Tweet-A-Watt alternative solution

One of my long time interests has been to log my power usage. I have an ongoing project  which will enable logging the whole house power consumption. That’s cool but I also wanted a way to log data for individual appliances. So I bought a Tweet-a-Watt starter kit (TAW) from Adafruit. This was a really cool solution to allow me to enhance a Kill-A-Watt power monitor (KAW) and make it transmit the power usage wirelessly so I could graph/log it.

But there was a problem… unfortunately the TAW didn’t work 🙁

After assembling everything, as soon as I would plug in the KAW, the screen would be blank and slowly come to life but still quite unstable and flickery. Data would start coming on the other end but after about 1 minute, the KAW screen would flicker really bad, then go blank and data would stop streaming. I tried a few things but nothing worked. The conclusion was that the XBee radio used by the TAW was drawing too much power and the KAW could not handle its needs. Apparently I had a newer revision KAW which was lower power which could not supply enough power for the TAW. Forums revealed that different revisions of the KAW had similar problems.

So I designed an alternative transmitter circuit using a low power Arduino clone that I designed for other wireless data logging, the Moteino. The operation principle is the same as the circuit from Adafruit. Power comes from the KAW through a Shottky diode to keep dropout low and stop current flow back into the KAW. There’s also a capacitor to help stabilize the circuit when it transmits. The Moteino runs at 3.3v so the current/voltage sampling from the KAW is fed to analog pins on the Moteino through a simple 10kohm<>4.7kohm voltage  divider. An LED is used for transmission indication.

A python script was adapted from Adafruit’s version, to read data from a receiver Moteino, graph power and voltage/current waveforms, and also transmit power data to my Emoncms which is running on my RaspberryPi. The results are quite nice, you can see it in action in this video:

MikeT over at posted his version of this project, this time including a Moteino!

4 thoughts on “Tweet-A-Watt alternative solution

  1. I’m also a big fan of RFM12B, bought two of them and made shields for Uno, and then bought two JeeNodes. Now I see your Moteino as even better (smaller) than JeeNodes, so I’ll be checking the out. I worked a reliable circuit to power up the JeeNode, and it’s posted on my blog (above), though only as an example of TinyCad. I’ll be posting a couple or few articles on the JeeNode Kill-A-Watt as well. One thing I found was the current sense line is very noisy if I use the 4.7K + 10K divider, especially with a light load. I experimented with a lot of things, even active buffers, but ultimately decided a 1K + 9K resistor divider is optimum. One last point, if you haven’t already, check out, JC has a great library for the RFM12B, that would probably work great with your Moteinos. He even has RF wireless code update covered. Good stuff and on your site. Later…

      • Cool, thanks! And BTW, I struggled quite a bit getting the KAW to tolerate any extra load, even a few mA on either the +5 or the +6V rails. Definitely need low power (sleep) mode between measurements and transmission. That meant no delay in the Reset, need to get that sketch running ASAP!, contrary to LadyAda’s design with the Xbee. And even so, it was very finicky. The JeeNode 3V3 regulator would tend to get stuck just under 3V, drawing a lot of current and prevent the KAW PS from fulling coming up. My solution was to go back a stage, to the raw 13V supply, with an added transistor to charge the big cap. Then the only remaining issue was when you unplug the KAW, the JeeNode supply would take a really long time to discharge, which was a pain if you wanted to plug it back in right away, say to reset the KAW elapsed time. So I added another transistor to discharge the big cap. Now it works every time.

        • Hey Mike,
          I had similar issues with even a few mA being TOO much for the new KAWs to handle. The XBee power draw was no match. Moteino works flawlessly inside the KAW, provided I used the low power sketch that you can find in my GitHub repo.

Comments are closed.