power supply replacement

On Thu, 08 Nov 2007 16:49:26 -0800, Robin wrote:

Came home the other nightto find everything quiet, too quiet because I'd
left the PC on downloading. Noone noticed anything but it is dead. The
PSU doesnt even flash the keyboard LED so I think it's a gonner. My old
one was a 600W PSU with a 120mm fan. It ran silently on a mates PC but
mine always seemed to have it howling. Was the load on it so high that
this caused the noise and premature failure.

Most have overload protection. Normally unplugging them for a minute and
plugging them back in will start them again. If it starts be sure and make
sure the fan(s) are running. If not then check the fans being stuck. This
will also cause the PSU not to start.

What size do I need.....

The bigger the better IMO. Less loads on a PSu mean longer life in
general. I buy the biggest rated I can get for under $25 for this machine.
Last one was a 680W running Nvidia 6100 with 1 GB ram (2 sticks), 4 power
hungry HDTV tuner cards, X2 3800+ cpu, 5 drives and 5 fans. For the other
systems I just buy the cheapest I can find 500W or more under $20. The
680W and another 550W are Chiefmax. Been running a year or more without
problems. I expect them to last at least 5 years. Or you can go out and
pay more for one PSU than all my cases and PSU's together.:-)

--
Want the ultimate in free OTA SD/HDTV Recorder? http://mythtv.org
My Tivo Experience http://wesnewell.no-ip.com/tivo.htm
Tivo HD/S3 compared http://wesnewell.no-ip.com/mythtivo.htm
AMD cpu help http://wesnewell.no-ip.com/cpu.php

Came home the other nightto find everything quiet, too quiet because I'd
left the PC on downloading. Noone noticed anything but it is dead. The PSU
doesnt even flash the keyboard LED so I think it's a gonner.
My old one was a 600W PSU with a 120mm fan. It ran silently on a mates PC
but mine always seemed to have it howling. Was the load on it so high that
this caused the noise and premature failure.
What size do I need for an NF4 mobo, FX55 CPU, Arctic Cooling Freezer64 Pro,
4x 512MB DDR400 simms, ATI HD2900XT, 2 optical dvd drives and three HDDs of
various sizes. Sound is onboard so shouldnt effect it loadwise.
I did check a Wattage calculator but it said that 480W would do but I think
Ati recommend 550W for their vga card alone.
What do you recommend?

Robin

Thanks to Wes and Paul. Its got me thinking. I am just not sure wether to
buy a named brand at a silly price or a highish wattage generic PSU that
should be running well under its rating. Oh decisions decisions. HaHa.

Robin

Robin wrote:
Came home the other nightto find everything quiet, too quiet because I'd
left the PC on downloading. Noone noticed anything but it is dead. The PSU
doesnt even flash the keyboard LED so I think it's a gonner.
My old one was a 600W PSU with a 120mm fan. It ran silently on a mates PC
but mine always seemed to have it howling. Was the load on it so high that
this caused the noise and premature failure.
What size do I need for an NF4 mobo, FX55 CPU, Arctic Cooling Freezer64 Pro,
4x 512MB DDR400 simms, ATI HD2900XT, 2 optical dvd drives and three HDDs of
various sizes. Sound is onboard so shouldnt effect it loadwise.
I did check a Wattage calculator but it said that 480W would do but I think
Ati recommend 550W for their vga card alone.
What do you recommend?

Robin



I cannot find a power measurement for HD 2900 XT on Xbitlabs. I thought it
was about 165W or so (mainly from +12V), but I could be wrong. This article
gives some total system power measurements, and under load, the full computer
draws 312 watts.

http://techreport.com/articles.x/12458/15

Here they got a total of 363W. They estimate worst case as 215W for the
video card ? The manufacturer estimates tend to be on the high side.

http://www.extremetech.com/article2/...2128869,00.asp

I'm not sure I believe the power measurements in the last article.
Xbitlabs measures a 26W difference between a 8800GTX and 8800GTS, while
the Extremetech total system power difference is 36W. Ah, maybe some of
that is the power supply efficiency (12W more waste heat from PSU) ?

They mention "over 160W" here.
http://www.xbitlabs.com/articles/vid...-1024mb_5.html

OK, I found a measurement of the card here, as 161W. 155.8W on the 12V rail.
The only problem with a measurement like this, is they only measure one
card, and there will be statistical variation amongst them.

http://www.xbitlabs.com/articles/vid...ecture_14.html
http://www.xbitlabs.com/images/video...power_full.gif

So now I don't know what to use for the video card...

*******

FX-55 is 104W. 104W/12V * (1/0.90) = 9.6 amps assuming a 90% efficient Vcore circuit.

http://products.amd.com/en-us/Deskto...ail.aspx?id=25

If we took 156w for video, 156W/12V = 13 amps.

Add in another 5 amps for storage devices and fans, the total is 28 amps or so.

This one offers 12V @ 49A and is $120 USD.
Silencer 610 EPS12V Technical Specifications
http://www.pcpower.com/products/view...view=techspecs

Silverstone SST-ST56ZF ATX12V / EPS12V 560Watts Power Supply 90V ~ 264V (Auto Range) $140 USD
+3.3 @ 38A, +5V @ 40A, +12 @ 38A
http://www.newegg.com/Product/Produc...82E16817163111

I like the single rail supplies, because they eliminate all the guess work.
If I had your system, I'd buy the Silencer if you could find it.

The wattage is not too much of an issue, and it is mainly an issue
of getting enough amps.

If you go with a dual rail supply, then 12V2 10 amps and 12V1 18 amps
would be bare minimum. A few amps margin doesn't hurt. It also
helps if the supply is high efficiency, at least in terms of the heat
that the power supply itself has to dissipate. Supplies with unstated
efficiency, will be around 68% efficient, while good supplies are now
in the 80% to 85% range. You pay more for the 80%+ designs.

This might be as small as I'd go in a dual rail type supply. The 12V
rails meet the 12V2 10 amps and 12V1 18 amps numbers.

ENERMAX EG495AX-VE FMA ATX12V Ver 2.2 485W Power Supply 100 - 240 V $80 USD
+3.3V @ 32A, +5V @ 32A, +12V1 @ 22A, +12V2 @ 22A, , +5VSB@3A
http://www.newegg.com/Product/Produc...82E16817194015

Note that the total power rating on 12V for that Enermax, works out to a total
of just 32 amps (as shown in the picture in the next link). Your load is
28 amps total, so it just fits nicely. There is no chance you'll make a dent
in the 170W combined limit for 3.3 and 5V consumption, and your combined
I would estimate at around 80W or less for them.

http://images10.newegg.com/NeweggIma...194-015-04.jpg

If you wanted both a 2x3 PCI Express connector and a 2x4 PCI Express connector,
that might be a bit harder to find. I've seen some supplies with quad PCI
Express connectors, that have both 2x3 and 2x4 connectors, but I don't recollect
a dual PCI Express power supply with one of each. It would probably take
a bit more searching, to find such a combination of connectors. I'm sure
that some must exist somewhere.

I think your card will function with two 2x3 connected to it, but since you
own the card, you know more about it than I do. One of the articles I glanced
at, mentioned that the ATI software only allows overclocking, if a 2x4 is
detected as being plugged in. Why you'd want to overclock that card, and
make it run even hotter and louder, escapes me :-) So maybe a pair of
2x3 connectors is the best option, when selecting the supply.

Hope that gives you a few ideas,
Paul

On Nov 9, 1:03 am, "Robin" wrote:
Thanks to Wes and Paul. Its got me thinking. I am just not sure wether to
buy a named brand at a silly price or a highish wattage generic PSU that
should be running well under its rating. Oh decisions decisions. HaHa.

More wattage does not mean a more reliable power supply - not even
close. But that myth is often heard when on never bothered to first
learn the technology. A 600 watt power supply should be grossly more
than enough for any computer. Furthermore, many 'discount' supplies
(that also tend to fail faster due to missing functions), may actually
be only 420 watts. They did not lie. They just provided a different
number to a market where knowledge comes only from speculation.

Meanwhile, the only useful load numbers are currents for each
voltage. Overall power supply wattage numbers says little that is
useful. And then the 600 watts may only be 420 watts - just to make
speculation even less useful.

As demonstrated from wattage numbers, those calculators cannot
provide useful information. Only way to learn what has failed is to
take measurements from what exists. That means not even disconnecting
any wires. These numbers are required to obtain a useful reply. Even
swapping power supplies can only provided 'maybe' answers. This two
minute procedure is posted in "When your computer dies without
warning....." starting 6 Feb 2007 in the newsgroup alt.windows-xp
at:
http://tinyurl.com/yvf9vh

In your case, especially important are numbers from the green, gray,
and purple wires both before and when power switch is pressed.

No load on a minimally standard power supply will cause a supply
failure. That was standard long before PCs existed. In fact, all
output wires from a power supply must be shorted together - and still
no damage can result. Another standard for power supplies known by
those who first learned the technology. Intel specs even list the
wire size (or larger) to short out any supply - without damage.

A power supply that includes all required functions would cost at
least $60 full retail. However, America is full of computer experts
who would not even know the load (computer) cannot harm any standard
supply. Many supplies missing essential functions are dumped into a
market chock full of the technically naive.

Nothing in a brand name or that $60 figure says a power supply is
even sufficient. But among those products are something far more
important - numeric specs. To dump supplies that are missing
essential functions, the manufacturer must not claim to include those
required functions; must not provide a long list of numeric specs.
For example, if a shorted power supply might fail, then provide no
specs. When it fails shorted, well, it did not even claim to do what
all power supplies did even 35 years ago.

Provided is a two minute procedure to get answers. No other post
has provided a solution for definitive answers. You need only
'definitive' answers; not speculation. Post those numbers here to
obtain additional useful information. Your replies will only be as
good as the information you provide. That means numbers. It is only
a two minute procedure.

thanks Wes for your suggestion of retrying the PSU. I had previously tried
everything but after stripping it and checking for blown fuses and whatever
but there were none. Obvious thought was then to try a different lead in
case of blown lead fuse or faulty plug and hey presto it switched on. All
is well in the house again.

Thanks

Robin

"Robin" wrote in message
...
thanks Wes for your suggestion of retrying the PSU. I had previously
tried
everything but after stripping it and checking for blown fuses and
whatever
but there were none. Obvious thought was then to try a different lead in
case of blown lead fuse or faulty plug and hey presto it switched on. All
is well in the house again.

Thanks

Robin


While you were ratting around inside the power supply did you think to look
at the caps to see if any of them were leaking?
I have found a couple of power supplies with leaking caps in the last year
or so.

On Sun, 11 Nov 2007 10:52:44 -0800, Robin wrote:

There was plenty of fluff and dust but no sign of leaking caps. At
least it's working now.

One should clean the insides at least once a year. remove dust buildup
from fans and heatsinks. Check fans for freedom of movement. Fan bearings
will dry out often with original lubrications. Oiling them (1 small drop)
with a lightweight oil, like sewing machine oil, will extend their life
considerably. It will also fix any that fail to start because of dried
bearings. This is very common and most people just trash them.:-)

--
Want the ultimate in free OTA SD/HDTV Recorder? http://mythtv.org
My Tivo Experience http://wesnewell.no-ip.com/tivo.htm
Tivo HD/S3 compared http://wesnewell.no-ip.com/mythtivo.htm
AMD cpu help http://wesnewell.no-ip.com/cpu.php

On Nov 11, 1:52 pm, "Robin" wrote:
There was plenty of fluff and dust but no sign of leaking caps. At least
it's working now.

A properly constructed power supply needs no dusting every year if
operating in a standard room environment. Properly designed supply
must work just fine when heavily coated with dust. Fan bearings do
not need periodic oiling. Seized fans are more often traceable to
manufacturing defects such as the placement of a hall effect sensor
inside the fan. The resulting low torque will not be solved by more
oil. But many want to cure symptoms rather than first identify the
problem. Fans provided in minimally acceptable supplies should never
need oiling.

Defined in that other post are how to find a reliable supply that
includes essential functions. Also noted are why overall wattage
numbers say so little that is useful.

Some manufacturers forget to include essential functions, sell an
inferior supply using only two numbers - wattage and dollars.
Actually increase their profit margin on that lower priced supply by
'forgetting' parts. They can do this because so many computer
recommendations are without technical knowledge.

Long before caps appear to be leaking, the caps are already creating
detectable problems. That two minute procedure would have identified
the problem long before more system crashes. Most every electronic
failure leaves no visual indication. Caps could already be failing
and would be detected by that two minute procedure that identifies too
many problems to list.. Your current failure, if due to failing caps,
will only happen again.

Was the problem solved? A defective power supply can still power up
a computer. A common myth is a power supply is OK if the computer
powers. It could have failing caps, a partially shorted transformer,
a failed filter, etc. All those problems also detected by a two
minute procedure that is also mentioned in A+ Certified Technician
books.

You have no reason to believe anything was solved because the
problem was not identified. What may have been a safety lockout in
the power supply controller (not part of a power supply) may simply
happen again. Buy a new power supply and still have the same power
supply controller problem?

There was plenty of fluff and dust but no sign of leaking caps. At least
it's working now.

Robin