Power Graph Explanation

The power graph is generated from a small Linux system using a TS-7250. There are three power transducers, one on each of the two branches ("Left" and "Right") and one on the neutral. Each transducer generates a DC voltage in proportion to the current it senses (50 millivolts per sensed AMP). That voltage is filtered with a capacitor to eliminate high frequency noise and fed to one of the input channels of a Maxxim 197 A/D converter. A C program reads 50 samples per second for each of the three channels (one per transducer) and accumulates the power over appoximately two minutes. It then emits a line of text that is input to the gnuplot graph you see.

Source code of power monitoring software

Here is a picture of my electrical panel with the cover off. In it you can see the three transducers. The left branch transducer has a yellow piece of tubing on its wire, the right transducer has blue tubing, and the neutral, the top of the panel, has a piece of black tubing. I tried to separate the transducers as much as possible to avoid mutual interference. For some reason I still do not understand, the current transducer on the neutral generated a negative voltage, so I had to reverse the leads on that channel.

The CFLs we installed in 2007 were GE FLE152DVR30 - 15 Watt Soft White Dimmable Reflector Indoor Floodlight.

For the last few years (writing this in July 2009), my house has taken about 500 Watts even at night with most things off. I call this the base load and it consumes about half of the roughly 30 KWh per day that my house uses. I wanted to find out what this base loaded consisted of, so I went about measuring each device that was on through the night. Here is that list.

Appliance	Watts	Notes
Fridge	60	Average over an 18 hour period in late June
Home-monitor	54	standby 5
Front office PC complex (idle)	20	active 100-150
iMac complex (idle)	15	10 w/o extra disk
Cisco/Linksys switch	14.6
12VDC supply in Munchkin closet	13	Can be replaced with solar-provided 12VDC
Taco Box (for heating system)	10	Can be unplugged in summer
Power-monitoring ARM Linux box	7	Could be solar-powered with 12VDC system
GE clock radio	6.5
2Wire DSL modem	6.1
WWV Clock	5
Gate power	5
Motorola phone base	5
Stereo	5	75 when on
Motorola portable phone stand	4.5
HA7Net Ethernet-to-1-Wire bridge	4.2
Ten-Tec Shortwave Radio	3.7	8 when on
Apple Airport Express	3.2
HP-45 wall wart	3
Canon MF printer	2.2 (standby)
Power-over-Ethernet	1.8
iPhone clock radio	0.6
Toothbrush stand/charger	0
Total	249.4

These are estimates.

Appliance	Watts
GE Spacesaver Microwave	10
Pool pump timer	10
Smoke alarms	10
Sprinkler controller	6
Skylights	5
Door bell	5
HVAC idle	5
Hango MP3 player	5
Washing machine	5
Clothes drier	5
MBR motion sensing light	5
Bosch dishwasher	5
Total	76

The total measure plus estimated base load is 325 Watts, which is about 75 fewer than the overall power measured by the current clamps. I haven't figured out why yet. One thing to do would be to make measurements of the devices for which I have only estimated the power consumption. Another would be to measure the overall power consumption another way (and hopefully more accurately). At 325 Watts, that's 7.8 KWh per day of base load. At 12.8 cents that $364 per year that the base load consumes. If the 400 Watt number is is the real value, that's 9.6 KWh/day or $448 per year in base load.

Back to graphs (top level page)

Last updated July 3, 2009. For more information, contact robert-pool@bedichek.org