open source power meter

[NiHaoMike]

For those who have used the Kill-a-Watt power meter before, it is a surprisingly good device (especially for $20!), but it has several flaws:
* It only works on 120V AC and up to 15A.
* Resolution is really poor at the low end, greatly limiting the use of the device for measuring standby losses.
* Only one measurement can be viewed at a time.
* It blocks the other outlet. (For those outside the US, US power outlets are almost always pairs.)
* The lack of backlighting makes it difficult to read in certain situations.
* Accumulated energy measurement is lost if there is a power outage or the unit is unplugged. Additionally, unplugging it is the only way to intentionally reset the energy measurement - a real problem if the device you're measuring is to remain on.
* No wireless capability. There is a kit (Tweet-a-Watt) to add it, but it's expensive.
* It will not measure reverse power flow, making it useless where power flow could be in either direction. Grid tie inverters and induction motors operating as regenerative brakes, for instance.

I have taken mine apart, and it uses a simple analog front end connected to some sort of microcontroller. It's exactly what I expected for a cheap device. And easy to build your own.

So let's start the design:
* The fundamental frequency, for AC, is either 50Hz or 60Hz. That translates into a minimum analog bandwidth of 500Hz or 600Hz (to capture harmonics) and a minimum sample rate of 5kHz or 6kHz, not very difficult. For DC, it would only need to sample a few times a second.
* 1% or so accuracy in order to keep costs low. That is particularly important for low cost wireless sensors. The standalone bench version could have somewhat higher accuracy.
* To accommodate the wide range of current measurement, it will use a switchable current shunt amplifier. Possible implementation could be to use two opamps set to different gains, and use a multiplexer to switch between them.
* Low cost microcontroller with required A/Ds integrated, probably would also have the required multiplexer integrated. Some TI MSPs even have wireless integrated, but those are surface mount.
* Capacitive power supply for low cost AC versions, just like the Kill-a-Watt. Small switcher for other AC versions and some DC versions, linear regulator for very low voltage DC.
* Wireless versions should use an open protocol and be able to share a channel.

And as for possible versions:
* Low voltage DC with small 7 segment or text LCD, for alternative energy and battery operated circuits. Good starting point as it is low voltage and simple.
* 120V or 240V AC, 15A wireless sensor.
* 120V or 240V AC, 15A bench unit, with graphic LCD and RS-232.
* Split 240V AC, Hall effect current sensing for whole house monitoring, intended for permanent install in a breaker box.
* High voltage DC, Hall effect current sensing or high current shunt for electric vehicles and large alternative energy applications.

[NiHaoMike]

http://enerjar.net/
That's similar to what I have in mind, but with a more flexible design.

[scrat]

Once by chance I found  this chip..
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en520376
I never used it, but I think that if they've done a chip just for that, it could be a good thing to take a look, perhaps they have already solved some precision issues...

[NiHaoMike]

That chip would probably be a great choice for a bench meter or whole house meter where accuracy is important, but not so much for a simple wireless sensor. You'll still need a microcontroller to operate the display or wireless interface.

[TheWelly888]

Is there really nothing commercially available that meets most of the specs given?

[NiHaoMike]

Nothing I'm aware of that is reasonably priced. I'm looking for something like $5-10 per wireless sensor and maybe $20-30 for a whole house meter.

[saturation]

Hello NiHao,

Some ways I've overcome your comments, only those in red I've been able to do something about:

Below it minimum resolution, if there is actually a draw, leaving the unit on for many hours will result in a cumulative wattage/hours that you can use to calculate its actual kwh

The no-input power consumption of the Killawatt itself in negligible and can be ignored,  I reconfirmed it last night, its 0.01kw/7.75h

Use an extension cord to unblock the outlet and make for better viewing angle, so backlight isn't strictly necessary

The updated version has a memory so it retains the last wattage before a power off; it needs to be purposefully zeroed to measure from zero

On you lead, I opened the unit.  Its a series resistor with the neutral lead, measuring I= V/R then calculating power as VI=W as AC rms. 

For those who have used the Kill-a-Watt power meter before, it is a surprisingly good device (especially for $20!), but it has several flaws:
* It only works on 120V AC and up to 15A.
* Resolution is really poor at the low end, greatly limiting the use of the device for measuring standby losses.
* Only one measurement can be viewed at a time.
* It blocks the other outlet. (For those outside the US, US power outlets are almost always pairs.)
* The lack of backlighting makes it difficult to read in certain situations.
* Accumulated energy measurement is lost if there is a power outage or the unit is unplugged. Additionally, unplugging it is the only way to intentionally reset the energy measurement - a real problem if the device you're measuring is to remain on.
*No wireless capability. There is a kit (Tweet-a-Watt) to add it, but it's expensive.
* It will not measure reverse power flow, making it useless where power flow could be in either direction. Grid tie inverters and induction motors operating as regenerative brakes, for instance.

I have taken mine apart, and it uses a simple analog front end connected to some sort of microcontroller. It's exactly what I expected for a cheap device. And easy to build your own.

So let's start the design:
* The fundamental frequency, for AC, is either 50Hz or 60Hz. That translates into a minimum analog bandwidth of 500Hz or 600Hz (to capture harmonics) and a minimum sample rate of 5kHz or 6kHz, not very difficult. For DC, it would only need to sample a few times a second.
* 1% or so accuracy in order to keep costs low. That is particularly important for low cost wireless sensors. The standalone bench version could have somewhat higher accuracy.
* To accommodate the wide range of current measurement, it will use a switchable current shunt amplifier. Possible implementation could be to use two opamps set to different gains, and use a multiplexer to switch between them.
* Low cost microcontroller with required A/Ds integrated, probably would also have the required multiplexer integrated. Some TI MSPs even have wireless integrated, but those are surface mount.
* Capacitive power supply for low cost AC versions, just like the Kill-a-Watt. Small switcher for other AC versions and some DC versions, linear regulator for very low voltage DC.
* Wireless versions should use an open protocol and be able to share a channel.

And as for possible versions:
* Low voltage DC with small 7 segment or text LCD, for alternative energy and battery operated circuits. Good starting point as it is low voltage and simple.
* 120V or 240V AC, 15A wireless sensor.
* 120V or 240V AC, 15A bench unit, with graphic LCD and RS-232.
* Split 240V AC, Hall effect current sensing for whole house monitoring, intended for permanent install in a breaker box.
* High voltage DC, Hall effect current sensing or high current shunt for electric vehicles and large alternative energy applications.

[cksa]

Once by chance I found  this chip..
http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en520376
I never used it, but I think that if they've done a chip just for that, it could be a good thing to take a look, perhaps they have already solved some precision issues...

I've used that chip. Quite easy to use. Integrated PGA for the current channel makes ranging easy. Quite cheap too (from farnell).

[Polossatik]

I'm not sure about your price point with a wireless interface, if you want a decent range it's going to be far more expensive (zigbee for example is $$$) and then I'm not talking yet about stuff like all the blocking you have in non-wooden houses

I think you might want to look for some form of PLC , but that of course needs a receiver that can handle multiple fazes for non-mono faze installations and a point where this can be installed (means in the actual mains switchboard), adding to the complexity of installation.

[NiHaoMike]

The wireless sensors are going to be used in the same room or nearby rooms, so 50 feet range would be enough. The whole house sensor will need a longer range, but it also has a larger budget.

One very simple and cheap method is to take a cheap FM transmitter and send the current (lowpassed to 1kHz or so) directly and the voltage on a subcarrier at 10kHz or so. The main problem is that analog transmission would result in low accuracy.

I'm thinking about having the microcontroller operate a cheap and simple FSK transmitter in the 900MHz ISM band, using Reed-Solomon ECC to reduce dropped packets.

As for PLC, I once bought 4 PLC transcievers for $1 each at a garage sale, originally intended to extend a phone line. Since I don't have an old style phone line anymore, I used a bench power supply and a resistor to simulate a line for testing. My next step was to modify the transmitter to operate as a VoIP endpoint so I can use my old analog phones for VoIP, but maybe it would be better to modify them for low speed data transmission?

[TeraHz]

NiHaoMike, how is the project going? I'd like to put together something similar and would rather not reinvent the wheel if you have something complete.