[Beowulf] Re: power usage, Intel 5160 vs. AMD 2216

Jim Lux James.P.Lux at jpl.nasa.gov
Sun Jul 15 17:39:46 PDT 2007

bAt 03:46 PM 7/15/2007, David Mathog wrote:
>Greg Lindahl <lindahl at pbm.com> wrote
> > So, the thing that logs power usage over time seems to be the "Watts
> > Up Pro", which says that it plugs into USB and has a Windoze program that
> > graphs power usage. Does anyone have one, and can you access the data
> > sans Windoze program? Are there any other cheap logging power meters?
>The Watts Up Pro is only $130 at Amazon right now - if your
>time is worth anything just buy one.
>If this is a one time deal and you  have a USB Camera you
>set it to snap pictures of the display at fixed intervals, and
>then scan through those later.
>Or, if you already own a USB data logger and a Kill-a-watt, try
>this at your own risk. Most likely you'll end up filing the effort
>under "I wish I'd just spent the money on the data logging version"!
>On opening up a Kill-a-watt you'll find that one side has all the power
>goodies and a tiny amount of logic, and the other side contains the
>display, the buttons, and what looks like the logic to control the
>display.  The two sides are connected by a 6 wire cable, where
>all of the wires are the same diameter, with no shielding and no
>twisted pairs.  It seems likely that if one attached a voltmeter
>to those 6 wires one would find a ground, 5V (or 12V) dc, and
>4 measurement lines, probably representing Line Voltage, Current, Line
>Frequency, and Power.  If you're lucky the measurement values are
>encoded as a DC voltage. That would be easy to test, and if true,
>you could solder leads onto those 6 lines and bring them outside
>the case to attach to the input leads of your existing USB data
>logging device.

Whoa there cowboy!!  You're risking death, dismemberment, and 
egregious excitement from the destruction of your datalogger.  The 
entire insides of the Kill-A-Watt are floated at line 
potential.  Providing isolation to a connection to the outside world 
(i.e. to provide a serial or USB port) costs money, not only for the 
parts, but because the design process is more rigorous and you need 
to have a lot more regulatory oversight (testing labs, etc.)

There's a good description of the insides of this beast posted by 
Terry on the Tesla Coil Mailing List at http://www.pupman.com/ where 
he figured out how to recalibrate it, the differences between the 
110V and 220V versions, etc.  There's been a post on this list with 
the exact URL, too.

It's not just a analog current and voltage measurement that you can 
log, because you need to measure power factor, and that requires 
digitizing the waveform (or at least fancy electronics to measure 
instantaneous I*V)..

If one wanted an interesting project.. Get two isolation transformers 
and send V and I to Left and Right inputs on the sound card, digitize 
it, and process it in software.  SOX could do a nice job digitizing, 
Matlab, C, Python, LISP, what haveyou to do the processing.

For isolation transformers:  Current transformers are available new 
from Newark, Mouser, etc..  For a voltage transformer, get a low 
power transformer used in power supplies.  A 5-10V output (from 120V) 
is typical, then use a resistive divider to get it down to something 
reasonable for the sound card (about 1 V p-p)...  Check the waveform 
with a resistive load, because some of those small cheap transformers 
saturate.  If you get a little PC mount transformer with 120V/240V 
dual primary, you can run it as a 240V transformer across the 110V 
line, which will almost guarantee no saturation (doubling the turns per volt)

James Lux, P.E.
Spacecraft Radio Frequency Subsystems Group
Flight Communications Systems Section
Jet Propulsion Laboratory, Mail Stop 161-213
4800 Oak Grove Drive
Pasadena CA 91109
tel: (818)354-2075
fax: (818)393-6875 

More information about the Beowulf mailing list