Power supplies?

[Yousuf Khan]

It's possibly time to upgrade my power supply. Haven't shopped around
for one in ages. What's the new features that I have to look for these
days? My last PS was a 450W ATX from maybe 5 years ago. I expect my new
one would have to be at least 500W now.

I'm thinking I should get one with the low-speed 120 mm fans. And maybe
one with a motherboard fan speed connector?

Yousuf Khan

 

[Jack]

Yousuf Khan wrote:
:: It's possibly time to upgrade my power supply. Haven't shopped
:: around for one in ages. What's the new features that I have to
:: look for these days? My last PS was a 450W ATX from maybe 5 years
:: ago. I expect my new one would have to be at least 500W now.
::
:: I'm thinking I should get one with the low-speed 120 mm fans. And
:: maybe one with a motherboard fan speed connector?

For sure, Yousuf, if I was still living in the States I would be buying my
PS from PC Power & Cooling. Check out the link:
http://www.pcpower.com/index.html

Geez, a PS from five years ago? Was this the PS involved with your computer
that you posted problems with SATA drives, Nvidia chipsets and corruption?
Hope not! Cheers.

J.

 

[Yousuf Khan]

For sure, Yousuf, if I was still living in the States I would be buying my
PS from PC Power & Cooling. Check out the link:
http://www.pcpower.com/index.html

Well, okay, I'll note that down.

Know anything about Zalman PS's?

Geez, a PS from five years ago? Was this the PS involved with your computer
that you posted problems with SATA drives, Nvidia chipsets and corruption?
Hope not! Cheers.

Well, yes it's the same machine, since one of the suggestions was to
replace the PS, I've decided to give that solution a shot.

Looking at the specs these days, they seem to have multi-rail +12V
support. They range from dual-rail upto quad-rail. Not sure whether to
go with more or if two is enough?

Yousuf Khan

 

[Jack]

Yousuf Khan wrote:
:: Jack wrote:
::: For sure, Yousuf, if I was still living in the States I would be
::: buying my PS from PC Power & Cooling. Check out the link:
::: http://www.pcpower.com/index.html
::
:: Well, okay, I'll note that down.
::
:: Know anything about Zalman PS's?

Never used one personally but friends who have built their own systems and
installed Zalman seem to be happy with them. They have a good rep here in
Spain.

::: Geez, a PS from five years ago? Was this the PS involved with
::: your computer that you posted problems with SATA drives, Nvidia
::: chipsets and corruption? Hope not! Cheers.
::
::
:: Well, yes it's the same machine, since one of the suggestions was
:: to replace the PS, I've decided to give that solution a shot.

With the problems you posted and now the fact you're using a five year old
PS, that's exactly what I'd do too.

:: Looking at the specs these days, they seem to have multi-rail +12V
:: support. They range from dual-rail upto quad-rail. Not sure
:: whether to go with more or if two is enough?

Hmmm. Read up on PC Power & Cooling's "Power Supply Myths Exposed." They
seem to indicate multirail designs have at least one major drawback. Maybe
DAYTRIPPER would jump in here???
http://www.pcpower.com/technology/myths/

J.

 

[daytripper]

Yousuf Khan wrote:
:: Jack wrote:
::: For sure, Yousuf, if I was still living in the States I would be
::: buying my PS from PC Power & Cooling. Check out the link:
::: http://www.pcpower.com/index.html
::
:: Well, okay, I'll note that down.
::
:: Know anything about Zalman PS's?

Never used one personally but friends who have built their own systems and
installed Zalman seem to be happy with them. They have a good rep here in
Spain.

::: Geez, a PS from five years ago? Was this the PS involved with
::: your computer that you posted problems with SATA drives, Nvidia
::: chipsets and corruption? Hope not! Cheers.
::
::
:: Well, yes it's the same machine, since one of the suggestions was
:: to replace the PS, I've decided to give that solution a shot.

With the problems you posted and now the fact you're using a five year old
PS, that's exactly what I'd do too.

:: Looking at the specs these days, they seem to have multi-rail +12V
:: support. They range from dual-rail upto quad-rail. Not sure
:: whether to go with more or if two is enough?

Hmmm. Read up on PC Power & Cooling's "Power Supply Myths Exposed." They
seem to indicate multirail designs have at least one major drawback. Maybe
DAYTRIPPER would jump in here???
http://www.pcpower.com/technology/myths/

J.

Not with both feet - too close to a religion discussion for me ;-)

That PC P&C page is certainly as self-serving as possible, but there are
utility tradeoffs in isolation vs flexibility between the two designs for
which the user must account, regardless of any price advantage.

In any case, at the end of the day, proper sizing for the intended
configuration, and user-provided reliability history gleaned from any number
of forums, are much more important than any implementation differences.

/daytripper (That - plus blue leds ;-)

 

[YKhan]

:: Know anything about Zalman PS's?

Never used one personally but friends who have built their own systems and
installed Zalman seem to be happy with them.  They have a good rep herein
Spain.

Well, there is a deal on a refurbed Zalman PS that I've seen.

::: Geez, a PS from five years ago?  Was this the PS involved with
::: your computer that you posted problems with SATA drives, Nvidia
::: chipsets and corruption? Hope not!  Cheers.
::
::
:: Well, yes it's the same machine, since one of the suggestions was
:: to replace the PS, I've decided to give that solution a shot.

With the problems you posted and now the fact you're using a five year old
PS, that's exactly what I'd do too.

I usually never replace my PS unless I get a new one with a new case
or if the PS blows up due to an electrical storm or something. The one
I have currently was relatively high-end for its time, 420W. It's gone
through two processor and two GPU upgrades without any problems. So it
seemed a little random that a simple hard drive addition should push
it over the edge when previous processor and video upgrades didn't
faze it.

:: Looking at the specs these days, they seem to have multi-rail +12V
:: support. They range from dual-rail upto quad-rail. Not sure
:: whether to go with more or if two is enough?

Hmmm.  Read up on PC Power & Cooling's "Power Supply Myths Exposed."  They
seem to indicate multirail designs have at least one major drawback.  Maybe
DAYTRIPPER would jump in here???http://www.pcpower.com/technology/myths/

I'll check it out.

Yousuf Khan

 

[YKhan]

Not with both feet - too close to a religion discussion for me ;-)

That PC P&C page is certainly as self-serving as possible, but there are
utility tradeoffs in isolation vs flexibility between the two designs for
which the user must account, regardless of any price advantage.

Sure seemed like it. Are these excuses for their products not having
all of the features these days? They're railing against multi-rail
power supplies (pun), but doesn't the current ATX v2.x power supply
specs require you to have at least two rails by default?

In any case, at the end of the day, proper sizing for the intended
configuration, and user-provided reliability history gleaned from any number
of forums, are much more important than any implementation differences.

/daytripper (That - plus blue leds ;-)

All of the big high-end ones seem to have the blue LEDs these days.

Yousuf Khan

 

[daytripper]

Sure seemed like it. Are these excuses for their products not having
all of the features these days? They're railing against multi-rail
power supplies (pun), but doesn't the current ATX v2.x power supply
specs require you to have at least two rails by default?


All of the big high-end ones seem to have the blue LEDs these days.

Yousuf Khan

In spite of some rather mushy "should" statements early in the spec, a strong
case could be made that the whole of ATX12V spec v2.2 absolutely requires more
than one 12V rail - for capacity in excess of 18A. That 18A value is driven by
the *requirement* of meeting UL & CSA safety 240VA regs:

3.2.4. Power Limit / Hazardous Energy Levels
"Under normal or overload conditions, no output shall continuously provide
more than 240 VA under any conditions of load including output short circuit,
per the requirement of UL 1950/CSA 950 / EN 60950/IEC 950."

The requirement to meet UL 1950/CSA 950 / EN 60950/IEC 950 is also provided in
section 3.1.4 "Regulatory".

fwiw, I've been using Rosewill supplies of late, particularly their
RP600V2-S-SL 600W supply - which is a dual-12V rail design. I've built three
well-loaded systems with these in the last couple of years with zero issues
and think they provide good value...

/daytripper

 

[Yousuf Khan]

fwiw, I've been using Rosewill supplies of late, particularly their
RP600V2-S-SL 600W supply - which is a dual-12V rail design. I've built three
well-loaded systems with these in the last couple of years with zero issues
and think they provide good value...

That previous PC P&C "myths" article mentioned that having unused +12V
rails would be a problem, as power supposedly gets trapped in the unused
rails creating efficiency losses. Any truth in this stuff? I can't see
how this might happen myself, a capacitor might get filled, and once
it's filled no more electricity goes into it until it gets utilized.

Yousuf Khan

 

[daytripper]

That previous PC P&C "myths" article mentioned that having unused +12V
rails would be a problem, as power supposedly gets trapped in the unused
rails creating efficiency losses. Any truth in this stuff? I can't see
how this might happen myself, a capacitor might get filled, and once
it's filled no more electricity goes into it until it gets utilized.

Yousuf Khan

It's nothing that involved - there's no "trapped" electrons, and there's no
real effect on "efficiency" in any classic sense. That whole spiel simply
points out the effect of "partitioning" 12V capacity into multiple isolated
rails, verses having all 12V capacity on a single rail.

For instance, say you have a single-rail supply with 36A of 12V capacity, and
another supply with 2 12V rails rated at 18A each. Now build a system that
requires 12A of 12V for the motherboard and peripherals, plus a pair of 12A
graphics cards.

Both supplies are rated at 36A of 12V, but hooking up the dual-rail supply is
probably going to be more challenging than the single rail supply. In fact, it
probably could not be done - hence you would have unusable ("trapped")
capacity in the dual-rail supply...

/daytripper

 

[YKhan]

It's nothing that involved - there's no "trapped" electrons, and there's no
real effect on "efficiency" in any classic sense. That whole spiel simply
points out the effect of "partitioning" 12V capacity into multiple isolated
rails, verses having all 12V capacity on a single rail.

And you mentioned that UL & CSA safety requirements don't allow for
more than 240VA, which would mean a maximum of 20A @ 12V. So I don't
see how they can avoid having multiple rails if they want more than
240VA on the system?

For instance, say you have a single-rail supply with 36A of 12V capacity,and
another supply with 2 12V rails rated at 18A each. Now build a system that
requires 12A of 12V for the motherboard and peripherals, plus a pair of 12A
graphics cards.

Both supplies are rated at 36A of 12V, but hooking up the dual-rail supply is
probably going to be more challenging than the single rail supply. In fact, it
probably could not be done - hence you would have unusable ("trapped")
capacity in the dual-rail supply...

But that's probably why they have triple rail power supplies.

Yousuf Khan

 

[daytripper]

And you mentioned that UL & CSA safety requirements don't allow for
more than 240VA, which would mean a maximum of 20A @ 12V. So I don't
see how they can avoid having multiple rails if they want more than
240VA on the system?


But that's probably why they have triple rail power supplies.

Yousuf Khan

And more, even. As we all know, "size matters" ;-)

Bottom line is you can't simply buy capacity - a box of watts - without
understanding potential partitioning effects as applied to a given system
build...

/daytripper

 

[Robert Redelmeier]

It's nothing that involved - there's no "trapped" electrons,
and there's no real effect on "efficiency" in any classic
sense. That whole spiel simply points out the effect of
"partitioning" 12V capacity into multiple isolated rails,
verses having all 12V capacity on a single rail.

For instance, say you have a single-rail supply with 36A
of 12V capacity, and another supply with 2 12V rails rated
at 18A each. Now build a system that requires 12A of 12V
for the motherboard and peripherals, plus a pair of 12A
graphics cards.

Both supplies are rated at 36A of 12V, but hooking up the
dual-rail supply is probably going to be more challenging
than the single rail supply. In fact, it probably could
not be done - hence you would have unusable ("trapped")
capacity in the dual-rail supply...

I thought there was a serious problem with underloading
(<5%) a switcher (or any given rail) in that the transistor
switch-on was very short and there were LC reactances that
made regulation difficult.

-- Robert

 

[YKhan]

And more, even. As we all know, "size matters" ;-)

Yeah, well the one I'm looking at is seriously blinged-out. Quad-rails
and blue LEDs and all.

::: Zalman, leading the world of Quiet Computing Solutions :::
http://www.zalman.co.kr/eng/product/Product_Read.asp?Idx=196

It's a refurbed unit though, so it's a good deal.

Yousuf Khan

 

[Yousuf Khan]

Why? I have seen power supplies at Fry's rated for 1 kW (!) and up, but how
much of this is actually needed by the average system and how much of it is
little more than stuff to enable dick-size contests among gamers? My
fastest system is a Core 2 Quad Q6600 with a couple of hard drives and a
9500GT; it runs just fine on a 380W power supply (Antec Earthwatts, if
you're curious). I've built servers with sixteen hard drives configured as
hardware RAID-5, and a 650W RPS is sufficient for that (staggered spin-up
helps here).

Getting a good-quality power supply from a reputable vendor is more
important than going for the biggest number. The high-efficiency power
supplies that have become available in the past year or two (look for "80
Plus" on the box), in addition to using less power for a given output, also
run cooler, which should help them last longer.

Well because I've been having some errors on my hard drives (another
thread on another newsgroup), especially when I added more hard drives
to the system. The drive problems ranged from the firmware puking its
guts out to pending sectors waiting to be remapped but never remapping.

My existing 420W unit was from long before the current ATX12V 2.x era.
So it seemed like a good place to start to resolve these issues. I
personally resisted the power-supply as the solution for a long time, as
it seems like the default solution to every problem these days. But I'm
getting the feeling it may have some validity in my case.

Yousuf Khan