"Paul" wrote in message
...
Ray wrote:
Ray wrote:
I have a ASUS P5B-E mo, a GeForce 9500 GT video, 4 sata hard drives
(with
sata and Molex connectors) and 5 usb peripherals and need a power
supply
to
run it. The GF9500GT video card wants 18 amps so I am thinking of
getting
one of these supplies:
1. OXZ OCZ500MXSP
(skip)
there are better choices with about the same cost, what would you
suggest?

Thanks,
Ray

Opps, I made a mistake and wrote "-5v" when I meant -12v. It seems most
supplies have 0.8 amp for the -12v line. Should have typed "1. OCZ"
(not
OXZ) also.

I guess I'm dizzy from looking at so many specs and comments. I didn't
remember the ones for the CoolerMaster supplies. I did see the comments
for
the OCZ but they looked better than a lot of others I've seen. I had a
EnerMax 420W supply but it broke down after a few years and that is why
I
need a new supply. It also seemed to be somewhat under powered for my
system too. I have an Intel Core 2 Duo E6400 CPU. The GeForce 9500GT
(PCI-E but no separate power connector) spec sheet says a minimum of 350
W
power supply with a min recommended +12v current rating of 18 amps. so I
assume that means for that power rail that supports the mob. Isn't this
right?

I have been trying to find supplies with +12v rails supporting 18 amps
and
there don't seem to be many under $100. Any suggestions?

Ray

The value the video card advertising recommends, is based on some
assumptions
about the contents of your computer case. The assumptions may not be
valid.

Let's try some numbers.

E6400 = 65W. 65W/12V * 1/0.90 = 12V @ 6 amps from 12V2 rail, (90% Vcore
efficiency)


http://ark.intel.com/Product.aspx?id...ec-codes=SL9S9,
SL9T9,SL9TY,SLA97

The rest of the loads will be from 12V1.

(4) hard drives. While idle in desktop, 0.6 amps each.
Total 12V @ 2.4A from 12V1. When the hard drives spin up,
the current is 2.5A for each drive, or about 10A total.
When there are enough drives, a separate calculation should be
carried out. Four drives happens to be the limit, before a person
should calculate startup power requirements. Many supplies
can withstand an overage on current during startup, but if
you have enough hard drives, the startup current can crush the
supply, and OC shuts off the power before the computer can POST.
Overcurrent detection on an ATX supply, is likely disabled for the
first 35 milliseconds, but is enabled from then on.

(1) CDROM. About 12V @ 1.5A boiler plate rating. I've measured 12V @ 1.0A
while media was spinning at top speed. This is from 12V1. If you have
no media in the tray, the current is zero.

(1) 9500 GT. 12V @ 4A approx from 12V1.

(x) Cooling fans. Guess 12V @ 0.5A as an estimated total. Read the
current rating on each fan hub, for an exact number. From 12V1 rail.

Total so far, at idle in the desktop

12V2 @ 6 amps CPU
12V1 @ 2.4+1.5+4.0+0.5 = 12V1 @ 8.4A HDD/ODD/video/fans

That's a total of 14.4 amps from both rails.

If we look at startup current, for the first ten seconds, the
processor is not running full power. The number is less. The BIOS
might be using one core. So this would be during the first 10 seconds.
The video card is also not at max power, because it isn't in 3D mode.
We'll cut the power on that in half as well (if Xbitlabs had numbers,
I could provide actual values for this).

12V2 @ 3 amps CPU
12V1 @ 10.0+1.5+2.0+0.5 = 12V1 @ 14A HDD/ODD/video/fans

In terms of total system power (watts), I need to include some 5V current
for the HDD and ODD. 5V @ 1A per hard drive. 5V @ 1.5A for optical.
Also, I include 50W for motherboard chipset and RAM DIMMs. Some DIMMs
now, are 2W each. Include 10W for +5VSB to power USB bus devices.
I use the "long term power" numbers for this calculation, with CPU
at full power, video card at full power.

12V * (6+8.4A) + 4*5*1 + 1*5*1.5 + 50 + 10 =
172.8 + 20 + 7.5 + 50 + 10 = 260.3W

so that would be an upper limit. I have an Intel processor here,
that is rated at 65W, that won't draw more than 36W under actual
usage, so the number on a real system is likely lower than the
260W I just calculated.

Looks like my response didn't get to the newsgroup so I'll repeat it here.
Sorry if it shows up twice.

Thanks for the detailed analysis. I forgot to mention an additional tray
mounted HD and I have a DVD as well as a CD drive but those should fit in
without much additional current demands. I like the modular designs and
would like to find a supply that has sufficient ASATA and Molex connetors to
handle my loads without a lot of extentions needed. I'll take a look at the
supply you mention below. Except for cables, it seems to handle enough
currents. Thanks again.

I just bought this yesterday, but the shipper hasn't delivered it
yet. This is to give you an example of how cheaply you can meet the
numbers.

http://images17.newegg.com/is/image/newegg/17-103-013-S03?$S640W$

http://www.newegg.com/Product/Produc...82E16817103013

3.3V @ 30A
5V @ 28A
12V1 @ 18A (enough to not only spin up the drives, but run
continuously at that level)
12V2 @ 18A (plenty for just the processor)
+5VSB @ 2A (not a lot, but sufficient. May require unplugging some
USB if there is trouble)
-12V @ 0.5A (don't care...)

(3.3V & 5V = 150W max) compare to 50W + 4*5*1 + 1*5*1.5 = 77.5W
(12V1 & 12V2 = 348W max) compare to our 172.8W figure, about half load
(DC output 400W) compare to 260W calculated

The supply meets our needs.

And that power supply is $45. I'm putting that power supply in
my oldest system, so I can move a better supply up into another
computer. The supply is likely not going to meet 80% efficiency,
so I wouldn't necessarily want to be running it 24 hours per day.
The better supply, has a higher efficiency, so I can afford to
leave it running if necessary. Another advantage of that SPI supply,
is no Active PFC, so no nasty reactions to my UPS switching to
battery operation.

I suspect my UPS makes the red waveform shown in this picture,
when the UPS is running from battery, and that little SPI supply
has no power factor correction, to get tied in knots by a square wave.
The core of an ATX supply, doesn't care that the AC waveform is a
square wave, but the recently introduced Active PFC on power supplies
(which sits in front of the regular ATX power supply design), does
care about waveshape, as the Active PFC reacts to waveshape in an
attempt to correct it. So far, I have *no* active PFC supplies
in the house. It's not that I don't like Active PFC, merely that
it isn't that compatible with my ancient UPS.

http://i285.photobucket.com/albums/l...ppedPure-1.jpg

You can certainly cure that problem, with a purer sine wave UPS, but
I really don't have the money to waste on that. Even if my UPS died
tomorrow (and it may), based on economics, all I could afford is
another square wave one. So I can't afford to buy any Active PFC
ATX supplies, and find out the combination is not happy. Sometimes, you
can spot those problems, by reading reviews on Newegg, for an
Active PFC supply.

This is a short blog on PFC, which doesn't show what happens. But
it does warn about mixing them.


http://www.dougv.com/blog/2010/03/01...re-not-compata
ble-with-most-low-end-ups/

"This I learned the hard way, as my Antec EarthWatts EA 500 PSU,
which I adore, will go dead the second the battery on my APC
Back-UPS BX1300LCD kicks in."

And that behavior, kinda negates why we have UPSes on computers :-)
My UPS is there, primarily to cover the one second outages we get
here, rather than power the computer for a couple hours.

HTH,
Paul