for the components on each of those connectors. You will then likely find that these split up rails typically have more than twice the current the components need available to those components.

And it's not like every customer has to educate them in this manner before buying a PSU. For the average customer the inquiry is a kin to asking what kind of tree was used to make the cardboard holding their 12 pack of Heineken togheter. I'm telling you that you need to educate yourself on it if you want to discuss it effectively. The average customer doesn't need to discuss it simply because it simply shouldn't matter. But because marketing people choose to make it a bullet point (single and multiple +12V rails, both) everyone asks.

 

[jonnyGURU]

Originally posted by: flexy
(He could, however, certainly also get a split-rail with multiple 25A rails....but..i am just guessing the single rail 60A would be cheaper.)

25A... 20A... either one is more than enough. 25A just means the PSU would shut down later in the event of a short. You would think the single 60A would be cheaper for the customer, but the truth of the matter is the manufacturer isn't handing that cost savings to the customer. That's why they have to make it a bullet point that the PSU has a single +12V rail in the first place. In order to make YOU feel better about spending more money on the unit!

 

[Modelworks]

The atx power supply is something of an oddity in power supplies.
If you see a bench power supply that says separate rails believe it because they almost always are separate rails.

I think the way pc power supplies use the term multiple rails is somewhat of a cheat on the whole concept compared to the way its generally used in engineering.

 

[flexy]

actually, if the separate rails *do* have independent voltage regulation (even if the rails are only "virtual") it doesnt matter then whether there is only one 12V rail "internally". This actually serves the purpose of the rails being semi-independent. Correct me if i am wrong.

 

[jonnyGURU]

Originally posted by: flexy
actually, if the separate rails *do* have independent voltage regulation (even if the rails are only "virtual") it doesnt matter then whether there is only one 12V rail "internally". This actually serves the purpose of the rails being semi-independent. Correct me if i am wrong.

If the rails are truly separate and kept separate (like the Thermaltake Toughpower 1000W, for example) then they serve MORE of a purpose. You have the "safety" concerns addressed (unless the individual output can put out enough juice to catch a wire on fire) and you get the self regulation, elimination of cross rail noise, etc.

But they only tend to do true separate output on higher wattage units because there comes a point where one medium sized transformer is still cheaper than two smaller ones... but when you have to put out 1000W+ of power, two medium size transformers is cheaper than one big one.

 

[Modelworks]

Originally posted by: flexy
actually, if the separate rails *do* have independent voltage regulation (even if the rails are only "virtual") it doesnt matter then whether there is only one 12V rail "internally". This actually serves the purpose of the rails being semi-independent. Correct me if i am wrong.

Thats what I hate about the way pc power supplies use the term rail.
They take a single voltage source , feed it through whats essentially a circuit breaker, and call it separate rails.

Thats like me telling everyone that their home wall receptacles have separate rails( if it were dc).

 

[HOOfan 1]

Originally posted by: Modelworks

Thats what I hate about the way pc power supplies use the term rail.
They take a single voltage source , feed it through whats essentially a circuit breaker, and call it separate rails.

Thats like me telling everyone that their home wall receptacles have separate rails( if it were dc).

As Jonny already said, that is marketing doubletalk. Just like when a car company says their car has 350 Horsepower...well maybe at the fly wheel, but once it gets to the road through the tires it might only be 290 Horsepower.

Why do you really only see this with consumer level ATX power supplies? Because that is where the gullible people are and that is where most of the individual sales are.

 

[Martimus]

Originally posted by: flexy
actually, if the separate rails *do* have independent voltage regulation (even if the rails are only "virtual") it doesnt matter then whether there is only one 12V rail "internally". This actually serves the purpose of the rails being semi-independent. Correct me if i am wrong.

I would say that is wrong, because you wouldn't want one supply to have a drop in its voltage because you are loading the other supply - which is what would happen in the case that you illustrate.

 

[jonnyGURU]

Originally posted by: HOOfan 1

Originally posted by: Modelworks

Thats what I hate about the way pc power supplies use the term rail.
They take a single voltage source , feed it through whats essentially a circuit breaker, and call it separate rails.

Thats like me telling everyone that their home wall receptacles have separate rails( if it were dc).

As Jonny already said, that is marketing doubletalk. Just like when a car company says their car has 350 Horsepower...well maybe at the fly wheel, but once it gets to the road through the tires it might only be 290 Horsepower.

Why do you really only see this with consumer level ATX power supplies? Because that is where the gullible people are and that is where most of the individual sales are.

Actually, server level EPS power supplies use the term rails in the same way.