ADBU-Journal of Engineering Technology

Experimental Performance Assessment of a low-cost

single-phase Arduino-based Power Meter
Loganaden Madalamootoo1, Heman Shamachurn2
1,2Department of Electrical and Electronic Engineering, University of Mauritius
Reduit 80837. MAURITIUS
[email protected]
[email protected]

Abstract: A single-phase, low-cost, plug and play, portable and wireless meter has been
implemented to measure the electrical parameters of typical household electrical loads
ranging from a few watts to a few kilowatts. ADE7753 smart measuring IC was employed to
acquire load current and voltage information through sensors. The Arduino Mega 2560
board, alongside the Arduino Wi-Fi shield were used for wireless communication and for
logging of information on an external memory card. Measurements performed on eight
typical office loads using the proposed meter and a calibrated Fluke 345 power quality
clamp meter revealed an acceptable performance for most of the considered electrical
appliances. The absolute percentage error in measured active power and cumulative energy
consumption were less than 5 % for most of the considered appliances. The total cost of the
system was 139 USD, which makes it affordable to measure the power consumption of
different typical domestic loads.

Keywords: Power meter, Arduino, Fluke 345, Single-phase, ADE7753

(Article history: Received: 5th August 2017 and accepted: 25th April 2018)

Several state-of-the art electronic meters have

1. Introduction been proposed in literature. A Wi-Fi smart plug for
The built environment consumes a significant 40 monitoring the energy consumption of remote
% of the total energy demand worldwide. Intelligent electrical appliances via Web, as well as switching
buildings are becoming more popular as they have them ON/OFF via consumer interaction was
the capability to monitor, manage and control many developed using the ESP-WROOM02
aspects to improve their energy performance. microcontroller and STPM01 chip [1]. A smart plug
Moreover, as the electricity infrastructure migrates to provide power reduction without turning off the
towards smart grids, conventional electromagnetic load was investigated in [2]. A power meter based on
based energy meters are being replaced by state of Arduino, current and voltage transformers was
the art electronic systems with advanced proposed in [3] and [4]. A low-energy consuming
functionalities in order to extract more useful electric energy meter reader based on
information about the energy consumption than MSP430FG4618 controller and comprising a module
merely numerical data. For instance, most for communication through SMS was proposed by
conventional meters have no ability to provide [5]. An Arduino and Ethernet-based smart plug
detailed consumption information for specific hours, which provides real time update of the energy
days or weeks, given that only accumulated energy consumption at the device level is proposed in [6].
use data is available. Moreover, the absence of An electronic energy meter using the ADE7751
communication option leads to the tedious, expensive metering IC and having the capability of automated
and time consuming task of manual meter reading progression of measurement through wireless
which can be erroneous and can involve corruption. communication was proposed in [7]. The authors of
Traditional metering techniques consisted of [8] presented a low-cost wireless GSM energy meter
measuring the energy consumption of an entire and a web interface for automatic billing and
building. Nowadays, it can be more advantageous to management of collected data. The MCP3905A
monitor the energy performance of individual devices metering IC, an 8-bit PIC16F877A Microcontroller, a
or a group of devices for a more effective energy GSM modem and the EEPROM Atmel 24C256
management. external memory were used. An Arduino-based

Madalamootoo, AJET, ISSN:2348-7305, Volume 7, Issue 1, June, 2018,007010604 (7PP) 1

ADBU-Journal of Engineering Technology

power factor metering system was proposed in [9], 2500:1 was selected for this application as it is
whereby it was concluded that the system does not relatively cheap, it is ideal for noisy environment, it
perform well with predominantly inductive loads provides galvanic isolation and it requires no external
made up of lamps. A wireless current sensor node power supply. Based on the maximum current to be
was proposed in [10] and [11], but the issue is that measured by the system, the calculated burden
voltage was not measured and therefore, devices’ resistance RB to be connected to the CT was found to
performances could not be fully described. An be 200 Ω. A low-pass filter was included to reduce
Arduino-based smart meter employing ADE7753 the noise on the secondary of the CT. A 240 V AC to
metering IC, and having information storage and 12 V AC voltage transformer (VT) was used to
wireless transfer capabilities was proposed in [12], measure the supply voltage and to provide galvanic
but limited investigation was carried out on the isolation between the high voltage on the supply side
performances of the device. and the low voltage side of the electronic circuitry.
Many of the cited works tested the developed
meters on simple loads such as bulbs, but other 2.2. Current and Voltage Measurement
common loads in a typical domestic context have
been disregarded. Moreover, very few studies exist The ADE7753single-phase multifunction
whereby the performance of the proposed meter has metering IC was used in this work.The chip has 2
been compared against a calibrated industry-grade input channels; one for the current and one for the
meter. Furthermore, for cases where the limited voltage. Each channel is equipped with a 16-bit Σ-Δ
results have been cross checked with other reliable analogue to digital converter (ADC). The IC
measurements, no investigation was carried out to incorporates RMS calculation for the voltage and
identify causes of the errors so that improvements current. Channel 1 is used to measure current and
can be carried out. As per the authors’ knowledge, so channel 2 is used to measure voltage. Given that the
study so far has critically assessed the performance of maximum differential voltage supported by the
a designed wireless low cost meter based on voltage channel was ± 0.5 V and that the output
ADE7753 IC and Arduino with appliances having voltage of the VT was 12 V, a suitable voltage
different power ratings, power factors and involving divider circuit was designed to maintain the voltage
switch-mode power supplies. The contribution of this within the required range. A low pass filter was
paper is to address the above knowledge gap. added to reduce noise.

2. Methodology 2.3. Microprocessor

The block diagram of the developed system is To be able to extract the processed information
shown inFigure 1. The domestic utility supply from the ADE7753 through Serial Peripheral
voltage of the island of Mauritius is 230 V ± 6 % at a Interface (SPI) and to provide wireless
frequency of 50 Hz and the earthing system is of T-T communication function, Arduino Mega 2560 was
type. selected. This microprocessor is known to be cheap,
reliable and has many well-known applications such
as power system substation monitoring and control
[14]. The Arduino Wi-Fi shield was used as add-on
board to allow the Arduino Mega to connect to the
internet wirelessly using the 802.11 protocol. An on-
board microSD card reader allows storage of
information on an external memory. Arduino Mega
communicates with both the Wi-Fi shield’s processor
and the SD card using the SPI bus. The shield is
powered by the Arduino Mega, which is itself
powered by an independent 5 V power supply.

2.4. Circuit Implementation

Figure 1.Block diagram of proposed meter
After simulations on Proteus software, sections
2.1. Current and Voltage Measurement of the entire circuit were practically tested to confirm
The ZMCT116A [13]CT having a turns ratio of
adequate operation. Figure 2 and Figure 3 show the

Madalamootoo, AJET, ISSN:2348-7305, Volume 7, Issue 1, June, 2018,007010604 (7PP) 2

ADBU-Journal of Engineering Technology

connections to the voltage and current channels of the microprocessor to fetch the different parameters from
ADE 7753 respectively. Figure 4 shows how the the ADE7753; setting up the connection between the
different components were assembled excluding the Wi-Fi shield and the network; and finally logging the
microprocessor, and Figure 5 shows the portable data to an excel file on the SD card. Logged
meter. parameters were RMS voltage, RSM current, active
power, apparent power, power factor, and cumulative
active energy consumption. Though the sampling
mode of the ADE7753 was 14 kHz, data was
obtained by the Arduino Mega from the ADE7753 at
a frequency of 100 Hz as per the Nyquist-Shannon
sampling theorem [12] and then averaged to get the
parameter value for 1 s for the RMS voltage, RSM
current, active power and apparent power. The
Figure 2. Connection to voltage channel average power factor for 1 was then calculated by the
ratio of average active power to the average of
apparent power, given that the ADE7753 does not
have special registers for the power factor. The
energy for each time period was finally obtained by
the product of the active power and the time period.

2.6. Testing
The developed meter was tested on some
commonly available electrical appliances: Light bulb,
kettle, fan, laptop, soldering iron, refrigerator,
computer (PC) power supply, and LED PC screen.
Figure 3. Connection to current channel The appliances are essentially different types of
electrical loads. For instance, the bulb was an
incandescent lamp, which is basically a resistive load.
The refrigerator contains an induction motor, which
is predominantly inductive. Most of the laptops and
PC screens make use of switch mode power supplies.
The duration of the test for each load varied
depending on the availability of the equipment.
Moreover, using the 1 s data, averaging over different
time periods were obtained as shown in TABLE1.

TABLE1. Measurement scenarios

Testing Measurement
Electrical Load duration averaging period
Figure 4. Circuit implementation (s) (s)
Light Bulb 720 45; 90
Fan 540 45
Kettle 420 30; 60
Refrigerator 720 45; 90
Soldering Iron 16 1
Laptop 720 45; 90
PC power supply 720 45; 90
PC screen 720 45; 90

Figure 5. Completed circuit Measurement results were compared to those

provided by a calibrated Fluke 345 Power Quality
2.5. Software design clamp meter, which has a sampling frequency of
The software design comprised 3 parts: 15.625 kHz. An averaging time of 1 s was selected to
establishing the SPI communication for the match the averaging time period of the developed

Madalamootoo, AJET, ISSN:2348-7305, Volume 7, Issue 1, June, 2018,007010604 (7PP) 3

ADBU-Journal of Engineering Technology

meter. Figure 6 shows how measurements were % Average voltage error

performed on the PC screen using both the developed 3
and the Fluke meters. The measurement from the 2
Fluke 345 meter was taken to be the true value and 1
the percentage measurement error of the proposed 0
meter was calculated as per equation (1). -1
-2
DMM − FMM -3
% Error = X 100 % (1)

Light Bulb (45 s)

Light Bulb (90 s)
Fan (45 s)
Kettle (30 s)
Kettle (60 s)
Refrigerator (45 s)
Refrigerator (90 s)
Soldering Iron (1 s)
Laptop (45 s)
Laptop (90 s)
PC supply (45 s)
PC supply (90 s)
PC screen (45 s)
PC screen (90 s)
FMM

Where DMM is the Developed Meter Measurement

and FMM is the Fluke Meter Measurement.

Figure 7. % Voltage Error

% Average current error
25
20
15
10
5
0
-5
-10 Light Bulb (45 s)
Light Bulb (90 s)
Fan (45 s)
Kettle (30 s)
Kettle (60 s)
Refrigerator (45 s)
Refrigerator (90 s)
Soldering Iron (1 s)
Laptop (45 s)
Laptop (90 s)
PC supply (45 s)
PC supply (90 s)
PC screen (45 s)
PC screen (90 s)
Figure 6. Testing on LED PC screen

The average error for each parameter is the mean

of all the measurement errors for that corresponding
parameter over the considered testing duration. The
Wi-Fi transmission of the meter over a university
network was eventually tested through a laptop and a Figure 8. % Current Error
mobile phone to verify the connection. The system
was assigned a special IP address of 192.168.100.8 % Average active power error
and it was verified if a web browser can display load 30
measurement information. 25
20
15
3. Results and Discussion 10
5
The average percentage measurement error for 0
each parameter is shown in Figure 7 to Figure 12 and -5
Light Bulb (45 s)
Light Bulb (90 s)
Fan (45 s)
Kettle (30 s)
Kettle (60 s)
Refrigerator (45 s)
Refrigerator (90 s)
Soldering Iron (1 s)
Laptop (45 s)
Laptop (90 s)
PC supply (45 s)
PC supply (90 s)
PC screen (45 s)
PC screen (90 s)

each figure shows the error for all the measurement

scenarios as per TABLE1.

Figure 9. % Active Power Error

Madalamootoo, AJET, ISSN:2348-7305, Volume 7, Issue 1, June, 2018,007010604 (7PP) 4

ADBU-Journal of Engineering Technology

% Average apparent power error measurement, 0.3 % to 23.9 % for apparent power
30 measurement, 0.4 % to 23.4 % for power factor
25 values, and 0.1 % to 27.8 % for cumulative energy.
20
15 The voltage measurements were relatively close for
10 the Fluke and the developed meter, with positive and
5
0 negative errors. Similarly, currents were measured
-5 with a decent level of accuracy for most of the loads.
-10 The measured current, voltage, powers and power
Light Bulb (45 s)
Light Bulb (90 s)
Fan (45 s)
Kettle (30 s)
Kettle (60 s)
Refrigerator (45 s)
Refrigerator (90 s)
Soldering Iron (1 s)
Laptop (45 s)
Laptop (90 s)
PC supply (45 s)
PC supply (90 s)
PC screen (45 s)
PC screen (90 s)
factor by the developed meter tended to vary
frequently and by larger amounts compared to the
measurements by the Fluke meter. Errors in current
and voltage caused error propagation when combined
to get the apparent power. Significant errors (> 5 %)
were obtained in the cumulative energy consumption
Figure 10. % Apparent Power Error for the PC screen and PC power supply, due to large
errors in the current and power factor measurements.
When considering the different measuring time
periods for the same test appliance, no definite
% Average power factor error conclusion can be drawn. The errors in active power
25
20 and cumulative energy being less than 5 % suggest
15 that the developed meter is providing sufficient
10 accuracy for most of the considered loads. The
5
0 average error for all the scenarios, all the parameters
-5 and all the appliances was 5.0 %. Figure 13shows the
-10 screen shot of a mobile phone accessing the
Light Bulb (45 s)
Light Bulb (90 s)
Fan (45 s)
Kettle (30 s)
Kettle (60 s)
Refrigerator (45 s)
Refrigerator (90 s)
Soldering Iron (1 s)
Laptop (45 s)
Laptop (90 s)
PC supply (45 s)
PC supply (90 s)
PC screen (45 s)
PC screen (90 s)

measurement information from its browser. Figure 14

shows a screenshot of the Wi-Fi transmission on a
laptop screen. Detailed electrical parameters for the
light bulb connected to the metering system were
displayed and the page would refresh every 10
seconds so as to get new data accordingly.
Figure 11. % Power Factor Error

% Average cummulative energy error

30
25
20
15
10
5
0
-5
Light Bulb (45 s)
Light Bulb (90 s)
Fan (45 s)
Kettle (30 s)
Kettle (60 s)
Refrigerator (45 s)
Refrigerator (90 s)
Soldering Iron (1 s)
Laptop (45 s)
Laptop (90 s)
PC supply (45 s)
PC supply (90 s)
PC screen (45 s)
PC screen (90 s)

Figure 13. Viewing measured parameters through

Wi-Fi on a mobile phone browser

Figure 12. % Energy Error

The percentage absolute average error (PAAE)

varied in the ranges of 0.1 % to 2.6 % for voltage
measurement, 0. % to 19.2 % for current
measurement, 0.2 % to 23.9 % for active power

Madalamootoo, AJET, ISSN:2348-7305, Volume 7, Issue 1, June, 2018,007010604 (7PP) 5

ADBU-Journal of Engineering Technology

Optimize Usage of the Electric Power," Cartagena,

Colombia, 2014.
[5] A. A. Adhau, N. M. Patel, A. T. Zaidy, S. L. Patil and A. S.
Deshpande, "Low Cost Electricity Meter Reading System
using GSM," Tamilnadu Nagercoil, India, 2013.
[6] A. H. Shajahan and A. Anand, "Data acquisition and control
using Arduino-Android Platform : Smart plug," Nagercoil,
India, 2013.
[7] T. Maity and P. S. Das, "Intelligent Online Measurement
and Management of Energy Meter Data Through Advanced
Wireless," Mesra, Ranchi, India, 2011.
[8] K. Ashna and S. N. George, "GSM BasedAutomatic
Figure 14. Viewing measured parameters through EnergyMeter Reading System with Instant Billing,"
Wi-Fi on a web browser Kottayam, Kerala, India, 2013.
[9] P. Machado Jr, T. Abud, M. Z. Fortes and B. Borba, "Power
factor metering system using Arduino," in 2017 IEEE
4. Conclusion Workshop on Power Electronics and Power Quality
Applications (PEPQA), Bogota, Colombia, Colombia, 2017.
A single-phase, low-cost, and plug and play, [10] D. Porcarelli, D. Balsamo, D. Brunelli and G. Paci,
portable and wireless power meter has been "Perpetual and low-cost power meter for monitoring
residential and industrial appliances," in 2013 Design,
developed in this work. The meter has the ability to Automation & Test in Europe Conference & Exhibition
measure, log and transmit via Wi-Fi essential (DATE), Grenoble, France, 2013.
electrical parameters of an electrical load such as [11] A. Reinhardt, D. Burkhardt, P. S. Mogre, M. Zaheer and R.
Steinmetz, "SmartMeter.KOM: A low-cost wireless sensor
RMS voltage, RMS current, active power, apparent for distributed power metering," in 2011 IEEE 36th
power, power factor and cumulative energy Conference on Local Computer Networks, Bonn, Germany,
consumption. It has been tested with different types 2011.
[12] C. Klemenjak, D. Egarter and W. Elmenreich, "YoMo: the
of commonly available electrical loads and the results Arduino-based smart metering board," Computer Science -
were compared to measurements provided by a Research and Development, vol. 31, no. 1-2, pp. 97-103,
calibrated Fluke 345 meter. For most considered test 2016.
[13] L. Nanjing Zeming Electronics Co., "ZMCT116A Ultra-
loads, the obtained errors in average power micro current transformer--Nanjing Zeming Electronics Co.,
measurement and the cumulative energy are less than Ltd," Nanjing Zeming Electronics Co., Ltd, unknown
5 %. It is important to note that the Fluke meter has unknown unknown. [Online]. Available:
accuracy values associated with all its measured http://www.zeming-e.com/English/prodviewtype5-255.html.
[Accessed 04 August 2017].
parameters as per its datasheet[15]. Therefore, the [14] A. Talukdar, K. Laishram, K. Tayu, J. K. Barman and G.
actual parameter values can be lower or higher than Das, "Design of RTU & SCADA," ADBU-Journal of
those measured by the Fluke meter. In the near Engineering Technology, vol. 5, no. 2, 2016.
[15] Fluke, "Fluke 345 Single Phase Power Quality Clamp
future, the meter will be compared to a much more Meter," Fluke, unknown unknown unkown. [Online].
accurate measurement system based on LabVIEW, Available: http://en-us.fluke.com/products/power-quality-
better current and voltage sensors and more analyzers/fluke-345-clamp-meter.html#techspecs.
[Accessed 5 August 2017].
sophisticated energy metering IC. Moreover, a three- [16] L. Nanjing Zeming Electronics Co., "ZMCT116A Ultra-
phase metering system will be developed. On overall, micro current transformer--Nanjing Zeming Electronics Co.,
it can be deduced that the developed meter, being Ltd," None, Unknown Unknown Unknown. [Online].
Available: http://www.zeming-
cheap and sufficiently accurate, can be applied in the e.com/English/prodviewtype5-255.html. [Accessed 19 April
efficient building electrical loads management and 2017].
control.
[17] A. Devices, "Search | Analog Devices," Unknown Unknown
Unknown. [Online]. Available:
http://www.analog.com/en/search.html?q=ADE7753.
References [Accessed 19 April 2017].
[18] Arduino, "Arduino - ArduinoBoardMega2560," Unknown
[1] Y. Thongkhao and W. Pora, "A Low-cost Wi-Fi Smart Plug Unknown Unknown. [Online]. Available:
with On-off and Energy Metering Functions," Chiang Mai, https://www.arduino.cc/en/Main/ArduinoBoardMega2560.
Thailand, 2016. [Accessed 19 April 2017].
[2] O. ELMA and U. S. SELAMOĞULLARI, "A Home Energy [19] Arduino, "Arduino - ArduinoWiFiShield," Unknown,
Management Algorithm with Smart Plug for Maximized Unknown Unknown Unknown. [Online]. Available:
Customer Comfort," Sharjah, United Arab Emirates, 2015. https://www.arduino.cc/en/Main/ArduinoWiFiShield.
[3] N. Tamkittikhun, T. Tantidham and P. Intakot, "AC Power [Accessed 19 April 2017].
Meter Design for Home Electrical Appliances," Hua Hin,
Thailand, 2015.
[4] M. Maryori Sabalza, D. Jes´us, O. Borr´e and J. C. M.
Santos, "Design and Construction of a Power Meter to

Madalamootoo, AJET, ISSN:2348-7305, Volume 7, Issue 1, June, 2018,007010604 (7PP) 6

ADBU-Journal of Engineering Technology

Author Profile

H. Shamachurn, received his BEng (Hons) degree in

Electrical and Electronic Engineering from the University
of Mauritius in 2010.He received his MSc degree in
Renewable Energy Systems Technology with distinction
from Loughborough University in 2012. He is presently a
Lecturer in the Department of Electrical and Electronic
Engineering at the University of Mauritius. His research
interests include Renewable Energy, Energy Management
and Energy Efficiency.

Madalamootoo, AJET, ISSN:2348-7305, Volume 7, Issue 1, June, 2018,007010604 (7PP) 7