Power Wattage Calculator

  • Thread starter Thread starter gawderho
  • Start date Start date
"gawderho" said:
Looking for power Supply Units and reviews, I came across this handy
calculator to determine total PC wattage comsumption; as a reference to
judge a proper size PSU for our PC's.

gawderho

http://www.jscustompcs.com/power_supply/
Click to expand...

Finally, some numbers for video cards.
Their site says:

32mb AGP/PCI Basic Video = 30 Watts
ATI Radeon 7000/7500/8500 = 30 Watts
ATI Radeon 9000/9100 = 35 Watts
ATI Radeon 9500 = 45 Watts
ATI Radeon 9700/9800 Pro = 54 Watts
nVidia GeForce 4 MX Series = 30 Watts
nVidia GeForce 4 Ti Series = 30 Watts
nVidia GeForce FX = 75 Watts

Note that these power numbers don't break down dissipation from
the individual rails. The cards might be using some 3.3V as well
as 12V. But for the purposes of a conservative estimate, you could
take all the current from +12V. Looks like my 3A from 12V guess is
low for the high end cards.

I believe the FX number, because I read 70 Watts for the first
5900 on some website (unfortunately I didn't bookmark the article).
Judging by the reuse of the 30 Watt numbers, I would take those as
wild ass guesses. A card which doesn't use heatsinks is not burning
30 Watts.

BTW: Thanks for the link :-)

Paul
 
Hey Paul!

Yes - I have just calculated I need 532 watts, with 5 Drives! Blimey I'm
right on the edge with the Antec 550. I removed my Q-Tec 550 watt psu last
night and the label quotes: - 16A Max and 20A Peak on the +12v rail - what
"did" this mean?

TA
Paul (Rozel)
Thanks gawderho

--------------------------------------------------------------------

This footnote confirms that this email message
has been swept by Norton AntiVirus for the
presence of computer viruses.

This computer is protected
by GIANT Company's Spam Inspector

---------------------------------------------------------------------
 
I looked at some of the numbers and I don't believe them. I don't
think that an PC-133 SDRAM stick draws 12 or 24 watts. I could be
wrong, but it seems too high. Also, as you've pointed out, this
doesn't partition the current among the various woltage rails. I guess
it is better than nothing.

Arnie
 
Arnie Berger said:
I looked at some of the numbers and I don't believe them. I don't
think that an PC-133 SDRAM stick draws 12 or 24 watts. I could be
wrong, but it seems too high. Also, as you've pointed out, this
doesn't partition the current among the various woltage rails. I guess
it is better than nothing.

Arnie
Click to expand...

It's a little late but there is a disclaimer saying at the top:

Please Note: The Wattages listed below are maximum potential
wattages for each item... It is important to bear in mind that
this amount will never be reached under typical operation.


jo
 
"Paul" said:
Hey Paul!

Yes - I have just calculated I need 532 watts, with 5 Drives! Blimey I'm
right on the edge with the Antec 550. I removed my Q-Tec 550 watt psu last
night and the label quotes: - 16A Max and 20A Peak on the +12v rail - what
"did" this mean?

TA
Paul (Rozel)
Thanks gawderho
Click to expand...

Not every estimate on the jscustompcs site is golden. For disk drive
estimates, for example, remember to get the info directly from the
disk drive manufacturer, as they do give numbers. The jscustompcs
site uses the peak startup current, and the drives don't stay at
that level for more than 10 or 15 seconds or so. As I said in a previous
post, if you have five drives, four will be idle at any one time,
and their consumption could be 7-10 watts when idle (consult manufacturer
numbers for more accuracy). Since the spindles are all stably spinning
at rated speed, the drive motors only have to make up from frictional
losses. So, to start, you can knock about 75 Watts off that number.

In a way, that relates to the purpose of the 16A Max and 20A Peak
numbers. The Max number is for continuous consumption, after the
computer has booted. The 20A Peak number applies for the time it
takes the disk drives to spin up. Basically, the peak time is limited
by the time it takes to overheat the components inside the PS,
so the more massive the heatsinks inside it, the longer the PS
can take the Peak number.

Another glaring error on the jcscustompcs site, is they are
calculating the "input power" - this is the power drawn from
the wall. They are including the inefficiency of the power
supply itself - the power supply generates waste heat when it
is making the output voltages. But the rating of the power
supply itself, is in terms of its outputs, not its inputs. So
the jcscustompcs site is overestimating the whole consumption
number by 20%. As a result, even with gross overestimation,
you won't be over 400 watts of PS output power. (You have to
remember that JCS makes more money by selling big power supplies!)

Processor 1.5V Vcore Conversion 12V AC-DC Conversion Wall
81.8 Watts ------ 80% efficient -------- 70%-80% efficient --- Power
81.8/12=6.8A ^ 12*8.5=102.3W ^
6.8/0.80=8.5A _/ 102.3W/0.70=146W_/

The JCS site is working out the wall power, whereas the PS is
actually rated in terms of the number in the center of the
picture above, which is the power provided to the motherboard.

That 550W is plenty for your system. In fact, a 550W is typically
used for dual processor systems, so you've got at least another
processor's worth of capacity. (Don't forget to include the power
used for your water cooling system, as JCS won't include that.)

Paul
 
There is another factor to consider, power supplies don't get
stronger with age, I suspect ones operating close to the margin fail faster.
I stuck a 52x burner in an old computer with a perfectly good 250w PS,
the PS failed within a month!
 
Arnie said:
I looked at some of the numbers and I don't believe them. I don't
think that an PC-133 SDRAM stick draws 12 or 24 watts. I could be
wrong, but it seems too high. Also, as you've pointed out, this
doesn't partition the current among the various woltage rails. I guess
it is better than nothing.

Arnie
Click to expand...

It does seem high, but then I checked another good site at
http://www.firingsquad.com/guides/power_supply/page2.asp
and for 128 MB of RAM they use a figure of 2 amps off the 5 volt rail, which
is equal to 10 watts. I assume they consider the 128 Megs as being on one
stick. Sooo, a figure of 10-12 watts per stick may not be very far off.
 
Most of those numbers are just wrong. An AGP card without an extra power
connector, which is everything except the highest-end Radeon and FX cards,
cannot draw more than 20 watts out of the AGP slot, as per the
specification.

Most of the power-estimating guides I've seen tend to overestimate
components' power consumption, sometimes wildly.
 
Yes, I am tending to agree that there is a bit too much emphasis being
placed on PSU's. I have read with extreme interest a lot about PSU's
following the other Paul's excellent postings. My system crashed recently
causing "Unrecoverable Errors" on my two S-ATA drives which had been
installed in a RAID 0 configuration. My other volume was made up with two
further, larger, S-ATA drives and the cause of the crash "could" have been
put down to the PSU. Indeed after a few days on here, I convinced myself
that it was the PSU, a Q-Tec Dual Fan 550 watt model. As indicated higher
up this thread, I managed to remove the psu yesterday and saw that it was
rated on the +12v rail as 16A max and 20A Peak and not merely 14A as had
been quoted elsewhere. Given what you have now said Paul, along with what
others are saying here, I am very doubtful that the psu was to blame at all.
If the psu can cope with "short pulls" of up to 20A, then this surely would
have been ok, even with 5 drives and accessories in my system.

Don't get me wrong, I am sure the Antec "TruePower" 550 watt" psu is better
and indeed I will be using it following your advice and my research, but
maybe the Q-Tec was blamed for something unfairly. I am convinced that it
was a sudden failure of my graphics card that caused my crash and surely,
once mosaics have started to appear on the monitor, there is no way that a
fresh psu is going to breath fresh life back into it? Curiously and
interestingly my supplier no longer has any stock - it was a Gigabyte Radeon
9800 Pro (128) - and has now been replaced with the Hercules model, which
has received great reviews.

That said if one is going to build a higher spec PC, then it is important to
research a compatible psu, especially the +12v rated power, given
particularly that I have read that PIII's and Athlons draw their power from
the +5V rail whereas P4's draw exclusively from the +12V rail - and some psu
manufacturers have not changed their build regimes. Other than this aspect,
maybe, just maybe all this talk about power is just theoretical? and that
overall quoted output is not merely "marketing speak" but enough spec for
one to shop around?

Paul (Rozel)

--------------------------------------------------------------------

This footnote confirms that this email message
has been swept by Norton AntiVirus for the
presence of computer viruses.

This computer is protected
by GIANT Company's Spam Inspector

---------------------------------------------------------------------
 
"Jim Banks" said:
It does seem high, but then I checked another good site at
http://www.firingsquad.com/guides/power_supply/page2.asp
and for 128 MB of RAM they use a figure of 2 amps off the 5 volt rail, which
is equal to 10 watts. I assume they consider the 128 Megs as being on one
stick. Sooo, a figure of 10-12 watts per stick may not be very far off.

Jim

www.onandaga.com
Click to expand...

I can get some numbers for you -

PC100 128MB SDRAM using 16 chips - 4 Watts from 3.3V
http://www.valueram.com/datasheets/KVR100X64C2_128.pdf

PC133 128MB SDRAM using 8 chips - 2 Watts from 3.3V
http://www.valueram.com/datasheets/KVR133X64C2_128.pdf

PC133 512MB SDRAM using 16 chips - 4 Watts from 3.3V
http://www.valueram.com/datasheets/KVR133X64C2_512.pdf

You can look up other Kingston products here, or use the info
to estimate what another vendors DDR sticks might draw:

http://www.valueram.com/resources/datasheets.asp

I'm not really sure whether the Kingston numbers are representative
of an active stick or an idle stick. You can end up in endless
debates about this stuff, as you need to know what the "cycle mix"
is to compute the power exactly.

For DDR, you can go to the Micron website and pull a datasheet for
a module. I just got a 512MB PC3200 module and the "write current"
is listed as 1.6 amps at 2.6V. This is 4.16W. To check this, I
also pulled the datasheet for the individual RAM chip used on the
module, and it lists 0.2 amp. Even though there are 16 chips, only
the 8 chips on one side of the module can be writing at one time.
The other side of the module will be idling. So, in fact, even DDR
is pretty good stuff, and no where near the 10 to 20 watt numbers
on the JDS site. No one will want these gory details, but here
they are:

(chip - page 53)
http://download.micron.com/pdf/datasheets/dram/ddr/256Mx4x8x16DDR_G.pdf

(module - page 14)
http://download.micron.com/pdf/datasheets/modules/ddr/DDA16C32_64_128x64AG_C.pdf

(how to calculate accurately, not important for this exercise)
http://download.micron.com/pdf/technotes/TN4603.pdf

And, no, the "Auto-Refresh" currents are not representative, as
the percentage of the time spent refreshing is insignificant (i.e.
it is a low duty cycle). I tried to pick a pathological case, where
the module is reading or writing all the time, and the power still
isn't all that great.

The "OPERATING CURRENT: Four device bank interleaving READs (BL = 4)"
number is interesting. First off, I'm not smart enough to know whether
that pattern can occur in the real world or not. Depending on how
address busses are hooked up, it seems like a large "stride" for a
typical Windows application. In any case, if we use that number of
3.792 amps for the DIMM, that gives 9.86 Watts, and may take a specially
written program to achieve that power consumption. I wish I knew
what thetaJA was for the 66 TSOP chip package, because if it was
say 100C/W, the chips would burn you.

I found an answer here:
http://www.flotherm.com/technical_papers/t322.pdf

It seems the industry practice is to take a portion of the two
power numbers. It still looks like 5W for an active PC3200 DIMM.
That article is amusing and worth a quick glance.

If the module dissipated 10 watts, the whole module has a thermal
resistance of 5.5C/W, so the module ends up 55C above room temp,
for a total of 80C (too hot!).

HTH,
Paul
 
Back
Top