Accuracy
The accuracy of the energy monitor was tested by comparison with a cheap plug-in domestic energy monitor. A series of resistive and reactive loads were chosen to give 25 data points up to a maximum of 2238W. Voltage, Current and real power values for each load were read from display of the reference meter and the energy monitor serial terminal and logged in a spreadsheet. From which the error, percentage error and standard deviation of the error were calculated.
It was impossible to note down all the values for both meters simultaneously which is a significant source of experimental error. The values would vary significantly from second to second, and its difficult to say how much of this second to second variation is due to changes in voltage on the mains supply and load draw variations (for example as heaters heat up) or due to meter measurement resolution. It would be interesting to test the meter on a professional test apparatus.
That said the results are an improvement on the previous meter design and are as follows:
Current
The current was measured up to 9.28A. The absolute error has a max: +0.01A and min -0.03A
The standard deviation of the error: 0.011A
In the range of 0.6A to 9.28A the percentage error for current stays within 1%
Voltage
Voltage was measured in the range 241V-250V. The absolute error has a max: +3.7V and min -0.5V
The standard deviation of the error: 1.236V
The graph of voltage vs error shows a definite sloping trend which suggests the result can be improved with further calibration.
Real Power
Real power was measured in the range 13.7W - 2238W. The absolute error has a max: +9.6W and min -2.1W
The standard deviation of the error: 2.624W
These appears to be two possible experimental errors at around 560W. With out these the max absolute error would be: +3.50W
In the range of 170W to 2238W the percentage error stays within 2%
Apparent Power
Apparent power was measured in the range 39VA - 2239VA. The absolute error has a max: +6.23VA and min -7.54VA
The standard deviation of the error: 3.898VA
In the range of 150VA to 2239VA the percentage error stays within 2%
Power factor
Power factor was measured in the range 0.45 – 1.00. The absolute error has a max: +0.03 and min -0.02
The standard deviation of the error: 0.008
In the range of 0.45 – 1.00 the percentage error stays within 3.5%
non-linear loads
Have you tested the accuracy using non-linear loads like fluorescent lamps, laptop adapters, desktop pc and switching power supplies ?
I have used a killawatt, a digital multimeter and my curent sensor to measure some switching pwr supplies and all have shown very different results.
An idea for gathering data
For your future data gathering and comparing, perhaps you can take pictures or a video of the devices performing. If you can frame your shot to include the data from the Arduino and from the other power device, then you can replay the video to get your data points within the same second of each other.
Re: An idea for gathering data
Nice idea!