In order to monitor our power consumption at our house, I set out to install a TED (The Energy Detective) system
. The system monitors current flow into the house using inductive taps. In our case, we have two panels each of which required two inductive clamps (on the incoming power lines near the top of the photo).
Each pair of inductive clamps connects to a small box at the base of the power panel. The TED system, at the time I purchased it, used a powerline protocol to send the current readings back over the powerline in the house to a small monitoring device that could be plugged into both the wall and a USB port on the computer. This would have made things very easy except for one thing: the TED powerline protocol interfered with powerline protocols like X10. That meant I had to get the TED powerline signal back from the sensor boxes in the power panels to the computer somehow without the interference. My solution: create an isolated circuit just for powering and communicating with TED. The photo below shows a box I added nearby the power panels. Inside is an isolation transformer (often used with medical devices) which gets power from a breaker in the panel but then creates a separate isolated circuit that provides power to the TED boxes in the panel. In addition, you'll see a power outlet I installed in the top of the box that's also on the isolated circuit. A long power cord leads from there to the central data area of the house (where all the CAT5 cables converge).
There you can see the TED remote monitoring box being happily powered by its isolated circuit. A USB extender is then used to remotely connect the monitoring box over CAT5 to the main computer (in another room) which is the central home automation controller. That computer records the data and serves up the TED web interface for other computers and devices. The two measurement devices each take a current measurement every second (yielding a kind of heart beat on the monitor's LED as the data is received). Although this is a whole house power consumption monitor, some of the more power hungry individual devices can be detected by their unique power consumption signatures.
It was actually Limor Fried's Tweet-a-Watt project
that I built from plans in Make magazine issue 18
that first got me thinking about collecting power consumption data around the house. The TED system was the easiest way to collect a more complete set of data (without placing a lot of individual sensors).