Open Source Wireless Mesh Networking Energy Meter

[Jay Kickliter] writes in to tell us about his open source energy/power meter. With his buddy [Frank Lynam] they designed a small device that crams into existing power boxes and uses and 8 core propeller (P8X32A) microcontroller to perform true RMS voltage and current measurements using a current transformer. [Frank] and [Jay] don’t stop there. The meter also features an xbee pro 900 MHZ to provide wireless (and even mesh networking) capabilities to the whole ordeal.

[Jay and Frank] estimate a total unit cost of around $80 (US) per prototype. With volume the price goes down by about half. With a larger number of units, and the magic of mesh networking, we could see cheaper xbee’s driving the cost down some. Check out the Google code page for details or the schematic (pdf) if you are interested.

So far the project is in the beta stages, and only features a single module sending data to a PC running an OS X Cocoa application. [Jay] is about to be otherwise occupied by the Merchant Marines and [Frank] the Navy, so they figured we could have a go at it for awhile.

We have seen other hardware used to monitor power consumption, but cramming this circuit into each power box is a neat idea.

[Frank] explains the whole project in the video after the jump.


35 thoughts on “Open Source Wireless Mesh Networking Energy Meter

  1. We’re using XBee’s because this was a proof of concept for utilities to use as actual home meter’s. I guess I didn’t get that across in the description. If I were to make one specifically for single outlet use I’d definitely use a much cheaper transmitter. $80 is what it cost us to prototype on such a short schedule (1 month idea to product). A lot of that was overnight PCB’s and much too expensive op amps and voltage regulator. Cheaper IC’s can be used to drop the price significantly.

  2. Those are some interesting modules, Ill have to remember them for future RF projects! :) The zigbee is far more powerful (+17dBm) than the RFM23B(+13dBm), TR24A(+2dBm) or the nRF24L01 (+0dBm). Also the zigbee has mesh networking capability, the module’s micro builds the whole network and manages packet transmission all on its own, less fuss.

  3. I would never use this because of various reasons.

    HV measurement is directly in the AC part without any galvanic isolation.
    You could fry your electronics/yourself easily.
    The use of an opto-coupler is advised.

    further the current op-amps could be connected in another way, so the op-amp with the virtual ground could be ditched.

    I would advise to connect the op-amp as an differential amplifier.
    And use the a load resitor for an power measurement

  4. I get xbees for 25 bucks a unit single order and I’m nobody special..I could get the rest of the parts around 10 bucks from most dealers single unit order. Thus why I ask what the other 40 is for..

    Something like this though you want to go as cheap as possible on digital logic parts.

  5. I love how everyone is so quick to jump onto how much better we’d make it, without considering the reasoning behind the design choices that were made :)

    As they were trying to complete something quickly and simply, they opted for less risky/more expensive components like xbee’s and an op amp for virtual ground. Just because you have a different definition of “better” doesn’t mean it’s right ;)

    Rather than uselessly picking at other people’s designs online, please spend less time stroking your e-peen and more time building projects! :)

  6. I am sorry if my reaction sounded like that Davd.

    I just was looking through a cool idea and cool project and saw some room for improvements and a little protection for the schematics.

    I don’t want to bash it in any way, just to improve it.

  7. The HAD summary is a little misleading, the board pictured and built by the guys is only good for 12Amps, not exactly whole house rated like my current clamp meter mentioned at the end of the HAD article.

    O wait, I get it. The want to cram this thing into every outlet. That can get expensive, but gives more detailed about your power flow I guess. I’ve had some requests to add features to my project that would allow multiple circuits to be monitored, which would give room level detail about power flow depending on how your house is wired. So there is some demand for it.

    I got sent a free Xbee smart meter RF kit, and when I have time, I’m going to look into integrating it into V2 of my power meter, along with the multi-circuit monitoring. I’ll have to thumb through this code to see if there is anything I can use for my project. Good Job guys.

    Also, Go Navy! (I’m an engineer for the Navy)

  8. RFM22B has +20 dB. Mesh networking can be implemented on any module.

    +13 Db is already quite enough for in-house networks, don’t spam on public frequencies with high-power transmitters plox.

    The project is great no matter how much money was spent since energy accounting is a good task to invest time and money in.
    There’s a SEGmeter project which uses Hall effect sensors instead of transformers.

  9. This is FAIL because of even $40 is too expensive (come on, compare it with possible savings). Idea of having this in every socket is insane (just do basic math how much it would cost (well man can dream of course)).

    It is also FAIL for using wireless instead of power line communication.

    Still huge success for being open source and true RMS.

  10. I have build a wireless energy monitor with various sensors, smart meter readers, thermostat and ac panel monitor using the RFM12B modules. More work to insure reliability and mesh networking, but you only have to do that once so it’s worth the financial savings. I’ll post the project one my website one day.

  11. You need to sample faster than 100 samples per second. Your platform is clearly capable of it. In the US you need to sample at at least 120 samples per second, and even higher would be better. I understand that this is a proof of concept, but it matters. You aren’t getting even a decent approximation of the voltage and current waveforms at that sampling rate…

  12. After looking again, your FFT idea is pointless if you are only sampling at 100 Hz. Read up a bit on how the FFT works, as well as nyquist theory and you may learn something.

  13. @Bill. The unit was just configured for 12A, it’s easy to change the max current capacity. All you have to do is put in a different current transformer and change the burden resistance. And maybe change the gain current sense op amp circuit. The next generation board will be able to switch either gains or burden resistance, for a wide dynamic range.

    @bigbob, not sure where you got that number, maybe we wrote down some bad specs somewhere? We’re sampling at 6 kHz, or 100 samples per voltage cycle. The FFT it is 1024 Hz.

    @Kris. We have no intention on on installing this in every socket in a house. The only reason we the current transformer was installed inside the box in that demo unit is for portability for demonstrations. I’m going to install the next generation design I come up with on the 2 phases coming into my house.

    This isn’t a drop in product. It’s a prototype, budding platform, that will most likely need to be modified for whatever use-case you might come up with. And $40 is a lot cheaper than what utilities are paying for smart meters.

  14. @Jay

    I understand that this is only the prototype.

    The problem is in my opinion that used components are inheritly expensive and to get the price (really) lower (like into $10-$15 range) you have to completelly redesign your project.

    But thumbs up for very good documentation. This is very inspiring!

    FYI Plugwise is a commercial product that does basically the same (check

    But there is one very big catch – THEY STEAL YOUR DATA. Seriously, they secretly upload the data to their servers ( So I am unable to buy their product.

  15. @ Jay

    I misunderstood the video. At about 1:40 it is noted that the waves are sampled at “100 samples per cycle” and I misunderstood that as “100 samples per second”.

    My bad…

  16. I priced an exact replicate at 37.89+$10 in S& can do it under 25 by getting rid of the xbee..

    If you don’t like criticism don’t’s simple. People aren’t going to be quiet cause you’re over-sensitive about your work.. it’s public and we’ll say what we want about it..get use to it or GTFO.

  17. Great work.

    How about a client that runs on open platforms?

    I’d really appreciate a device like this if I can build it for <$25/ea including the CT. I think the ZigBee module probably precludes that, but it might just take a bit of hacking to switch to one of the Nordics or similar.

    Is there room to increase the sampling rate?

  18. Oh yeah, and do you even need the virtual ground buffer at all? I don’t see where you’re drawing any current from it. I only see it being used as a reference at the high impedance noninverting inputs. Why not just use the divider?

  19. @Duane, sorry, I forgot to reply. It’s a CR8420-1000-G. You can use any CT you want, with modifications.

    @xorpunk, you’re right. Thanks for doing the pricing, $40 was just a rough estimate. Did you find cheaper IC’s, like op amps, or used the original parts list?

    @error404. I only have experience writing OS X programs. But, there’s Google Power Meter. If anyone else to write an interface for it, that would be awesome. About the VGnd buffer. You might be able to use just a divider, and we did in prototypes. But we got inconsistant results. Then I found an old TI app note about single supply op amp circuits, that recommended using a buffer to feed the same virtual ground to multiple circuits. I’ll attach the link below. And yes, you can increase the sampling rate significantly. I don’t see much of purpose of going much faster, unless you wanted to work with a greater than 60 Hz system.

  20. Come’on guys, i hurts to see the lot of cheaping and bitching comments. we are missing the point here that is OpenSource, basically leaving it to us to improve and adapt to our particular situations… and contribute back!

    will we have to duplicate the voltage and CT circuit in order to monitor the other 120 vac line in US installations? or better yet, find a way to sync more than one unit?

  21. Lovely project :)
    I have thought about the same thing my self.
    Would be really awesome if you added a solid state relay on each phase. So it would be possible to wireless turn the socket on and off. Another cool safety feature would be to add temperature sensor, so if something goes wrong and the socket temperature gets to high it would automatically shut off the power to prevent fire

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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