How to know which rail in PSU powers what connections?

[64K]

Absolutely true 64K!

The point is the MUCH higher initial price versus other PSUs that were out which are just as good or better with a better warranty.

Yeah, I know. That is what I meant by my last sentence of my post. Guess I didn't word it well. A 400 watt PSU would have been plenty for his rig and put him in the peak efficiency range as well.

 

[cdawall]

Not something I recommend to anyone that is not experienced... it can kill you. Id sooner flip the thing to single rail mode (problem solved) or email/call the company and see what is what.

I will tell you I am not sold on the numbers corresponding with the rails though...

The rail numbers list as 70A max across all 4. The others are just OCP bull. I am not saying anyone should go rip a psu open, but if you want to end all argument that's how. PSU caps are tiny compared to what I am used to.

 

[RejZoR]

But we don't know that for sure. It is likely, but there are enough out there to make one wonder. Really though, all it takes is OCP on the rails not reaching the full output to make it a 'true' multi rail PSU
You still get OCP... its is not any more safe with multiple rails at this power level.

There is no OCP in single rail mode. You get other protections, but not OCP. The whole point of OCP is limitation of components on individual rails from drawing ridiculous amounts of amps. If it's a single 80A rail, a faulty component can draw all of it to itself and turn itself into supernova. But if the rail only has 25A, the disaster will most likely be way smaller because PSU will shut down at 25A current draw already. Didn't know that until recently.

 

[rtwjunkie]

There is no OCP in single rail mode. You get other protections, but not OCP. The whole point of OCP is limitation of components on individual rails from drawing ridiculous amounts of amps. If it's a single 80A rail, a faulty component can draw all of it to itself and turn itself into supernova. But if the rail only has 25A, the disaster will most likely be way smaller because PSU will shut down at 25A current draw already. Didn't know that until recently.

And there you've hit upon the primary benefit of multi-rail PSU's!

 

[EarthDog]

There is no OCP in single rail mode. You get other protections, but not OCP. The whole point of OCP is limitation of components on individual rails from drawing ridiculous amounts of amps. If it's a single 80A rail, a faulty component can draw all of it to itself and turn itself into supernova. But if the rail only has 25A, the disaster will most likely be way smaller because PSU will shut down at 25A current draw already. Didn't know that until recently.

Yes there is....Its a quality unit and will have OCP... just somewhere above the total output, not sliced up on the rails. You should still only lose the component that is going bad causing that load, not anything else... with a QUALITY PSU.

Jonnyguru's own Oklahoma Wolf has mentioned in multiple forums multiple times that there really isn't a worry about single rail PSUs until you get into the monster sized PSUs... and its only because of the sheer volume of output.

Let me find you a link.............

http://www.overclockers.com/forums/...swer-inside!?p=5492844&viewfull=1#post5492844

Ok... What's the bottom line?

The bottom line is, for 99% of the folks out there single vs. multiple +12V rails is a NON ISSUE. It's something that has been hyped up by marketing folks on BOTH SIDES of the fence. Too often we see mis-prioritized requests for PSU advice: Asking "what single +12V rail PSU should I get" when the person isn't even running SLI! Unless you're running a plethora of Peltiers in your machine, it should be a non-issue assuming that the PSU has all of the connectors your machine requires and there are no need for "splitters" (see Example 1 in the previous bullet point).

The criteria for buying a PSU should be:

• Does the PSU provide enough power for my machine?
• Does the PSU have all of the connectors I require (6-pin for high end PCIe, two 6-pin, four 6-pin or even the newer 8-pin PCIe connector)?
• If using SLI or Crossfire, is the unit SLI or Crossfire certified (doesn't matter if a PSU is certified for one or the other as long as it has the correct connectors. If it passed certification for EITHER that means it's been real world tested with dual graphics cards in a worst case scenario).

Yes... a bit old. But the point should be clear?

Actually, not sure that proved anything except its an old link, hahalololol.

The rail numbers list as 70A max across all 4. The others are just OCP bull. I am not saying anyone should go rip a psu open, but if you want to end all argument that's how. PSU caps are tiny compared to what I am used to.

You did just say that, actually.

 

[thebluebumblebee]

I still say, that with a modern, well built PSU such as this, just plug the plugs in and go. I know I've only gotten in trouble when I've used adapters. Each of those plugs have a rating, and I'm quite sure that if you loaded every single plug to it's design limit, that the PSU would just be fine.
Well, that's what I normally say, but not for this PSU. It comes with 6X 6+2 pin PCI-e connectors (150 watts each) plus an additional 6 pin PCI-e (75 watts). If you loaded them to their max, that's 975 watts!

 

[EarthDog]

I messaged Oklahoma Wolf...

Originally Posted by EarthDog
Im losing my marbles... I recall you mentioning that its better to have multi rails on large PSUs (1KW+?) but less than that its not as big of a deal. I looked at the stickies, tried my google foo, but came up short...can you refresh my memory on why that is?

Thanks in advance for your time!

It's just because the big watt monsters have so much current output capacity... some of them don't shut down soon enough if the fault is below their overpower protection threshold.

 

[newtekie1]

But we don't know that for sure. It is likely, but there are enough out there to make one wonder. Really though, all it takes is OCP on the rails not reaching the full output to make it a 'true' multi rail PSU

The fact is that it just doesn't make sense to make true multi-rail units anymore. So no one does it. It is rather expensive to include multiple 12v transformers in a unit, not to mention it makes then very heavy.

I think the Antec HCP is a true quad rail.

Not as far as I know it isn't.

 

[thebluebumblebee]

its better to have multi rails on large PSUs (1KW+?)

Because the amperage on a large, single rail PSU, can be dangerous and extremely harmful if a short occurs.

 

[EarthDog]

The fact is that it just doesn't make sense to make true multi-rail units anymore. So no one does it. It is rather expensive to include multiple 12v transformers in a unit, not to mention it makes then very heavy.



Not as far as I know it isn't.

Oh, don't misunderstand me, I agree with you...

And here's the label with the power specifications on it. Since each 12V rail is given overcurrent protection above 40A, you should have zero problems with this unit shutting down on you in any normal configuration. I'd test that number to see where the trip point is, but it's probably somewhere around 50A and I can't manage that without going to a lot of hassle. Since my time is very limited this week (seriously, all this soda isn't going to chug itself), I'll forgo that.

Antec HCP 850W ^^

Each rail cannot support the full output due to OCP.

 

[newtekie1]

Antec HCP 850W ^^

Each rail cannot support the full output due to OCP.

Yes, but it still isn't a true multi-rail design. It is a single 12v rail, with multiple OCP circuits split between different connectors.

 

[EarthDog]

Lol, minutia, but yes. Oklahoma wolf said railS, plural. I see what you are saying, but the reality is that because there is ocp on the 'rails' it acts like a multi rail to the user. So for all intents and purposes, it is a multi rail because of its behavior.

 

[95Viper]

From BeQuiet's lexicon:

The P8 plug is an extension of the P4 plug. It is equipped with eight contacts (pins), of which four provide 12V power supply and four are intended for earthing. The P8 plug is connected to the mainboard and is used for multi-core processors.

The P4 plug is an important connector for a power supply. The plug was introduced by Intel with the Pentium 4 and has two contacts (pins) each for 12V current and earthing. It was introduced so that the processors receive power from a separate, uninfluenced 12V rail.

The "rail splitting info" on page 41 of the manual, found at BeQuiet's site:

 

[RejZoR]

Excellent, you've answered exactly what I was asking Strangely, this schematic wasn't in the paper manual that arrived with the PSU.

So, my prediction was indeed correct. PCIe ports have the highest amp output rails and the weak rails are dedicated to less power hungry devices.

 

[Jetster]

Its doesn't matter

 

[RejZoR]

It matters if you're tripping the OCP and you have no clue about the actual power configuration. That's how I had to fiddle with connections on my LC Power 550W years ago when ASUS soundcard with molex was tripping the protection and I had no clue what to use apart from just blindly trial and error trying stuff, something that can be done in 10 seconds if you have the schematic...

 

[BiggieShady]

Really? So PCI-e connectors 1-4 line up with 12V1-4? So where do the P8/P4 and main 24 pin connectors get their power from? I'm not telling you that you are wrong, I'm just questioning your logic and pointing out the quandary of multi-rail PSU's.

Yeah, well I was very much wrong because 4 PCI-e connectors are on 2 stronger rails, and the rest is on weaker rails. I'm having more "I should have known that" moments lately.

 

[RejZoR]

Well, for me it wasn't entirely certain where CPU gets the power from. The strong rail or the weaker one...

 

[Jetster]

It matters if you're tripping the OCP and you have no clue about the actual power configuration.

I was referring this actual PSU.

 

[EarthDog]

It matters if you're tripping the OCP and you have no clue about the actual power configuration. That's how I had to fiddle with connections on my LC Power 550W years ago when ASUS soundcard with molex was tripping the protection and I had no clue what to use apart from just blindly trial and error trying stuff, something that can be done in 10 seconds if you have the schematic...

You would be hard pressed to do that with your setup. Very. Good to know, but no need to worry.

 

[Mussels]

Yeah, that's not very bright. You know the power efficiency curve is very bad at the lower end of the spectrum? You are wasting lots of energy there.

jumping in the thread just to correct that one as entirely untrue, go test the power consumption at a wall meter. you might lose 5% efficiency being at the low end, but hey - 5% of 100W is only 5W. i've got a 1000W PSU and it idles at 65W, and i'm definitely devastated at the 10W tops i could have saved with a 600W PSU.

 

[EarthDog]

Someone covered that already.

Where you WOULD save the cash is buying an appropriately sized PSU. For your rig for example, even with TWO r9 290's overclocked and the CPU, I would have bought an 850W PSU. Single r9 290.. 600W is PLENTY. That would have saved you at least $50 if not a lot more. I had a 295x2 and an overclocked 5820K running happily on a 750W PSU.

 

[Vayra86]

jumping in the thread just to correct that one as entirely untrue, go test the power consumption at a wall meter. you might lose 5% efficiency being at the low end, but hey - 5% of 100W is only 5W. i've got a 1000W PSU and it idles at 65W, and i'm definitely devastated at the 10W tops i could have saved with a 600W PSU.

The point is, many PC's are in idle the vast majority of the time. It adds up. In addition, most PSU's have a very hard efficiency falloff in the lowest part of the curve. The difference can be FAR greater than 5%. Recent PC's have a very low idle power draw, much lower than older PC's do, which further increases the power efficiency loss.

So, definitely not entirely untrue, and a factor of influence in PSU choice. Why would you pick a PSU that is inefficient for the vast majority of the time? Of course a PSU with a higher price suffers from lower falloff across the whole curve in general, but I don't think idle power draw should be ignored.

I don't like linking tom's, but this page says it all.

http://www.tomshardware.com/reviews/power-supply-psu-review,2916-8.html

Honestly the only reasons that remain for buying a PSU that goes far beyond your system's max power draw, is future plans for upgrades and silence.

 

[EarthDog]

Honestly the only reasons that remain for buying a PSU that goes far beyond your system's max power draw, is future plans for upgrades and silence.

Im with you here for the most part. About the silence thing, a lot of high end gold/plat level PSUs don't even turn on the fan until a certain load or temperature anyway, not to mention I don't EVER recall hearing a PSU fan over my system (which isn't loud in the first place) so there is little need to overbuy for silence.

But you really need to look at the REAL differences in idle power draw and that steep slope. If you work out the math, it really isn't much at all. I mean If you sit at 25W idle, and you are at 65% efficiency, with each PSU listed you are you are pulling about 42W from the wall from the crap PSU and around 34W as an average for the other PSUs. That is a difference of 7W. I pay 10 cents per KW/h... you do the math on that difference. While I am not saying it should be ignored, the monetary differences are damn near negligible if you take the time and do the math. And if you are that worried about it. Shut your PC off. Most PC's with SSD's are up and running in well under 30s anyway...

I personally run my PSU's around the 80% level of its capacity. I lose 1% or so by not running it around 50%, but I also didn't overbuy. That large amount of money I saved from getting a 750W PSU versus a 1KW PSU will NEVER be made up through that difference of efficiency UNLESS YOU RUN ON LOAD 24/7 LIKE WITH F@H or distributed platforms of the like. For most Intel setups, a quality 550W is PLENTY even with overclocking your GPU and CPU...assuming ambient cooling and no bios mods on GPUs. For AMD and its power sucking heat blowing FX octos, I would go 650W...this is single GPU for the record. Dual I bump that up to 750W/850W.

 

[newtekie1]

Lol, minutia, but yes. Oklahoma wolf said railS, plural. I see what you are saying, but the reality is that because there is ocp on the 'rails' it acts like a multi rail to the user. So for all intents and purposes, it is a multi rail because of its behavior.

Not exactly. The benefit of true multi-rail PSUs used to be that when you loaded one of the rails, it didn't affect the other rails. So when you fired up that beefy video card and it started sucking down the amps, the 12v rail it was on might drop to 11.5v. But the other rails would remain at say 12.1v. So, if you have your CPU overclocked, that overclock might be stable with the GPU idle because the CPU is receiving a clean 12.1v. But load the GPU on a single rail unit, voltage drops to 11.5v, and suddenly your CPU overclock is no longer stable. But on a true multi-rail PSU that wouldn't happen, the CPU would still be happily receiving 12v even if the GPU was dragging its rail down below 12v. This actually use to be pretty common back in the day, people would have stable overclocks on their GPU when the CPU was idle, and vise versa, but when you tried to load both one(or both) would be become unstable. This was actually the original reason why multi-rail PSUs became popular. Of course this was back in the day when even the best PSUs out there were not even as good as Corsairs CX line today. The PSUs we have now, at least the decent ones, all have good enough regulations that the single rail doesn't fluctuate that much under load, so this isn't really a problem anymore. Which is why they are all single rail designs now.