The Internet Enabled Kill-A-Watt

The Internet of Things has been applied to toasters, refrigerators, Christmas lights, Barbies, and socks. Unsurprisingly, the Internet of Things has yet to happen – that would require a useful application of putting the Internet in random devices. One of the best ideas is a smart electric meter, but the idea behind this is to give the power company information on how much electricity you’re using, not give you an idea of how much power you’re pulling down. The answer to this is the Internet-enabled Kill-A-Watt, and that’s exactly what [Solenoid] is building for his entry into the Hackaday Prize.

Modern power meters have an LED somewhere on the device that blinks every time a Watt is used. This is the data [Solenoid]’s creation is pushing up to the Internet to relay power consumption to himself or anyone else in the world.

The hardware, like many upcoming Hackaday Prize entries, we’re sure, is based on the ESP8266 WiFi module, with a light sensor, SD card reader, and OLED display. It’s meant to mount directly to a power meter, recording power consumption and pushing that data up the network. It’s simple, but it also allows for very granular monitoring of [Solenoid]’s power consumption, something the electric company’s smart meters can’t compete with.

The HackadayPrize2016 is Sponsored by:

47 thoughts on “The Internet Enabled Kill-A-Watt

  1. I swear, the next time I see a Hackday post say the words “Internet of Things” I will not only unfollow you on FB, I’ll never visit your site again. And I’m a regular, in case anyone cares. I get buzzwords. I wasn’t born yesterday. But to regurgitate them EVERY FUCKING DAY makes this guy GO AWAY.

    1. I remember back in 2007 when it was called “Wireless Sensor Networks”, and everyone was enamored with the idea of “smart dust”, which would use energy scavenging.

      Actually, not, the pipe drive of energy scavenging has come back around, and is being proposed again by people who want to add mystique and gravitas to any idea they propose. I bet we’ll start hearing about the “smart dust” again!

  2. I was hoping for a kill-a-watt hack, to push kill-a-watt stats out rather than a kill-a-watt equivalent. Bummer.

    Still a clever idea, and well executed, though I’m thinking my utility wouldn’t be thrilled with an attachment to their meter.

  3. Has anyone come up with a small, simple module I can put in my circuit breaker box with, say, a dozen wraps of enameled wire around the hot line going to each breaker to give more detailed breakdowns?

    1. You’d need a professional install in the circuit breaker box. The sensor has to go around a single AC wire, not the pair. Pics I’ve seen usually have the sensors added to a length of stripped AC line tacked out to the wall under the box.

      1. Kinda like the idea of that emon-project. If only they would have something smaller, like a sensor that could be plugged between your cable and wall-outlet, or some power strips with included sensing…
        ‘Hacking around’ with 230VAC installations should not be required, because it can be lethal if you don’t really know what you’re doing and how it has to be done.

    1. There are two types of power meter LEDs, but none I’ve heard of blink when a ‘watt’ is consumed. Once per KWh is what they say. There is a less-common type which uses the LED to xmit a data stream of the current consumption rate.

        1. This led me to ask, and then have a look, do they make bi-colour LEDs with infra-red as one of the colours? And they do! Dunno how easy to get hold of, don’t really care either, but they do. That’s interesting innit?

  4. I’ve done something similar on my meter. My IR sensor is hardwired back to a Teensy, which in turn talks to a python script over serial and then up to The main problem I’ve had is the IR LED is actually quite weak, and will get washed out by direct sunlight. I resorted to securing it with adhesive foil tape, covering the top half of the plastic meter housing. Even so, if enough sunlight shines on the face of the meter, I still lose my reading then.

    Best way would be to just read the meter wirelessly. Either through whatever Zigbee thing the Itron uses or this

    1. Yeah, I like those Itron meters. The Zigbee output is very useful! It’s just tricky to get the utility to allow you to pair with it. But if you can get that done, you’ve got far more data than “a pulse every watt[hour]”.

      1. Also, if you have an Itron meter (or other zigbee enabled smart meter), you might try getting one of the cheap in-home-displays that are approved by your utility, and hacking that. In that way, you don’t need to worry about the pairing as much. Most in-home displays are pretty hackable, with an I2C, JTAG or other common interface.

  5. We have done something similar.
    We have made an ESP8266 based sensor network for heating and electrical meters (both Kamstrup devices).
    Currently we have about 6 in production with data up at:

    All sourcecode and pcb designs available at :

    Currently we have two devices in production:
    1: Kamstrup heating meter daughter-board feat. ESP8266:
    This device logs from the heating meter (via reverse engineered uart protocol).
    This device is also able to control a heating buffer using to triacs.
    2: Kamstrup electrical pulse meter daughter board.
    sits beside an electrical meter and aggregates the pulses.

    Both devices transmit their measurements to a central mqtt server and feature an onboard ring buffer for some retransmission tries.

    Cheers from Christiania Nabovarme, Copenhagen, Denmark

  6. A fine idea. But the solution that really needs to happen is: this needs to be built into every AC outlet, and then it needs to be commoditized. The box in the wall should be made into a socketed mounting point itself, and then a whole unit, either “smart” or “dumb” can be plugged into that. Then per-device energy audits can be done.

    This always makes me wonder what the absolute minimum, cheapest way is/could be to take 110VAC and generate a low average current 3.3v out of, 5-200mA average current or something close to that, since that is all you would need to run a micro and radio with a duty cycle.

    Other than that, IoT is a shit soup of bad ideas and platform wars. Zigbee is a failed standard. Thread is building on 802.15.4-2006. BTLE has no meshing, but they are scrambling to write a standard. Wifi is horrible, but it is the only thing that is pervasive.

  7. There is an assumption here being made about the blink rate and how it relates to energy usage. Depending on the meter it may be programmable value. The value is called kh and it relates to the old analog meters where the kh value would correspond to a unit of power per disc rotation over a set period of time. In modern meters they still use a kh value and instead of looking at the black line on the rotating disc you now look for a blinking LED. Many residential grade meters are too cheap to have a blinking LED but instead will have a black arrow that flashes on the display. A good rule of thumb would be that if there is no optical port on the front of the meter it is not field programmable. If that is the case the kh value is probably fixed and would be listed on the meter somewhere. The formula for calculating power is P=3600*kh/time (in seconds). If there is an optical port the kh value could be anything. Meters I’ve programed come out of the box with a kh of 1.8 but that could easily be changed. I know that a kh of 7.2 was common for older meters. Hope this helps.

  8. Energy company here offered a device with idea similar to this. It would watch the blinking led on the meter and send wireless data (probably using 433 Mhz radio) to display unit. Inspired on that, I built a simple monitoring device with couple of cards using Ti cc430 radios. It used to log data to database and I had a report on web page for energy usage.

    But that way it didn’t affect our energy usage habits at all. Just recently, I added a small display unit using similar OLED display as in this project and EMW3165 module ( I feel that this might really change something, as everyone in family can now easily see how much power we are pulling and maybe think if it is really necessary to have all those things ON that they have.

    1. Wanted to do the same thing, but i concluded that it is not feasible to rely on it for long period of times with low load.
      Second issue is that if you don’t tap the voltage signal as well, it is difficult to see the phase angle and will not estimate power consumption correctly. And if I tap the voltage, well, it is easier to just dump an always on small dc dc converter.

      1. I have noticed that many others seem to think voltage is important. I assumed it would be for voltage variations (which may be up to 10%, I’d guess) I am surprised it would be for phase angle. Do we really have that many motors spinning in our houses? I’m sure power factor is important for industrial plants, but I think I could count on one hand the number of big motors that I have in my house that would significantly affect the phase angle on a circuit. Am I missing something else?

      2. Do we really have so many things in our house that affect the power factor? PF is important for industrial facilities with many motors, but I would guess much of the power draw in the home is not reactive. I always assumed voltage monitoring was for voltage fluctuations, and I can live with a 10% error for discovering where I am using the most power. 10% is probably in the noise of my daily power consumption.

        I agree on long periods at low power, but I decided those situations are not as important to monitor. I’m shooting to harvest enough power from a 0.5 A current draw (60 W in the US) for hourly updates. If I’m drawing less than 0.5 A on a circuit, it will stop reporting and I’ll know it’s not a significant power dissipation.

        1. Most of the loads I have in the house are complex and have bad power factor, the majority have some sort of SMPS and I bet very few have power factor correction. So yes, voltage is important.
          It is also important that 10% voltage change will have a 20% impact in power for some simple loads like the electric cooker. I am not going for 0.01% accuracy, but I would not consider 20% to be enough.

          Overall the classical solution of powering the thing and measuring I and V is simple and ok enough for me. I confess that the idea of a tiny clip on device on the cable has interested me, but the compromises are not worth it.

          I bet @Capt. McAlister is measuring just the current because there is only analog input on the ESP module.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.