Power optimisation 

For the emonTx to run of two AA's for as long as possible the Mains AC: non-invasive 3.0 energy monitoring circuit and code was optimized for low power. See below for brief account of changes. 

 

 

1. Increase CT circuit voltage divider resistor values from 10K to 470K

This reduced quiescent power consumption across the resistors from 0.25mA to 5.3uA at 5V

2. Reduce number of samples from 6000 to 1490

This reduced the time needed for a reading from over 1s to 332ms. As the number of samples are reduced it is important that the time taken for the samples is an integer of the period of the waveform.

The period (one wavelength) of mains 50Hz is 20ms. Each samples was measured to take 0.188ms. This means that 106.4 samples/wavelength are possible. 1480 samples takes 280.14ms which is 14 wavelengths. 

3.  Put Atmega 328 into sleep mode in between readings 

We followed this example to use the watchdog timer to put the device into sleep mode for 4 seconds (watchdog value 8)  in between readings. In sleep mode the ATmega 328 consumes 0.03mA with an RF transmitter and CT monitor circuit. The power consumption during readings is 13mA

 

Rough battery life estimation calculations

Therefore the device is on for 332mS consuming 13mA then asleep for 4s consuming 0.03mA

This gives an on off ratio of 4/0.332=12. The battery life should be extended by this amount. 

An AA alkaline battery has between 2000-3000 mAhrs of capacity. If the device was not put into sleep mode the battery would last for 2500/13=192hrs=8 days.

Since the device is in sleep mode for a proportion of this time the average power consumption is 13/12=1.083mA plus the power consumption while in sleep mode of 0.03mA

This should give a battery life of 2500/(1.083+0.03)=2246hrs= 93.6 days = 3 months

In reality the battery will not last this long, at least not using two AA's since the Atmega 328 will stop working due to low voltage before the batteries are depleted. Using three AA's would be better but our enclosure does not allow for this. The ATmega 328 should work down to 1.8V. 

 

Real world battery life test 

A real-world test was setup to test the above calculation.

 

Here was the result: 

The test actually ran for 5 months from two AA batteries! A great success.  The test setup actually used a pair of simple RF modules not the RFM12B transceiver module which was used in the final design. According to the datasheets the power consumption for the two RF modules are similar, if not slightly less. We are therefore expecting battery life to be similar, or even slightly longer. 

Burden Resistor Power Dissipation 

Normal expected peak current through 18 ohm burden resistor when using SCT-013-000 CT measuring 100A  = 70.7mA

Peak voltage = VCC/2 = 3.3/2= 1.65V

Peak power dissipation in burden resistor = Peak V * Peak I = 1.65 * 70.7 =115.5mW = 0.1155W
 
Fault condition peak current through burden resisor @ 260A+ primary = 90mA (saturation limit + safety margin) see here
Peak power dissipation in burden resistor = 90mA * (3.4V/2)= 153mW
 
The conclusion is that at a 1206 SMT resistor can safely be used as the burden resistor. A 1210 resistor would give a bit more of a margin.