PC Review


Reply
Thread Tools Rating: Thread Rating: 3 votes, 5.00 average.

Single-rail vs. multi-rail power supplies?

 
 
Yousuf Khan
Guest
Posts: n/a
 
      21st Mar 2012
I currently have a desktop with a Zalman 600W PS, it was a pretty nice
PS a few years ago when I got it, but now it looks like the system has
been upgraded and grown again, and I'm starting to see signs that it's
not producing enough power for the components anymore. So I'm looking at
the market for bigger PS's, likely 750W+.

When I got the Zalman, multi-rail +12V were the de riguer feature of the
day. Now that I'm shopping for them again, I see that the manufacturers
have done an about-face, and they are advertising the advantages of
single-rail +12V. The old 600W had 4 rails at 16 Amps each. The new
single rails I'm seeing seem to have a rail at anywhere from 45A to 65A!
Why have the manufacturers done the about-face on single-rail vs.
multi-rail?

Yousuf Khan
 
Reply With Quote
 
 
 
 
Paul
Guest
Posts: n/a
 
      21st Mar 2012
Yousuf Khan wrote:
> I currently have a desktop with a Zalman 600W PS, it was a pretty nice
> PS a few years ago when I got it, but now it looks like the system has
> been upgraded and grown again, and I'm starting to see signs that it's
> not producing enough power for the components anymore. So I'm looking at
> the market for bigger PS's, likely 750W+.
>
> When I got the Zalman, multi-rail +12V were the de riguer feature of the
> day. Now that I'm shopping for them again, I see that the manufacturers
> have done an about-face, and they are advertising the advantages of
> single-rail +12V. The old 600W had 4 rails at 16 Amps each. The new
> single rails I'm seeing seem to have a rail at anywhere from 45A to 65A!
> Why have the manufacturers done the about-face on single-rail vs.
> multi-rail?
>
> Yousuf Khan


That's a good question.

The terminology is adjusted, to whatever sells power supplies.
That's a rule of thumb about marketing. You don't actually
have to be honest about what you're selling. You describe it
in any way, that will make you successful. if a competitor
touted single rail output, and he's selling supplies,
then you chime in with a matching story.

These are some examples of power supply designs.

AC ---- switcher --- 65A ---- 75A_limiter ----+---- 12V1
|
+---- 12V2
|
+---- 12V3

OK, let's take a Molex, and short +12V to ground on that one.
The whole yellow and black wires on the Molex, start to glow,
the plastic melts off the wire, and so on. While this is a
"true single rail" design, notice it's a safety hazard. The
design could provide around 12*75 = 900W of thermal energy,
if you adjusted the loading just right. If the short had a
resistance of 0.2 ohms, the supply likely wouldn't shut off.
A dead short, might trip it OK. Hard to guess...

Now, let's make a design closer to the intentions of the
IEC60950 spec, and limit individual outputs to a safer level.
This still isn't compliant, but its closer to the spirit
of the limitations you're supposed to apply to secondary outputs.

AC ---- switcher ----- 65A ---+-- 25A_limiter --------- 12V1
|
+-- 25A_limiter --------- 12V2
|
+-- 25A_limiter --------- 12V3

Now, the wires may still get hot, the connectors burn, but
the amount of power in any individual circuit, cannot exceed
12*25 = 300W. The limiter works, by switching off the supply,
if the current flow level is exceeded. An overload in any
branch, can switch it off.

Power supplies built as follows, are "true independent output"
circuits. A hint you're getting the real McCoy, is the chassis
is 1" to 1.5" longer than the competing power supply of the
same capacity. The density cannot be as great, because of
the replicated circuits. I don't think that many, were
actually built this way. If you open your Zalman, it probably
doesn't match this topology. Your Zalman, might be the
previous figure.

AC -- DC ---+-- switcher ----- 22A ------ 25A_limiter --- 12V1
300V |
+-- switcher ----- 22A ------ 25A_limiter --- 12V2
|
+-- switcher ----- 22A ------ 25A_limiter --- 12V2

I'd say that middle design, looks pretty good. No matter
whether my marketing department describes it as "one rail"
or "multi rail". The size of the supply, or the lack of
replicated circuits inside, can help you judge the
true topology.

I don't own a copy of IEC60950, so I can't copy/paste
the appropriate part. And the last time I looked, I couldn't
find a copy "floating" on the web. The keyword "SELV" comes
to mind, but that's about all I remember now.

HTH,
Paul



 
Reply With Quote
 
 
 
 
Rod Speed
Guest
Posts: n/a
 
      21st Mar 2012
Yousuf Khan wrote

> I currently have a desktop with a Zalman 600W PS, it was a pretty nice PS a few years ago when I got it, but now it
> looks like the system has been upgraded and grown again, and I'm starting to see signs that it's not producing enough
> power for the components anymore.


What signs are those exactly ?

That wouldnt normally be the case unless you have one hell of a video card in it.

> So I'm looking at the market for bigger PS's, likely 750W+.


> When I got the Zalman, multi-rail +12V were the de riguer feature of the day. Now that I'm shopping for them again, I
> see that the
> manufacturers have done an about-face, and they are advertising the
> advantages of single-rail +12V. The old 600W had 4 rails at 16 Amps each.


And I would be amazed if you are exceeding that.

> The new single rails I'm seeing seem to have a rail at anywhere from 45A to 65A!


And even the biggest is only 1A more than you currently have.

> Why have the manufacturers done the about-face on single-rail vs. multi-rail?


So they dont have to have 4 separate 12V regulators.



 
Reply With Quote
 
Rob
Guest
Posts: n/a
 
      21st Mar 2012
On 22/03/2012 8:52 AM, Yousuf Khan wrote:
> I currently have a desktop with a Zalman 600W PS, it was a pretty nice
> PS a few years ago when I got it, but now it looks like the system has
> been upgraded and grown again, and I'm starting to see signs that it's
> not producing enough power for the components anymore. So I'm looking at
> the market for bigger PS's, likely 750W+.
>
> When I got the Zalman, multi-rail +12V were the de riguer feature of the
> day. Now that I'm shopping for them again, I see that the manufacturers
> have done an about-face, and they are advertising the advantages of
> single-rail +12V. The old 600W had 4 rails at 16 Amps each. The new
> single rails I'm seeing seem to have a rail at anywhere from 45A to 65A!
> Why have the manufacturers done the about-face on single-rail vs.
> multi-rail?
>
> Yousuf Khan



Well get yourself one that has good weight and a three year warranty.

Bigger is not necessarily any better and depends on your PC configuration.
 
Reply With Quote
 
Yousuf Khan
Guest
Posts: n/a
 
      22nd Mar 2012
On 21/03/2012 6:34 PM, Rod Speed wrote:
> Yousuf Khan wrote
>
>> I currently have a desktop with a Zalman 600W PS, it was a pretty nice PS a few years ago when I got it, but now it
>> looks like the system has been upgraded and grown again, and I'm starting to see signs that it's not producing enough
>> power for the components anymore.

>
> What signs are those exactly ?


Well, we've discussed those on csiphs already, mainly the optical drives
sending controller error messages even when they are not being used, and
I'm also noticing some occasional spin retry errors on a few of my
internal HDD's.

> That wouldnt normally be the case unless you have one hell of a video card in it.


Well, the video card is not a monster of any kind, it's more upper
mainstream, an AMD Radeon 6870.

>> So I'm looking at the market for bigger PS's, likely 750W+.

>
>> When I got the Zalman, multi-rail +12V were the de riguer feature of the day. Now that I'm shopping for them again, I
>> see that the
>> manufacturers have done an about-face, and they are advertising the
>> advantages of single-rail +12V. The old 600W had 4 rails at 16 Amps each.

>
> And I would be amazed if you are exceeding that.


Well, I ran my system components through a few power supply calculators:

http://extreme.outervision.com/PSUEngine said I needed 580W.
http://www.thermaltake.outervision.com/Power said I needed 574W.
http://support.asus.com/powersupply.aspx said I needed 900W!!!

Obviously, the Asus rating is an outlier so I'm ignoring that one, but
the other two seem to agree pretty close to each other, and that level
is pretty close to the maximum power rating (within 96-97%) of my old
PS. And with the age of the PS advancing in years, its maximum capacity
might actually be decreasing over time.

When I first obtained this 600W PS, the PS calculator showed that my
system requirement was only in the upper 400W range. So I thought I had
more than enough leeway, but it looks like the various upgrades have
quickly taken up most of that leeway since then.

>> The new single rails I'm seeing seem to have a rail at anywhere from 45A to 65A!

>
> And even the biggest is only 1A more than you currently have.


When you add it up, yes the 4 separate rails come out to about 64A, but
the point is is it possible that some of the rails are overloaded by
themselves? Would it be easier to distribute the power if there was a
single larger rail rather than 4 smaller rails?

>> Why have the manufacturers done the about-face on single-rail vs. multi-rail?

>
> So they dont have to have 4 separate 12V regulators.


I found one possible explanation here:

http://www.overclock.net/t/88626/inf...iple-12v-rails

According to the above, the EU had mandated that no single +12V rail
could exceed 20A, and that later Intel also embraced that EU rule, and
further reduced that down to 18A. But now it looks like Intel no longer
requires the 18A rule. I'm not sure if the EU has also dropped its 20A
requirement, but here in North America there is no longer a need to have
separate rails anymore, so we can go to single super-rails.

Yousuf Khan
 
Reply With Quote
 
Paul
Guest
Posts: n/a
 
      22nd Mar 2012
Yousuf Khan wrote:

>
> Well, I ran my system components through a few power supply calculators:
>
> http://extreme.outervision.com/PSUEngine said I needed 580W.
> http://www.thermaltake.outervision.com/Power said I needed 574W.
> http://support.asus.com/powersupply.aspx said I needed 900W!!!
>
> Obviously, the Asus rating is an outlier so I'm ignoring that one, but
> the other two seem to agree pretty close to each other, and that level
> is pretty close to the maximum power rating (within 96-97%) of my old
> PS. And with the age of the PS advancing in years, its maximum capacity
> might actually be decreasing over time.
>
> When I first obtained this 600W PS, the PS calculator showed that my
> system requirement was only in the upper 400W range. So I thought I had
> more than enough leeway, but it looks like the various upgrades have
> quickly taken up most of that leeway since then.


The HD6870 is 124 to 147 watts. One of the cards quoted here could be
an overclock. Try doing your power calculation manually.

http://www.xbitlabs.com/articles/gra...u_4.html#sect0

Paul
 
Reply With Quote
 
KR
Guest
Posts: n/a
 
      22nd Mar 2012
On Mar 22, 7:52*am, Yousuf Khan <(E-Mail Removed)> wrote:
> I currently have a desktop with a Zalman 600W PS, it was a pretty nice
> PS a few years ago when I got it, but now it looks like the system has
> been upgraded and grown again, and I'm starting to see signs that it's
> not producing enough power for the components anymore. So I'm looking at
> the market for bigger PS's, likely 750W+.
>
> When I got the Zalman, multi-rail +12V were the de riguer feature of the
> day. Now that I'm shopping for them again, I see that the manufacturers
> have done an about-face, and they are advertising the advantages of
> single-rail +12V. The old 600W had 4 rails at 16 Amps each. The new
> single rails I'm seeing seem to have a rail at anywhere from 45A to 65A!
> Why have the manufacturers done the about-face on single-rail vs.
> multi-rail?
>
> * * * * Yousuf Khan




Traditionally it was always considered good practice to separate the
power for
digital and analog circuits as much as possible to prevent
interference.
This would especially be the case in audio systems.

In the case of a PC, this would mean keeping the supply for motors
(such as
Hard Drives, CD ROMS etc on a separate circuit, so as to minimise any
electrical
noise from these from interfering with digital circuits - assuming
this really is a problem in a modern PC.

Usually having separate supply cables from a common power source is
sufficient.



I would suggest it is more likely done to keep the currents manageable
on particular circuits however.
You might have 60A available, but in cases like this where one circuit
is never going to need more than (say) 15a,
it is safer to split it into 4 separate circuits.

Exactly the same is done in your home fusebox in having separate fuses
and circuits for your lights, power points,
air con, water heater, stove etc, rather than just having a single
200A fuse for the entire lot.







 
Reply With Quote
 
Rod Speed
Guest
Posts: n/a
 
      22nd Mar 2012
Yousuf Khan wrote
> Rod Speed wrote
>> Yousuf Khan wrote


>>> I currently have a desktop with a Zalman 600W PS, it was a pretty
>>> nice PS a few years ago when I got it, but now it looks like the
>>> system has been upgraded and grown again, and I'm starting to see
>>> signs that it's not producing enough power for the components anymore.


>> What signs are those exactly ?


> Well, we've discussed those on csiphs already, mainly the optical drives sending controller error messages even when
> they are not being used,


OK, I dont believe that those would be because you
are exceeding the 600W rating of the power supply.

I meant that the power supply may be failing, putting more noise
on the rails than is allowed. If thats the case, just replacing it with
another copy of the 600W supply should see that problem go away.

> and I'm also noticing some occasional spin retry errors on a few of my internal HDD's.


Thats unlikely to be because doesnt have enough current on the
12V rail even tho you do have quite a few hard drives from memory.

If that is the problem, it makes more sense ot replace some
of the smaller drives with new much larger ones than it does
to change the power supply, tho you should change the power
supply because of the optical drive symptoms you are getting
to see if thats due to the power supply.

You dont necessarily need such a big supply for that test tho.

>> That wouldnt normally be the case unless you have one hell of a video card in it.


> Well, the video card is not a monster of any kind, it's more upper mainstream, an AMD Radeon 6870.


Yeah, it only need 2 75W power connectors.

The 12V rails on that power supply are way above what it needs.

>>> So I'm looking at the market for bigger PS's, likely 750W+.


>>> When I got the Zalman, multi-rail +12V were the de riguer feature
>>> of the day. Now that I'm shopping for them again, I see that the
>>> manufacturers have done an about-face, and they are advertising the
>>> advantages of single-rail +12V. The old 600W had 4 rails at 16 Amps each.


>> And I would be amazed if you are exceeding that.


> Well, I ran my system components through a few power supply calculators:


I dont buy those. What matters is AMD's statement of what the video card needs,
http://www.amd.com/us/products/deskt...verview.aspx#3
and thats where that 2 75W power connectors comes from,
and that detail you have already provided on 4 12V 16A rails.

Those are 192W rails, each one of them.

> http://extreme.outervision.com/PSUEngine said I needed 580W.
> http://www.thermaltake.outervision.com/Power said I needed 574W.
> http://support.asus.com/powersupply.aspx said I needed 900W!!!


Yeah, thats obviously silly.

> Obviously, the Asus rating is an outlier so I'm ignoring that one, but the other two seem to agree pretty close to
> each other,


But nothing like what you get when you calculate the 12V rails explicitly.

That power supply handles the video card fine with 2 of the rails
and you have two more for your hard drives which wont take
anything like 384W even if they are all trying to spin up at once.

> and that level is pretty close to the maximum power rating (within 96-97%) of my old PS.


Thats the problem with those power supply 'calculators', they
dont actually calculate what matters, the 12V rail currents.

And the 900W supply you are considering only has 1A more
12V current available anyway.

> And with the age of the PS advancing in years, its maximum
> capacity might actually be decreasing over time.


Nope, that doesnt happen.

What you can get is a deterioration of the low ESR caps and that
sees a lot more noise on the rails than there should be, but you dont
see a reduction in the rail current capacity and even if you did you
are nowhere near the maximum currents on any of those 12V rails.

> When I first obtained this 600W PS, the PS calculator showed that my system requirement was only in the upper 400W
> range.


Do you mean that the calculator has changed, or that
what you have in that system has changed that much ?

> So I thought I had more than enough leeway, but it looks like the various upgrades have quickly taken up most of that
> leeway since then.


Or the calculator has changed since then. What have you changed upgrade wise ?

>>> The new single rails I'm seeing seem to have a rail at anywhere from 45A to 65A!


>> And even the biggest is only 1A more than you currently have.


> When you add it up, yes the 4 separate rails come out to about 64A, but the point is is it possible that some of the
> rails are overloaded by themselves?


Nope, not with that particular video card.

> Would it be easier to distribute the power if there
> was a single larger rail rather than 4 smaller rails?


No, in fact its harder because you cant do a remote sense
so that the highest current rail is seeing 12V at the pins
without increasing what the other connects get at the pins.

Not that that matter much, the specs on the variation in the 12V rails is pretty wide.

And you have the other problem with a single rail too, limiting the
current to say 75A can still see a decent fire with some shorts.

You dont get that with 4 seperate 16A max current rails. 200W
isnt that bad as long as there is someone around to turn it off.

800W can be pretty spectacular in the very small space of a
single molex nylon connector.

>>> Why have the manufacturers done the about-face on single-rail vs. multi-rail?


>> So they dont have to have 4 separate 12V regulators.


> I found one possible explanation here:


> http://www.overclock.net/t/88626/inf...iple-12v-rails


> According to the above, the EU had mandated that no single +12V rail could exceed 20A, and that later Intel also
> embraced that EU rule, and further reduced that down to 18A. But now it looks like Intel no longer requires the 18A
> rule.


Yeah, if thats accurate, it likely is the reason for the change.

> I'm not sure if the EU has also dropped its 20A requirement, but here in North America there is no longer a need to
> have separate rails anymore, so we can go to single super-rails.


I dont like the idea of a single rail that can in theory deliver 800W
being able to do short circuit protection adequately myself.

Corse most of the time a short would just blow off quite literally but that doesnt always happen.


 
Reply With Quote
 
Yousuf Khan
Guest
Posts: n/a
 
      22nd Mar 2012
On 21/03/2012 10:38 PM, Paul wrote:
> The HD6870 is 124 to 147 watts. One of the cards quoted here could be
> an overclock. Try doing your power calculation manually.
>
> http://www.xbitlabs.com/articles/gra...u_4.html#sect0


No, mine is a Sapphire 6870, which might be pretty close to a reference
design.

Yousuf Khan
 
Reply With Quote
 
Yousuf Khan
Guest
Posts: n/a
 
      22nd Mar 2012
On 21/03/2012 11:51 PM, Rod Speed wrote:
> Yousuf Khan wrote
>> Well, we've discussed those on csiphs already, mainly the optical drives sending controller error messages even when
>> they are not being used,

>
> OK, I dont believe that those would be because you
> are exceeding the 600W rating of the power supply.


Maybe not the overall power supply rating, but maybe some of the
individual rails might be undercharged.

> I meant that the power supply may be failing, putting more noise
> on the rails than is allowed. If thats the case, just replacing it with
> another copy of the 600W supply should see that problem go away.
>
>> and I'm also noticing some occasional spin retry errors on a few of my internal HDD's.

>
> Thats unlikely to be because doesnt have enough current on the
> 12V rail even tho you do have quite a few hard drives from memory.


Well, I do have six internal hard drives right now, and one optical
drive (Blu-Ray burner). It's the number of error messages that I'm
seeing on the BR burner that's got me most worried, but also recently I
saw a worrying pop-up message from Hard Disk Sentinel that it is
predicting an imminent failure of my boot drive too. I don't know which
of the rails all of these drives are connected to, but if they are all
connected to the same rail (very likely) then they might be all sharing
current from a diminished resource. I wonder how much current each of
the optical and hard drives use?

>>> That wouldnt normally be the case unless you have one hell of a video card in it.

>
>> Well, the video card is not a monster of any kind, it's more upper mainstream, an AMD Radeon 6870.

>
> Yeah, it only need 2 75W power connectors.
>
> The 12V rails on that power supply are way above what it needs.


One or two rails would go to the motherboard, another one would go to
the video card, and the last rail would be left for all of the rest of
the peripherals. I've mostly had no problems with motherboard components
(CPU, RAM, PCI cards), no problems with the video card, but the rest of
the system is all supplied by one rail, such as drives and fans and
lights. I won't really notice any power problems with the fans or
lights, but the drives might be pretty sensitive.

> That power supply handles the video card fine with 2 of the rails
> and you have two more for your hard drives which wont take
> anything like 384W even if they are all trying to spin up at once.


I don't think the video card gets two whole rails to itself. The
motherboard 24-pin connector is one rail which would power the PCI/PCI-e
slots, chipset and RAM; and maybe it'll feed a few Watts to the CPU too.
Then another 6-pin plug would be a rail for the CPU alone, which also
plugs into the motherboard. Then a couple of video power connectors
would go into the video card, which would likely come from one rail by
itself. The video card would also receive some power from the
motherboard through the PCI-e slot. So the video card might have at most
maybe 1.5 rails for its use (partial motherboard rail & full video
rail). And the last rail for everything else in the system.

>> and that level is pretty close to the maximum power rating (within 96-97%) of my old PS.

>
> Thats the problem with those power supply 'calculators', they
> dont actually calculate what matters, the 12V rail currents.
>
> And the 900W supply you are considering only has 1A more
> 12V current available anyway.


Well, I'm not really considering a 900W PS, more likely a 750W one.

>> When I first obtained this 600W PS, the PS calculator showed that my system requirement was only in the upper 400W
>> range.

>
> Do you mean that the calculator has changed, or that
> what you have in that system has changed that much ?


Yup, the stuff in the system has changed that much. This system is in a
constant state of evolution, including the case itself. I upgraded from
a mid-tower with a capacity for only four 3.5" drives to one with six
3.5" drives, and then it quickly evolved to to fill up those additional
drive slots. Also there was a video card upgrade, and a CPU upgrade
along the way too.

>> When you add it up, yes the 4 separate rails come out to about 64A, but the point is is it possible that some of the
>> rails are overloaded by themselves?

>
> Nope, not with that particular video card.


I don't think the video card is the issue here at all! Just those drives
in combination with all of the other powered peripherals inside that
system. I played around with the figures in one of the PS calculators,
and I found out that it's assuming 13W per 7200-rpm SATA hard drive, 24W
per 7200-rpm IDE HDD, 29W per Blu-Ray burner, 34W per DVD burner.

Based on that I currently have 129W in internal drives alone (4 SATA
HDD, 2 IDE HDD, 1 BR). When I previously had the dual DVD burners rather
than the single Blu-Ray, I had 168W worth of drives! This is now
starting to explain why my two DVD burners failed simultaneously.

I also have 4x 250mm case fans on the system (they came with the new
case). Each is regular fan is rated at 12W, and each LED fan is 13W.
Based on that I have 49W worth of fans (3 regular, 1 LED). Adding to the
previous figures of 129W and 168W brings them to overall totals of 178W
and 217W, respectively! If I only have 192W to play with per rail, then
I was well over when I had two DVD burners, and I'm sitting on the edge
still now.

>> Would it be easier to distribute the power if there
>> was a single larger rail rather than 4 smaller rails?

>
> No, in fact its harder because you cant do a remote sense
> so that the highest current rail is seeing 12V at the pins
> without increasing what the other connects get at the pins.
>
> Not that that matter much, the specs on the variation in the 12V rails is pretty wide.
>
> And you have the other problem with a single rail too, limiting the
> current to say 75A can still see a decent fire with some shorts.


That is obviously a worry, and that's why they didn't do this in the
past. I'm hoping that now that they are doing it, that they may have
found a way to keep it under control these days?

Yousuf Khan
 
Reply With Quote
 
 
 
Reply

Thread Tools
Rate This Thread
Rate This Thread:

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Low power software + high power hardware = high power dissipation? blackhead DIY PC 8 15th Feb 2008 01:51 AM
Permament disable for power options on press Power and Sleep button to Power Off or Standby teo9799 Microsoft Windows 2000 1 21st Sep 2004 04:01 PM
power scheme power management =?Utf-8?B?T21hcg==?= Windows XP Help 0 21st Jun 2004 11:52 PM
Power Management (Power Schemes) James B. Holladay Windows XP Help 0 27th Apr 2004 05:06 PM
No power from Power Supply sbowler Support 5 20th Nov 2003 10:00 PM


Features
 

Advertising
 

Newsgroups
 


All times are GMT +1. The time now is 05:41 PM.