[PLUG] UPS Education

Vram lamsokvr at xprt.net
Tue Nov 22 05:56:08 UTC 2005


On Mon, 2005-11-21 at 21:21 -0800, Wil Cooley wrote:
> On Sat, 2005-11-19 at 15:40 -0800, fh hillsboro wrote:
> > So, to continue this very interesting and informative training session, 
> > i got out my (not so?) trusty Radio Shack V-A meter and tested some of 
> > my systems to see what kind of current they draw.  Here's the results:
> > 1 - an old 233MHz Dell box:  never saw above 1 Ampere
> > 2 - a pretty old 2 processor X 800 MHz HP workstation:  never saw above 
> > 2 Amperes
> > 3 - a home assembled 2GHz system: never saw above 1 Ampere
> > 4 - a Sony 24" glass monitor (really heavy!!):  also under 1 Ampere even 
> > on startup
> > 
> > These numbers seem a bit light to me.  Any comments?  I used the meter 
> > and associated inductive pickup/multiplier on the 10X setting for better 
> > accuracy.
> 
> Yup.  I recently played shepherd for a bunch of HP DL360s with dual 15k
> RPM drives, dual 2.6GHz Xeon procs, lots of memory on metered APC PDUs
> (power distribution units; think high-end powerstrip) and rarely saw a
> draw of more than 6A with all 8 sockets in use (some of them were
> plugged in but used by the secondary power supplies, which I was told
> didn't draw if the primary was operational).

NOT the same


The measurement of electrical power that is computed by multiplying
volts times amps. In a DC circuit, volt-amps (VA) and watts are the same
because DC circuits do not add inductance and capacitance that take away
power. In AC devices, volt-amps ratings are higher than watts. For
example, in an uninterruptible power supply (UPS), the volt-amps rating
is approximately 60% higher than the actual wattage.



When a sinusoidal voltage is applied to a purely resistive load (an
electric heater) the current has exactly the same waveform and phase.
We use "rms" to indicate the magnitude of a sine wave voltage or current
because a sinewave of 120 volts rms produces the same amount of heat in
a resistor as a steady dc 120 volts.  A 120 volt rms sine wave has peak
values, plus and minus, of square root of two times 120, or plus and minus
170 volts.

For a purely resistive load, the power dissipated, in watts, is simply
volts rms times amps rms (watts == volt-amps).  Because then is no phase in DC.

If the load contains inductors (eg most motors or transformers), or 
capacitors, the current either lags behind or leads the voltage.  
Voltage and current are out of phase.  
When the voltage reaches its peak the current is not at its peak.  
If the difference in time between voltage peak and
current peak is given as an angle, theta (one complete cycle = 360 
degrees), then the power is volts rms times amps rms times cosine(theta).
Volt-amps is greater than watts.

When the alternating current supply is rectified to obtain the direct
current needed by most electronics, the current waveform may be far from
sinusoidal, for example "square-wave"-like or a series of impulses.
Here the power must be determined by calculating the product of the 
instantaneous voltage and current, and integrating the product over a full
cycle.   Watts will generally not be equal to volts times amps.

So, in sizing a UPS, make sure that the total load requires fewer watts,
AND fewer volt-amps than the rating of the UPS.  Unfortunately, most
devices do not have labels indicating both watts and volt-amps, so we
just have to guess, and use a somewhat larger UPS.





> 
> If my interpret my observations correctly, these high-power-draw servers
> were only actually drawing about 100-150W.  If I'm wrong, I trust
> someone who knows this stuff will disabuse me of these mistakes
> (Galen?).
> 
> People seem to think that that 350W or 400W rating on the computer's
> power supply indicates that it draws that continuously, like a
> lightbulb.  That is not the case AFAICT, which is why UPSen are rated in
> Volt-Amps instead of Watts--they are in fact the same, but VA is a
> little consensual lie that makes you think about the actual draw, as
> opposed to a power supply's rating.
> 

This is a good assumption...  The speed-o-meter on your car goes to 120
MPH...  How often do you go 120 MPH <OK.  Don't tell me.. I don't want
to know>


Then power supply manufacturer is assuming the user will draw at most
50% normal load....

Which means if you have a 400 Watt power supply  the PS vendor is
counting on you using only 200 Watts normally...

The PS is not designed to put out 400 Watts continuously.  
It can the life goes down rapidly!!!



HTH


Vram








> I've been thinking about getting one of these, because I'm very curious
> about this kind of thing: http://tinyurl.com/czjkx
> 
> Wil





More information about the PLUG mailing list