Archived Forum

Hello Klapauzius

I have also been looking at this problem.  Recent spot measurements indicate to me that it IS ok to measure one leg only.  Typically variations are less than 0.1V (in 120V) between the two phases (maybe less, this is the resolution of my voltmeter).  The nominal voltage (120V) goes up and down during the day due to the local (street, town) loading (A/C during the day, mainly), but both phases do move together.  If your house (or group of houses) had a very severe load imbalance across the phases then the voltages could differ but most heavy load items (tumble driers, A/C, pool pumps etc are connected across the two phases), and I believe this effect is small.

For those unfamiliar with the US system, the two phases are 180 degrees to each other, giving 120V+120V = 240V nominal.  The neutral is the common between these two phases.

I have been trying to figure out how to use the EmonTx with the US system.  I have come to the following conclusions: measuring the voltage on one phase is perfectly fine, as described above.  Using two current transducers, one for each phase, will work, however one current transducer will need to be reversed to account for the voltage signal being 180 degrees out relative to the phase used to measure the voltage.  This should not require any software mods - yippee!  However I have actually not had time to try this out.  I would like to measure the two phases (current, power, phase angle), and then also sum the two real power values to get the total house power consumption (this may require some software mods, or may be available via one of the PV system configurations (I have not had time to look at this).  Measuring the power for a 240V appliance will work, the current is measured twice (once in each supply leg), but the voltage is only half (120V), so the answer is correct - and have of the total power will be logged by each phase.

I hope this helps?  Please let me know how you get on.

Cheers

AndyB

I am in the UK and have no experience of the US system, but I think your (AndyB) take on the problem is correct. But as klapauzius pointed out, it depends on the two phases being balanced so that both are at the same voltage. Ideally, you would measure two voltages and two currents. It is reassuring to hear that your experience indicates that the assumption that balance is generally achieved is valid, and the errors resulting from measuring just one phase will be minimal. If as you say the voltage error is about 0.1 V, then the overall error is about 0.1%. I don't think I would worry about that, I doubt that your official supplier's tariff meter is better!

As regards the software, if you use the standard EmonLib routines you will measure voltage and current for 10 cycles (default) on the first phase, then the same voltage and the second current for the next 10 cycles. If that is a concern, you might want try creating a new library routine to read voltage and two currents at the same time and calculate the total power directly.

Thanks for the advice!

I am still waiting for the 2nd CT to arrive...I had to get some with a 2.5 cm ( 1")  opening to fit around the rather bulky mains lines going into my main panel.

To complicate matters, I have a PV system on one leg, but that should be fine, as the solar system measures its own power and I am really only interested in the net consumption.

I am using the arduino with an old WiShield 1.0, this is a WiFI shield, that allows me to write data directly into an IP socket.

Since the shield actually eats quite a bit of clock cycles, in terms of calculations I cannot do too much on the arduino itself, so I merely transmit the rms voltage, rms current in both legs, timestamp and the "real" power to know the direction of the flow.

I am mostly using my own functions for this...

I plan to do all subsequent data processing on a PC...actually a Rasperry Pi (which I should hopefully get in a few weeks).

As for the hardware:

I am using a little bit of analog signal conditioning before feeding it to the Arduino adc. One quad op amp (Lm324) serves fine for this purpose, i.e. 1 omp amp to provide a 2.5 V reference (I think this is better than the current resistor voltage divider), 2 op amps as inverted amplifiers  (10x) for the CTs. The last one could be used to buffer the voltage measurement. I did not, and it seems to work fine, but would probably be more robust to have it.

I also added to 60 Hz (corner) RC filters to the op-amp output for the CTs...not sure how I am going to fare with highly non-linear loads, but actually this should improve things a bit for the harmonics.

Just a short follow up:

2 CTs, one on each leg, and just one Voltage measurement near the main panel seems to work. Measured power

and energy seems to be close enough to the utility meter. Voltage on the one leg I measure typically varies  0.3 - 0.5 V around the mean (measured over a day). Resolution is 0.3 V/AD unit and the max and min are somewhere between 119 V and 122 V. More importantly , there is no significant correlation between load and voltage.

Total power is then simply V*(I1+I2), where the sign of I1 and I2 can determined from the power factor of each leg. Since I have solar on one, this can be actually negative. Note that V, I1 and I2 here mean RMS.

 Very interesting read.  

I was wondering which AC-AC transformer you were using.  I found an old Linksys Mdl. AD9/1C transformer that works but does not quite agree with the measured current and voltage on my Kill-A-Watt.  Calibration information is not listed anywhere on the openenergymonitor.org website for the AD9/1C.  The only transformers here are ones sourced in the UK with calibration factors for their mains voltage.   Is there some way to use a Kill-a-Watt, or even better a calibrated power measurement tool to determine the correct calibration value?  Sorry if I missed something in a FAQ/HOWTO...

Thanks in advance,

Marshall

See the installation and calibration page here: http://openenergymonitor.org/emon/buildingblocks/ct-and-ac-power-adaptor...

You need an accurate measurement of your local mains voltage and simultaneously (or as close as possible!) a measurement of your transformer output voltage - then substitute your numbers for "230" and "9 + 20%".  At the end of the day, if you can measure voltage and power accurately some other way, you still need to tweak the calibration factors to give the correct answers (because the theoretical calibration factors cannot take account of resistor tolerances that could still leave you in error by 10% or so).

I changed the correction factor based on what my Kill-a-Watt and multimeter were saying the the wall voltage is and now the reported RMS voltage from the emontx is spot-on.    

Thanks for the help!

Just to follow up on subject I have managed to do the same as you using the following for the USA market.

EmonTx

enable all 3 CT ´s in the Code and adjust Calibration factor to suit (measuring only one leg (AC-AC) adpater)

CT2=1

CT3=1

EmonGLCD (script EmonGLCDsolar)

change code to use Solar Inverter  on CT2 and double up in power calculation function measuring only one leg (since inverter balances that should be good enough)

gen = 2*emontx.power2;

Now Im using type 2 where I have 2 clips CT1 and CT3 which measure Grid Import and Export make sure clips are reversed relative to each other and show a Negative value when there is no Consumption just production from CT2.

  if (SolarPV_type==2)
  {
    grid=emontx.power1+emontx.power3;  // this makes sure that you have both CT1 and CT3 contributing to the grid measurement and CT2 for solar production doubled up.

ect....

and if you have EmonBase/NanodeRF (script NanodeRF_Power_RTCrelay_GLCDtemp)

Modify script to include both grid Legs sent to one input and double up on solar again.

  // JSON creation: JSON sent are of the format: {key1:value1,key2:value2} and so on
         
          str.reset();                                                   // Reset json string     
          str.print("{rf_fail:0");                                       // RF recieved so no failure
          str.print(",grid:");        str.print(emontx.power1+emontx.power3);          // Add grid reading  -ve when sun is out !!
          str.print(",solar:");        str.print(2*emontx.power2);          // Add solar power production reading
          str.print(",voltage:");      str.print(emontx.voltage);        // Add emontx grid voltage reading

Hope this Helps anyone trying to use this setup using USA style grids.

John

I just ordered a kit for installing in my house in the US to measure energy output of my grid-connected PV. Anyone in the forum who has done that in the US? I have ordered the following from the openenergy monitor shop. Can anyone guide me on how to get started.

Home sysem is a 10 KW usign Suntech 185 panels with PV Powered inverters. Standard net-metering configuration in New Jersey.

monTx Case - un-milled

Encapsulated DS18B20 temperature sensor kit

NanodeRF SMT 433Mhz - Pre Assembled

Programmer - USB to serial UART

Solar PV Monitor Kit Bundle 433Mhz UK plug
(emonTx 433Mhz: emonTx 433Mhz kit , CT 1: 100A max clip-on current sensor CT, CT 2: 100A max clip-on current sensor CT, AC-AC UK: AC-AC adapter - AC voltage sensor (UK plug), emonGLCD 433Mhz: emonGLCD 433Mhz kit)

What do you need to know that is not covered in the thread above, the various other threads about the US system, the assembly guides under Modules and in the articles in Building Blocks?

I guess you'll need to buy a couple of standard UK mains sockets and a few 5 V USB power supplies as well. Take care and get good quality ones (weird problems have been reported with poor quality 'noisy' ones). Here is a blog about 5 V USB power supplies.