the ultimate "do i need a 24pin PSU ?" thread

[jonnyGURU]

Method 3? Why are they putting SATA with the CPU?!?!

Aaargh!

If you want to know what drives can do to a 12V rail, hook a cold-cathode inverter to the same cable as a hard drive and watch what the light does when the drive spins up.

Sure that's an example from "worst case scenario," but on a small scale it's the same thing that happens inside the power supply when drives and fans are on the same rail as the CPU and video cards, etc.

Believe me, I'm NO advocate of dual rails. I like one big 12V rail so I get power where I need it. But someone has to explain the logic of the load division of these dual rails.

 

[Algere]

Going back & forth here now & method 3 is a possibility if the diagram does indeed indicate SATA as being on the same rail as the CPU.

Alright when first going into the dual rail PSU subject, I thought having the PCIe 16X gfx cards's power seperate from the rail the CPU uses was the intention or at least one of 'em. However according to Zepper, the motherboard has filters to clean out the power, I soon later went into it a bit further & found out that the capacitors are responsible for filtering out noise & what not. From this I gather the reason the PCIe 6P plug & PCIe slot are not seperate has to do with the main reason dual rails exist in the first place. Ya know, the bit about having too much power on one rail & prolong use under heavy loads. So now I think having rails seperate for cleaner power is one of a few side benefits for dual +12V rails, but not the main reason.

AFAIK at this point, the components that share the same rail with the CPU has/uses caps & the ones that don't have such are powered via the other rail.

 

[jonnyGURU]

I just took a look at the ATX12V2.01 specs on Formfactors.org. 12V2 is just for the CPU. Enermax is making sh!t up as they go and I think I know why.

To be "Intel 2.01" certified, you need to have a second rail if the 12V has more than 18A on it. And you can't have 20A or more on either rail if you have dual rails. So what Enermax is doing is splitting up the load so they can stay within Intel specs but still provide enough power across the board for anything you throw at it. In sense: A worthless dual 12V rail. They should've made one big 12V rail and called it a day. :-/

 

[Algere]

Checking it...

 

[Algere]

Where (what page) does it say the CPU is required to be seperate from other components?

 

[jonnyGURU]

Originally posted by: Algere
Going back & forth here now & method 3 is a possibility if the diagram does indeed indicate SATA as being on the same rail as the CPU.

Alright when first going into the dual rail PSU subject, I thought having the PCIe 16X gfx cards's power seperate from the rail the CPU uses was the intention or at least one of 'em. However according to Zebo, the motherboard has filters to clean out the power, I soon later went into it a bit further & found out that the capacitors are responsible for filtering out noise & what not. From this I gather the reason the PCIe 6P plug & PCIe slot are not seperate has to do with the main reason dual rails exist in the first place. Ya know, the bit about having too much power on one rail & prolong use under heavy loads. So now I think having rails seperate for cleaner power is one of a few side benefits for dual +12V rails, but not the main reason.

AFAIK at this point, the components that share the same rail with the CPU has/uses caps & the ones that don't have such are powered via the other rail.

The caps on the board ARE for filtering out the voltage, true.

But the point of dual rails is just to have cleaner power. It's not just one of a few side benefits. The 12V coming from the power supply is still the same all of the way up to the two transistors that make the two 12V rails. The only benefit to having two seperate transformers would be to have the devices on the one not create noise on the other.

 

[jonnyGURU]

Originally posted by: Algere
Where (what page) does it say the CPU is required to be seperate from other components?

I'm not pulling that beast up again. It's all over the document. 12V2 is for the 2x2 connector. No other connector listed in the entire document is using the 12V2.

 

[Algere]

Haha ha

Did a Adobe Reader search for "CPU" & "Processor". As far as I can tell, all it says about the +12V2 current is...

"12V2 supports processor power requirements and must have a separate current limit"

Don't think it means that components have to be seperate from the CPU.

P.S. Oh and this "Isolated current limit on 2x2 connector for 12V2 rail.", same thing I suppose.

Yep editing again . From what I can tell, it could mean that the CPU can't be powered by both rails.

 

[jonnyGURU]

Ok. That is repeated over and over again in the document that the 2x2 connector is for the processor and that the 12V2 is for the 2x2 connector.

But if that wasn't enough... (doesn't seem like it is) did you notice in the diagram of the power connectors (which looks similar to Enermax's) that all of the power connectors are getting their 12V from 12V1 except for the 2x2?

 

[jonnyGURU]

Originally posted by: Algere
From what I can tell, it could mean that the CPU can't be powered by both rails.

Haha... That's not what any of that means. That wouldn't even be something they'd point out because it's fairly obvious.

 

[Algere]

The only benefit to having two seperate transformers would be to have the devices on the one not create noise on the other.

Not the only.

From another post I posted in,

1) Cheaper to make. Would a single rail 700W PSU cost more than a 700W multi-rail PSU???
2) Easier to implement. That or there's the challenge PSU makers would face in attempts to regulate high amperage on a single rail. I suppose reasons 1 & 2 intertwine with one another.
3) Cooler operation, assumingly more important now than ever before. PSUs act as a secondary exhaust for a PC plus add to the fact that the higher the wattage the PSU is outputting, the more heat is produced. Both factor against a PSUs ability to provide full power & as some know, more heat = less power.
4) Prevents CPU power load (and/or other?) from straining the rails. In this case as load increases, voltage goes down. Now my guess is that voltage on the other rail doesn't go down/follow as well.
5) Cleaner power, or at the very least dirty power is isolated to it's own rail.

 

[Algere]

Originally posted by: jonnyGURU

Originally posted by: Algere
From what I can tell, it could mean that the CPU can't be powered by both rails.

Haha... That's not what any of that means. That wouldn't even be something they'd point out because it's fairly obvious.

Well a PCIe card powered by both PCIe slot & PCIe 6P power plug can be powered by both rails. Those cards also have 2 groups of capacitors, one can guess one group is dedicated to the power coming from the PCIe slot and the other group from the PCIe 6P plug & AFAIK the ATX12V 2.0 spec doesn't say you can't combine rails to power up a PCIe card.

P.S. It says 2x2 must have seperate current limit, which to me sounds like seperate current limit & not seperate current, especially from other components.

 

[jonnyGURU]

Originally posted by: Algere

The only benefit to having two seperate transformers would be to have the devices on the one not create noise on the other.

Not the only.

From another post I posted in,

1) Cheaper to make. Would a single rail 700W PSU cost more than a 700W multi-rail PSU???

That would be an assumption that two small MOSFET's would be cheaper than one big one. I don't think so.

2) Easier to implement. That or there's the challenge PSU makers would face in attempts to regulate high amperage on a single rail. I suppose reasons 1 & 2 intertwine with one another.

The additional MOSFET would also require more PCB space. Again, not cheaper... or "easier."

3) Cooler operation, assumingly more important now than ever before. PSUs act as a secondary exhaust for a PC plus add to the fact that the higher the wattage the PSU is outputting, the more heat is produced. Both factor against a PSUs ability to provide full power & as some know, more heat = less power.

Wrong. Each of two smaller MOSFET's will run cooler than one big one, but the heat generated by two is still greater than the heat generated by one.

4) Prevents CPU power load (and/or other?) from straining the rails. In this case as load increases, voltage goes down. Now my guess is that voltage on the other rail doesn't go down/follow as well.

Only if the load balance is extremely uneven. If a power supply is engineered for heavy 12V load, it does not require as much of a load on it's other rails to prevent the voltage dropping on the 12V.

Trust me. This "opinion" does NOT hold water. If I zero out all of the rails but the 12V on either a single or dual 12V rail power supply, the voltage STILL drops significantly. Again, it's not as if the power provided to the second rail is from a seperate rectifier or transformer. If it was, it would completely and totally blow 1, 2 and 3 out of the water exponentially (heat, cost, space.)

5) Cleaner power, or at the very least dirty power is isolated to it's own rail.

Bingo! They got one right.

 

[Algere]

That would be an assumption that two small MOSFET's would be cheaper than one big one. I don't think so.

I'm not the only one that said this & it makes you wonder why NewEgg would sell the dual rail version cheaper than the single rail version. Unless the Sparkle brand warrants a +$10 increase over the Fortron.

P.S. You said MOSFET's & I saw MOSFET's and in my head had dual rails.

Wrong. Each of two smaller MOSFET's will run cooler than one big one, but the heat generated by two is still greater than the heat generated by one.

Says otherwise

P.S. You said MOSFET's & I saw MOSFET's and in my head had dual rails. One and the same when it comes to heat?

Only if the load balance is extremely uneven. If a power supply is engineered for heavy 12V load, it does not require as much of a load on it's other rails to prevent the voltage dropping on the 12V.

The part in my post where it says "in this case", it was a hyperlink to the single rail vs. dual rail challenge thread & that was responded further in this thread.

Bingo! They got one right.

TY

TR, Mark R, & Zepper would say otherwise. Their main reason dual rails exist is because of safety, not clean power.


EDITED for grammar, links, etc.

 

[Algere]

Originally posted by: jonnyGURU
Ok. That is repeated over and over again in the document that the 2x2 connector is for the processor and that the 12V2 is for the 2x2 connector.

But if that wasn't enough... (doesn't seem like it is) did you notice in the diagram of the power connectors (which looks similar to Enermax's) that all of the power connectors are getting their 12V from 12V1 except for the 2x2?

Where is it reported? If it's reported all over the spec sheet then it shouldn't be hard to quote & does it say ATX12V 2.0 PSUs must follow that diagram? & what would you follow, the diagram (pg. 34) which shows...

RAIL 1: 24P plug only
RAIL 2: 4P Molex, 3P Floppy, P4/CPU 4P plug, & SATA

or the table (pg. 35-37) below it

RAIL 1: 24P plug, 4P molex, SATA, Floppy
RAIL 2: P4/CPU 4P

If you were to strictly follow the diagram/table for ATX12V 2.0 compliance. Then having 6P PCIe power plug(s) is out of the question since it's not even listed in the spec.

P.S. Shot an email towards Enermax/Maxpoint to see what they say about it. In the meantime, THG says,

"There is a second +12V voltage slot for the CPU, so the other 12V current doesn't get overtaxed if there is a sudden surge in the CPU load."

Based on that statement alone, it doesn't look like the main purpose of dual rails is for noise/ripple isolation (side benefit), nor does it say that the second +12V rail is for the CPU alone & if that's the case, having the P4/CPU connector seperate from the main 24P connector or not is moot.

Update: No email as of yet however I did stumble upon Intel's ATX12V PSU list & it has the "EG425P-VE" listed, which as you know follows the rail design of method 2 as well.

 

[jonnyGURU]

Originally posted by: Algere

P.S. You said MOSFET's & I saw MOSFET's and in my head had dual rails.

P.S. You said MOSFET's & I saw MOSFET's and in my head had dual rails. One and the same when it comes to heat?

Aargh! Why do you keep saying that? You wonder why I misunderstand and misquote you?

MOSFET's are the Metal-Oxide Semiconductor Field-Effect Transistors that you see on the heatsinks in a power supply. The little black boxes. Those regulate the voltage coming into the power supply and regulate the rails for the voltage coming out of the power supply.

Now I believe Zepper over TrustedReviews, even though we don't agree on the brands we use. But I'll stand corrected. Also the verbiage in your bleedingedge.com link is so matter of fact that I must stand corrected.

I suppose a MOSFET that runs 38A is more expensive than two MOSFET's that run 20A for the same reason an Intel Pentium 4 3.8 is more expensive than two 2.0's. And I can see the concern for "safety" when a single, high amperage MOSFET's gets very hot. It can and will explode. Fatal? I'm not sure about that. When they blow up the parts just richochet around inside the power supply. The AC to DC transformer prevents any high-voltage from going to your computer components and power is isolated to ground on the AC side where the MOSFET isn't going to change whether the power supply has single or dual rails.

Someone still has to explain to me how two MOSFET's can be cooler than one, though. Especially when they're on the same heatsink and the heatsink and cooling method for the power supply doesn't change. Obviously they're "safer" in that each one doesn't run "as hot," but like I said... you've got an aluminum heatsink that saturates only that much quicker with more MOSFET's on it and fans that don't spin any faster. And if I look in my Enermax Noisetaker (dual a8A rails), I don't see heatsinks any larger than those in my Ultra (single 34A rail.) Maybe I need to think outside of the box, but I'm not able to grasp the concept.

 

[jonnyGURU]

Originally posted by: Algere

Originally posted by: jonnyGURU
Ok. That is repeated over and over again in the document that the 2x2 connector is for the processor and that the 12V2 is for the 2x2 connector.

But if that wasn't enough... (doesn't seem like it is) did you notice in the diagram of the power connectors (which looks similar to Enermax's) that all of the power connectors are getting their 12V from 12V1 except for the 2x2?

Where is it reported? If it's reported all over the spec sheet then it shouldn't be hard to quote & does it say ATX12V 2.0 PSUs must follow that diagram? & what would you follow, the diagram (pg. 34) which shows...

RAIL 1: 24P plug only
RAIL 2: 4P Molex, 3P Floppy, P4/CPU 4P plug, & SATA

or the table (pg. 35-37) below it

RAIL 1: 24P plug, 4P molex, SATA, Floppy
RAIL 2: P4/CPU 4P

If you were to strictly follow the diagram/table for ATX12V 2.0 compliance. Then having 6P PCIe power plug(s) is out of the question since it's not even listed in the spec.

P.S. Shot an email towards Enermax/Maxpoint to see what they say about it. In the meantime, THG says,

"There is a second +12V voltage slot for the CPU, so the other 12V current doesn't get overtaxed if there is a sudden surge in the CPU load."

Based on that statement alone, it doesn't look like the main purpose of dual rails is for noise/ripple isolation (side benefit), nor does it say that the second +12V rail is for the CPU alone & if that's the case, having the P4/CPU connector seperate from the main 24P connector or not is moot.

I never said you COULDN'T use the two rails for whatever means you wish. I'm just saying, "what's the point?" IF your primary concern is clean power for the CPU.

I know my statement has a lot fo conditions, but it's based on this....

Initially, the 2x2 was introduced to supply additional power to the CPU. Initially, the second 12V rail was for the 2x2 in an effort to isolate it's power from the rest of the PC.

Now the whole spec is open to interpretation and has to be tweaked by the manufacture, because if you look at it you'll see that they won't approve a power supply with 20A or more on a single 12V rail in a dual rail arrangement. Why? Are we back to the heat, cost, safety thing? Are we in the same boat with two 20A MOSFET's as we are with a single 34A, 38A or 40A?

Unfortunately, that "limit" means that a power supply that's "Intel approved" that may be fine for an Intel rig w/o SLI will not necessarily be appropriate for a rig with an AMD and SLI unless you take your two rails and split up the load accordingly.

Think about it. What's 18A on 12V2 going to do for you if you only have it feeding the 2x2? Nothing, agreed? So put that power elsewhere. A PCIe connector? Sure. Why not. You're going to need to, because 18A on 12V1 isn't enough for a lot of high end systems and 18A on 12V2 IF you were to adhere to the ATX12V2.01 specs would be too much.

That's why Enermax split the load up the way they did. I'm not arguing with you. I'm not saying "they're not allowed to." I'm just frustrated that Enermax CHOSE to split the rails up the way they did.

Personally, I would put my drives and fans and lights on one rail (with appropriate filtering on the drive connectors just in case) and put the 20/24-pin, 2x2 and one SLI (not enough juice for two) on the other.

Otherwise, Enermax is only utilizing the "cheaper" and supposedly "cooler" advantages of dual rails and not the primary "cleaner" factor of it.

All of this is "in my humble opinion"... of course.

 

[Romano7]

better download our christian screensavers www.jesus-flower.com

 

[Algere]

Originally posted by: jonnyGURU

Originally posted by: Algere

P.S. You said MOSFET's & I saw MOSFET's and in my head had dual rails.

P.S. You said MOSFET's & I saw MOSFET's and in my head had dual rails. One and the same when it comes to heat?

Aargh! Why do you keep saying that? You wonder why I misunderstand and misquote you?

ahAhaH, ya got me there. Wanted to cover my bases in case I missed something + wasn't sure if there was a difference between MOSFETs & rails under the subject at hand.

Originally posted by: jonnyGURU
Now I believe Zepper over TrustedReviews, even though we don't agree on the brands we use. But I'll stand corrected. Also the verbiage in your bleedingedge.com link is so matter of fact that I must stand corrected.

I suppose a MOSFET that runs 38A is more expensive than two MOSFET's that run 20A for the same reason an Intel Pentium 4 3.8 is more expensive than two 2.0's. And I can see the concern for "safety" when a single, high amperage MOSFET's gets very hot. It can and will explode.

:thumbsup:

Originally posted by: jonnyGURU
Fatal? I'm not sure about that. When they blow up the parts just richochet around inside the power supply. The AC to DC transformer prevents any high-voltage from going to your computer components and power is isolated to ground on the AC side where the MOSFET isn't going to change whether the power supply has single or dual rails.

I'm gonna take a long shot here but... would it be possible a fire could start within a PSU & then possibly spread?

Originally posted by: jonnyGURU
Someone still has to explain to me how two MOSFET's can be cooler than one, though. Especially when they're on the same heatsink and the heatsink and cooling method for the power supply doesn't change. Obviously they're "safer" in that each one doesn't run "as hot," but like I said... you've got an aluminum heatsink that saturates only that much quicker with more MOSFET's on it and fans that don't spin any faster. And if I look in my Enermax Noisetaker (dual a8A rails), I don't see heatsinks any larger than those in my Ultra (single 34A rail.) Maybe I need to think outside of the box, but I'm not able to grasp the concept.

Like you said, "each one doesn't run as hot". Therefore you have 2 warm heat sources versus a single very hot heat source. With the increased surface area of the two heat sources, heat is more efficiently spread throughout the heatsink. Whereas with a single hot heat source situated at the middle of a heatsink for instance, the cooler parts of the heatsink would be on the outer regions of the heatsink & the hottest parts would be at the center. Now imagine if you had sufficient enough airflow to make contact with the entire heatsink. Air flow hitting the outer regions of the heatsink wouldn't be as important as the airflow hitting the center when it comes to cooling. Now if you count only the air flow hitting the center, it isn't as much as the amount of airflow hitting the entire heatsink. NEways that's the way I see it possibly going down.

Originally posted by: jonnyGURU
I never said you COULDN'T use the two rails for whatever means you wish. I'm just saying, "what's the point?" IF your primary concern is clean power for the CPU.

Aye, you didn't say that. However you did say the main reason for dual MOSFET's/Rails was for noise/ripple isolation. Which contradicts the sources I've quoted/listed.

Originally posted by: jonnyGURU
Initially, the 2x2 was introduced to supply additional power to the CPU.

:thumbsup:

Originally posted by: jonnyGURU
Initially, the second 12V rail was for the 2x2 in an effort to isolate it's power from the rest of the PC.

That's debatable & as you've said, open to interpretation. Follow the table (ATX12V 2.01; pg. 35-37) or follow the fact that it doesn't necessarily say the CPU's rail must be seperate from other components. Suppose having two +12V rails is open as well since IIRC it suggests having two rails if a PSUs total +12V current exceeds 18A.

Originally posted by: jonnyGURU
if you look at it you'll see that they won't approve a power supply with 20A or more on a single 12V rail in a dual rail arrangement.

From what I can tell (ATX12V 2.0 specs) it says if a PSU's total +12V current exceeds 18A, a second rail should be made available. Now one can imagine it could mean that the first rail must not exceed 18A but that the second rail can exceed 18A or that the rails can be split in two under any condition as long as it exceeds 18A e.g. if a PSU's total output is 22A. What's to stop a PSU manufacturer from making a PSU with one rail at 18A and the other at 4A while still claiming ATX12 2.0 compliance - doubtful it would ever happen but nevertheless.

However if it is how you say it is (not allowed to have 20A or more on 1 rail), then I'd like to see how the OCZ boyz in the CPU forum (IIRC) respond to that since the 600W Powerstream has 20A on one of it's rails and is marketed as ATX12V 2.01 compliant.

Originally posted by: jonnyGURU
Unfortunately, that "limit" means that a power supply that's "Intel approved" that may be fine for an Intel rig w/o SLI will not necessarily be appropriate for a rig with an AMD and SLI unless you take your two rails and split up the load accordingly.

That's probably the reason why Enermax's 600W Noisetaker PSU is certified by nVidia for SLI. Could be why that Enermax PSU has the nVidia logo & nVidia SLI certification label on their site. Now if by "Intel approved" you mean a PSU with the P4/CPU 4P connector with it's own dedicated rail. Then yea it would be a no go on high-end cards under SLI but then again, I haven't seen any dual rail PSUs with that design in combination with SLI certification/marketing. Nevertheless you've proven a point in this matter.

Originally posted by: jonnyGURU
Think about it. What's 18A on 12V2 going to do for you if you only have it feeding the 2x2? Nothing, agreed?

You bet'cha, no argument there.

Originally posted by: jonnyGURU
So put that power elsewhere. A PCIe connector? Sure. Why not. You're going to need to, because 18A on 12V1 isn't enough for a lot of high end systems and 18A on 12V2 IF you were to adhere to the ATX12V2.01 specs would be too much.

Doubtful or at least not efficient use of the rails IMO.

System Setup: A64 s939 dual core & 6800 Ultra (SLI) w/ Enermax Noisetaker 600W SLI PSU under 18A/17A (can go vice versa: 17A/18A) method.

Your example
Rail 1: P4/CPU connector (A64 DC=9.2A max), main 24P (PCIe 8X slot=2.1A, PCIe 8X slot=2.1A, PCIe 1X=0.5A, PCIe 1x=0.5A, fan headers=<1A = 6.2A total) = 15.4A
PCIe connector can only provide 2.6A.

Rail 2: drives, fans, other PCIe connector, etc. (17A available).

Problem: Rail 1 is overloaded/passes 18A limit, should you go dual core (under max. consumption specs).

Enermax design: method 2
Rail 1: P4/CPU connector (A64 DC=9.2A max.) + main 24P (PCIe 8X slot=2.1A, PCIe 8X slot=2.1A, PCIe 1X=0.5A, PCIe 1x=0.5A, fan headers=<1A = 6.2A total) = 15.4A

Rail 2: both PCIe, drives, fans, etc. = 17A available

Problem: See any (non-noise/ripple related) compared to yours?

If it matters, a 6800 Ultra under SLI can consume up to 14.8A. One could guesstimate under method 2, around 4A of that 14.8A load could be taken care of by the motherboard's PCIe slots, while 10.8A remaining will be taken cared of by the second rail. That leaves you with 6.2A left to be used for drives, fans, etc.

Also note Intel s775 dual cores will consume up to 130W/10.8A max. So if you were to switch the dual core A64 in the system setup above with Intel's dual core, coincidentally that'll be a bit under the 18A rail limit. Coincidence?

Originally posted by: jonnyGURU
That's why Enermax split the load up the way they did. I'm not arguing with you.

But I am , nothing personal of course

Originally posted by: jonnyGURU
I'm not saying "they're not allowed to." I'm just frustrated that Enermax CHOSE to split the rails up the way they did.

Yea you didn't say that, you said the way Enermax did it wasn't ATX12V 2.0 compliant.

Originally posted by: jonnyGURU
Otherwise, Enermax is only utilizing the "cheaper" and supposedly "cooler" advantages of dual rails and not the primary "cleaner" factor of it.

Sacrifice a bit more cleanliness for more usable power. Guess it's either one or the other.

 

[jonnyGURU]

I NEVER said that Enermax wasn't ATX12V2.01 compliant. I said that they were taking things and manipulating the specs to their own liking. My exact words were: "Enermax is making sh!t up as they go."

Intel is blurry about what you can and can not have on 12V2. All they say is "2x2 is on 12V2," so like I said... it's open to interpretation. Enermax could decide to put nothing but fan headers on the 12V2 for all Intel's concerned.

I was under the impression that the sole reason for dual 12V is rail cleanliness because that's what all of the PSU manufacturers SELL us. Do you think they'd advertise, "Hey! We're making power supplies CHEAPER by splitting the 12V rail!" ? I don't think so. I got sucked in like they expected me to.

As for catching fire.... I suppose that is a very good possibility. I've had MOSFET's on motherboards shoot out 2" high blue flames before. And those were rated at much lower wattage. I could only imagine what would happen if a big ol' 34A 12V MOSFET would do if it decided to turn into a blow torch!

And you don't have to give me examples. I've always taken issue with Intel's spec for ATX12V. I'm fully aware that a PCI-e card can suck up 7A off the 12V rail. Imagine if there were Penrium 4 boards with SLI (which there will be later this year.) The CPU and two video cards ALONE would take in 21A under load. Yet Intel wants to keep the 12V rails under 20A each. I think that's unrealistic. I guess we'll just have to wait and see what Intel does with the ATX spec once nVidia releases their SLI chipset for the Pentium 4 and people start using them.

 

[Algere]

Originally posted by: jonnyGURU
I NEVER said that Enermax wasn't ATX12V2.01 compliant. I said that they were taking things and manipulating the specs to their own liking. My exact words were: "Enermax is making sh!t up as they go."

You're right :light:, my mistake. Chalk one up on my end for misunderstanding.

Originally posted by: jonnyGURU
Intel is blurry about what you can and can not have on 12V2. All they say is "2x2 is on 12V2," so like I said... it's open to interpretation. Enermax could decide to put nothing but fan headers on the 12V2 for all Intel's concerned.

Agreed .....

Originally posted by: jonnyGURU
I was under the impression that the sole reason for dual 12V is rail cleanliness because that's what all of the PSU manufacturers SELL us.

Must admit I initially was under the same impression as well & would say the same thing you were (are?) saying when someone asked the main reason for dual (multi-rails if you wanna extend it further) rails.

Originally posted by: jonnyGURU
Do you think they'd advertise, "Hey! We're making power supplies CHEAPER by splitting the 12V rail!" ? I don't think so. I got sucked in like they expected me to.

Ditto, nevertheless despite what they sold me on. For the other benefits that dual rails offer. I'd purchase a dual rail PSU again unless someone can give me a good reason not to other than the small loss of usable power due to the splitting of the rails (compared to single rail), which isn't a major factor in my case.

Originally posted by: jonnyGURU
As for catching fire.... I suppose that is a very good possibility. I've had MOSFET's on motherboards shoot out 2" high blue flames before. And those were rated at much lower wattage. I could only imagine what would happen if a big ol' 34A 12V MOSFET would do if it decided to turn into a blow torch!

Fire, fire fire fire!! :evil:

/Beavis off

Originally posted by: jonnyGURU
And you don't have to give me examples.

I'm not attempting to insult your intelligence if that's how I'm coming off. Placing examples down just helps me order my thoughts & hopefully prevents misunderstanding.

Originally posted by: jonnyGURU
Imagine if there were Penrium 4 boards with SLI (which there will be later this year.) The CPU and two video cards ALONE would take in 21A under load. Yet Intel wants to keep the 12V rails under 20A each. I think that's unrealistic.

I sortof imagined the same except instead of a Pentium 4, I had a dual core Pentium instead (per my previous post). There's a limit to how much power a PCIe slot can provide on the motherboard's end & the limit AMD & Intel are set on the processor side. Under "method 2", the motherboard (main 24P) & CPU (4P 12V) doesn't exceed 18A with known Intel & AMD dual core power consumption specs & SLI.

Question: How do I keep usable power lost (compared to single rail) caused by splitting of the rails to a minimum while having the capability to power all the components under a system?

Answer: set a power envelope for components on one rail (e.g. Motherboard+CPU must not exceed 18A) & allow the second rail to cover the rest of the +12V power needs of a system; as system demands.


Setup: 2x18A rail PSU

Intel dual core
130W maximum or 10.8A (130W divided by 12V)

Motherboard
PCIe 8X = 2.1A (SLI)
PCIe 8X = 2.1A (SLI)
PCIe 1X = 0.5A
PCIe 1X = 0.5A
fan headers = <1A

Total: 6.2A

Source: PCIe power specs


CPU (10.8A) + motherboard (6.2A) = 17A


Following this, if a dual core Pentium on s775 + SLI does indeed exceed 21A total, I doubt you're gonna exceed the first rail & the remaining power (21A-18A=3A) will be supplied by the other rail via PCIe power connectors.

Now on the other hand, there's a possible downside to method 2 under a non-SLI situation.

Intel dual core
130W maximum or 10.8A (130W divided by 12V)

Motherboard
PCIe 16X = 5.5A max.
PCIe 1X = 0.5A
PCIe 1X = 0.5A
PCIe 1X = 0.5A
fan headers = <1A

Total: 8A


CPU (10.8A) + Motherboard (8A) = 18.8A exceeds 18A


Although based on a possible maximum. It's possible the rated amps aren't typical & actual power load may be lower in which a 18A/rail PSU could handle. For now it's a ? until someone can test a non-SLI Intel dual core (130W) setup with a 18A/rail PSU. AMD's dual core (what I intend to use) still fits under the 18A/rail limit on non-SLI AMD dual core setups.

Originally posted by: jonnyGURU
I guess we'll just have to wait and see what Intel does with the ATX spec once nVidia releases their SLI chipset for the Pentium 4 and people start using them.

Wait & see but in the meantime, I'll speculate.

They could either adopt 3+ rail PSUs into a new PSU spec or nVidia might say EPS12V PSU required for P4 SLI. They could also increase the rail limit a bit further - albeit doubtful but possible.

 

[pilryu]

I'm running all of equipment on 20 pin power source and it works perfectly, even when OC.

 

[cubeless]

right now i'm typing on a 9nda3j/a64 3000 retail@2.31@1.65v/2 80GB pata/1 cdrw/1 6800 oc on an achieve ax500 cheapass 20 pin psu... and it's running p95 torture and hotcpu in the background...

and my dfi nf4ut is running about the same with a 550 flower...

better psu's might be nice, but these old dogs still hunt...

 

[Azsen]

Well I looked in my Enermax 470W EG475P-SFMA24P manual for details about how to make use of the second 12v rail. They recommend I plug the special molex connector with "Extra" written on it into my AGP 8x high end graphics card to "provide it with stable current".

I assume this is how you run stuff off the second 12v rail. Unfortunately I looked over all the cables coming out of the PSU and none of them have "Extra" written on the molex cable. I think they might have replaced it with the PCI-E 6pin connector? So it would appear I cannot make use of the second 12v rail at all now seeing I have an AGP graphics card??

Here I am thinking I'm getting a good PSU that will make use of the 31A on both 12v rails and in actual fact my whole system is probably running off the one 12v rail rated at 16A? I need my 15A on the other 12v rail. How do I make stuff run off the second 12v rail?

 

[imported_Xavier]

Current Specs:

AMD Athlon XP 2500 "Barton" - Not Overclocked; To lazy
ATI Radeon 9800 Pro -Card has started to show signs of aging after 2 years
Maxtor 40GB 7200rpm HDD
1024mb Kingston "HyperX" Ram
Asus A7N8X Deluxe Retail Motherboard
Antec True430 PSU - 20-pin connector


I am now looking into the process of upgrading.

This is pretty much what I am wanting to buy:

MSI nVIDIA GeForce 6600GT Video Card, 128MB DDR3, 128-Bit, DVI/VIVO, AGP 8X,

DFI "LANPARTY NFII ULTRA B" nForce2 Ultra 400 Chipset Motherboard for AMD Socket A CPU

Western Digital Raptor 36.7GB 10,000RPM SATA Hard Drive, Model WD360GD, OEM Drive Only

Only problem is my PSU. I, like some others, have the Antec True430 PSU.

I was reading how I don't really need the adapter, but I've heard the 6600 GT does NOT have an alternate power cord that hooks straight up to the card itself; which from reading, if I dont have the 24 power connector than the video card essentially needs this connector to the card.

So, is there a card that has an alternate power supply, or is it actually on the card and I am missing it?