Low Power Contacless

Voltage Sensor for IoT Applications

Antonio Delle Femine, Daniele Gallo, Carmine Landi, Alessandro Lo Schiavo, Mario Luiso
Department of Engineering
University of Campania “Luigi Vanvitelli”
Aversa (CE), Italy
{antonio.dellefemine, daniele.gallo, carmine.landi, alessandro.loschiavo, mario.luiso}@unicampania.it

Abstract—Due to advances in the fields of low power
electronics, computer science and communications, the diffusion
of sensors based on the Internet of Things is more and more
increasing. In particular, contactless sensing principles
represent enabling technologies for the IoT. This paper presents
a new contactless power line voltage sensor (from 0 V to 300 V)
with low power consumption in order to be suitable for IoT
applications. It has been characterized with a high performance
electrical calibrator showing good accuracy (lower than 1 %) at
230 V.
Keywords—voltage measurement, contactless, low power,
sensor, internet of things
I. INTRODUCTION
The demand of smart measurement systems, and
innovative sensors based on Internet of Things (IoT) methods
has increased thanks to the advances in the fields of low power
electronics, information technology and communications [1].
It is possible to realize new devices for different applications
with big variety of sensing elements able to transmit the
information with minimal maintenance and without the
adoption of wires [2].
However, in order to allow the widespread diffusion of IoT
devices, they must be simple to use and to install. In this
context, measuring devices adopting contactless sensing
principles represent enabling technologies for the IoT.
Contactless voltage and current sensors have numerous
applications in electrical power systems. They are used for
state monitoring of high voltage transmission line [2]
appliance, load monitoring applications for smart home and
industries [3]-[8], smart metering applications [9] and for
condition monitoring and recording of electrical rotating
machines [10]. Established methods are available for non-
invasive current measurement for direct current (dc) and
alternating current (ac) applications such as Hall effect current
sensors [11], current transformers [12]-[15] with a high degree
of accuracy, etc. There are few non-invasive ac measurement
schemes using capacitive coupling based measurement of
transmission line voltage [16]-[19]. Even though some of the
cited solutions present high accuracy and some have been
employed for IoT applications, they all are characterized by a
high power consumption (related to IoT applications) which
makes them unprofitably usable for the IoT paradigm. In fact,
the high power demand makes necessary the use of a robust
power supply or, alternatively, a battery with a high capacity,
making the dimensions and the cost increasing: both of these
characteristics are against the widespread diffusion of the
solutions.
Starting from these considerations, in this paper, a new
power line contactless voltage sensor is presented. It has a
very low power consumption, compared to most solutions
presented in literature, allowing employing an energy
harvester for the power supply [20]. Moreover, it has small
dimensions and it is lightweight. All of its feature makes it
very attractive for the IoT applications. A similar approach is
presented in [19]: however, it makes use of non-linear devices
and phase shifters, which have power consumption not
suitable for IoT applications and, moreover, compromise the
accuracy of the whole device.
The paper is organized as follows. Section II discusses the
measuring principle and the architecture of the proposed
system. Section III presents the implementation and the
experimental characterization of the devices with a high
performance electrical calibrator. Finally, Section IV draws
the conclusions.
II. THE PROPOSED SYSTEM
The contactless system for power-line voltage
measurement is made up by a capacitive probe, a low-power
analog front-end and a low-power microcontroller, as shown
in Fig. 1. The contactless capacitive probe surrounds the line
cable with two metallic surfaces, the inner electrode and the
outer shield. This arrangement creates a capacitance
between the line and the electrode, a capacitance between
the electrode and the shield and a capacitance between the
shield and the earth.
As the shield is driven by the low-power analog front-end
at the same voltage of the electrode, no current flows through
the capacitance . Thus, the current flowing through the
capacitance is detected by the analog front-end in order to
determine the line voltage. Unfortunately, this current
depends not only on the line voltage, , but it also depends
on the value of the capacitance itself. In order to estimate
the value , the analog front end generates a reference signal
at frequency higher than the line frequency and injects
it into the electrode. By detecting the whole current flowing
through the capacitance , including both a line frequency
Fig. 1. Scheme of the low-power contactless voltage measurement
system.

177
978-1-7281-0429-4/19/$31.00 ©2019 IEEE
 Fig. 2. Block diagram of the low-power analog front end.

component and a reference frequency component, and by Finally, the negative peak detectors provide as output the
separating the two components, the measurement system is signal negative peaks, and , respectively.
able to estimate the value of the line voltage.
The microcontroller makes the difference between the DC
Differently from other implementations presented in voltage and , obtaining
literature [19], the analog front end is here designed in order
to minimize the power consumption, aiming at its use in a = − = (3)
wireless sensor node, possibly supplied by an energy Moreover, it makes the difference between the DC voltage
harvesting source [20]. This target is achieved by significantly and , obtaining
reducing the supply voltage down to = 3.3 , by using
only low-power operational amplifiers, MCP6242, and by = − = + (4)
devising a simpler analog processing circuit. An accurate In (3) and (4), and are the overall gains for the line
circuit design is developed in order to minimize the drawbacks signal and for the reference signal, respectively. is equal to
of the above choices, which consist in a lower signal-to-noise and it is a constant term due to the presence
ratio and in less performing integrated circuits. of the input capacitance . Equations (3) and (4) show that
The block diagram of the low-power analog front-end is the quantity , calculated by the microcontroller, depends
shown in Fig. 2. on and , while depends only on .
The oscillator, implemented through a simple Wien-bridge A calibration procedure, performed at an early stage by
circuit, generates the sinusoidal reference signal, , having disconnecting the probe to make null the capacitance ,
amplitude , frequency and a DC component = allows us to determine the value of . By subtracting this
/2, which is created by the DC level generator [21]. The value from (4), the microcontroller is able to determine
reference sinusoidal signal drives the non-inverting op-amp
= (5)
terminal, connected to the probe shield.
By using (3) and (5), the microcontroller can estimate the line
By applying the superposition principle at the voltage
voltage, that is
sources and , it is possible to calculate the output
voltage of the operational amplifier of the input stage, that is = (6)
_
=− +
where and are known quantities because they are
+ + (1) locally generated. Also and are known quantities. On
the other hand, can be measured by the microcontroller.
where is the gain resistance of the input stage and is
The initial calibration procedure can also be aimed at
the input capacitance at the inverting terminal of the determining the gain and offset errors affecting (6) and the
operational amplifier. The difference amplifier performs the resulting data can be used to correct (6) during normal
difference between the voltage in (1) and the reference
operation.
voltage. It also adds the DC voltage , in order to ensure a
proper bias. Thus, the output of the difference amplifier is From a design point of view, it should be noted that the
peak to peak voltage of signal should be lower than =
=− + 3.3 . This limits the amplitude of the signal , whose peak
+ + (2) to peak value is chosen to be equal to 2 . On the other hand,
the best accuracy is achieved if the two quantities to be
where is the gain of the differential amplifier. measured, i.e. the first two terms in the right hand side of (1),
The line-frequency low-pass filter, LPF, and the reference are comparable. Taking into account that is about 300
frequency band-pass filter, BPF, separate the two frequency times , the frequency of the reference frequency is chosen
components, generating the signals and , equal to about 300 times , i.e. = 2 15000 rad/s.
respectively. Finally, is chosen equal to the largest value that does not

178

Fig. 3. Prototype of the capacitive probe.
(a)
(b)
Fig. 4. (a) Prototype of the analog front-end with the probe. (b) Zoom of
the analog front-end circuit.
Fig. 5. Power supply of the low-power analog front end, measuring the
current drawn by the circuit.
saturate the operational amplifier with the maximum value of
.
In order to keep the signal-to-noise ratio as large as
possible, the gain of the difference amplifier is set equal to
= 3, while the gains of the LPF and of the BPF are chosen
equal to = 2. Thus, the resulting values for the overall
179
gains, neglecting the gain errors, are = = 6. In order to
ensure a suitable accuracy of the difference amplifier, this was
implemented by means of a four-resistance topology with
resistor arrays.
For the low-pass filter, a 4th order Sallen-Key filter is
chosen for its compactness and the cutoff frequency is set just
above the line frequency, in order to attenuate the undesired
line harmonics. For the band-pass filter, a 4th order
Tow-Thomas biquad filter is chosen, in order to accurately
control the bandwidth. This is a critical design aspect. As the
line noise at frequency is amplified by the derivative
behavior of the input stage, a very narrow-band is desired.
However, a too narrow band leads to stability problems.
With reference to the peak detector, the negative type is
chosen, instead of the positive type, in order to obtain an
output voltage that is lower than the DC voltage. Accordingly,
a lower reference voltage can be chosen for the A/D converter,
leading to a better quantization resolution.
Moreover, in order to allow a fast discharge of the holding
capacitor, the negative peak detectors have a reset input
controlled by the microcontroller. As the capacitor is quickly
charged by an operational amplifier and it is slowly discharged
by a resistor, the reset control allows a quicker and accurate
measurement.
III. EXPERIMENTAL RESULTS
A prototype of the capacitive probe, shown in Fig. 3, was
made with a rubber tube of 1 cm diameter and 15 cm length,
cut lengthwise to allow insertion of the line cable. The tube
was coated with silver paper both internally, to create the
probe inner electrode, and externally, to create the probe outer
shield.
A prototype of the low-power analog front-end was
implemented by means of a solderless breadboard, as shown
in Fig. 4. The value of the capacitance , measured by the
Keysight LCR meter U1733C, is 14 pF, while the value of the
capacitance is 140 pF.
The measured current drawn by the analog front-end
supply is less than 1.5 mA, as measured by the Keysight
E36313A shown in Fig. 5. Thus, the power consumption is
less than 5 mW, that is 100 times lower than the power
consumption of other analog front-ends presented in literature
for contactless power-line voltage measurement systems. In
particular, the full analog solution proposed in [19] is
characterized by a very high supply voltage and by a quite
complex circuit topology based on signal down-conversion by
mixing. Also the full digital solution proposed in [22] is
power-hungry, because it is based on a quite complex digital
signal processing (filtering, integration, scaling and so on) that
requires both a high-frequency and high-performing ADC
converters as well as a real-time signal processing by an ARM
Cortex-M4F microcontroller. On the other hand, the proposed
solution is aimed at simplifying the analog processing and, at
the same time, at minimizing the microcontroller operation,
which only performs low-frequency and low-power ADC
conversions of DC signals and very simple mathematical
operations.
The system has been characterized with a high
performance electrical calibrator Fluke 6105A (up to 1000 V,
50 A, best accuracy 50 ppm) and with a digital storage
oscilloscope (DSO) Lecroy MDA810 (12 bit, 1 GHz analog
bandwidth, 2.5 GHz maximum sampling rate). Voltages with
 Peak-peak amplitudes of filter outputs
(a)

(a)

(b)

Outputs of peak detectors and DC level

(c)

(b)
VLINE_EST [V]

(d) Error [%]

Fig. 6. Screenshots of the oscilloscope showing the main circuit signals

in case that the line voltage is (a) = 100V, (b) =
150V, (c) = 230V, (d) = 300V.

amplitudes in the range of 10 V ÷ 300 V, root mean square

(rms) value, and frequency of 50 Hz have been generated.
The output signals of the system were measured through
the DSO. Oscilloscope screenshots obtained for =
100 , = 150 , = 230 and for = (c)
300 are shown in Fig 6. The upper yellow curve is
Fig. 7. Measurement results obtained by varying the amplitude of the
proportional to the line voltage , the red curve shows the line voltage . (a) Peak to peak amplitudes of the filter
output of the LPF, , the upper blue curve the peak detector outputs, and . (b) Outputs of the peak detectors,
output, , the green curve the output of the BPF, , the and , and DC level, . (c) Estimated values of the line
lower yellow curve the peak detector output, , and the voltage and errors, obtained by using the proposed analog front-
end.
lower blue curve the continuous signal .
Numerical results obtained by varying the amplitude of the to peak amplitudes of the filter outputs, and , are
line voltage are reported in Fig. 7. In particular, the peak shown in Fig. 7(a).

180
 The DC outputs of the peak detectors, and , and
the DC level, , are reported in Fig. 7(b). The estimated
values of the line voltage calculated by (6) and corrected with
the gain and offset errors measured in the calibration stage, are
shown in Fig. 7(c). The resulting error, between the estimated
line voltage _ and the actual value , is also
reported in Fig. 7(c) and lies within a 4% band.
IV. CONCLUSIONS
This paper presents a new power line contactless voltage
sensor. Other than the possibility to measure voltage without
contact, it has the feature of low power consumption, which
makes it attractive for IoT applications.
It has been characterized with a high performance
calibrator, showing good performance over 230 V (lower than
1 %) whereas higher errors (up to about 4 %) with lower input
voltages.
Further research activities about the characterization
changing the frequency and the harmonic content of the input
voltages are still in progress.
REFERENCES
[1] A. Lo Schiavo, “Fully autonomous wireless sensor network for freight
wagon monitoring”, IEEE Sensors Journal, vol. 16, n. 24, pp. 9053 –
9063, Dec. 2016.
[2] X. Sun, Q. Huang, Y. Hou, L. Jiang and P. W. T. Pong, "Noncontact
Operation-State Monitoring Technology Based on Magnetic-Field
Sensing for Overhead High-Voltage Transmission Lines," IEEE Trans.
Power Delivery, vol. 28, no. 4, pp. 2145-2153, Oct. 2013.
[3] T. D. Huang, W. S. Wang and K. L. Lian, "A New Power Signature for
Nonintrusive Appliance Load Monitoring," IEEE Trans. Smart Grid,
vol. 6, no. 4, pp. 1994-1995, July 2015.
[4] L. Yu, H. Li, X. Feng and J. Duan, "Nonintrusive appliance load
monitoring for smart homes: recent advances and future issues," in
IEEE Instrum. Meas. Magazine, vol. 19, no. 3, pp. 56-62, June 2016.
[5] K. L. Chen, H. H. Chang and N. Chen, "A new transient feature
extraction method of power signatures for Nonintrusive Load
Monitoring Systems," in Proc. IEEE International Workshop on
Applied Measurements for Power Systems (AMPS), Aachen, Sept. 25-
27, 2013, pp. 79-84.
[6] M. Šíra and V. N. Zachovalová, "System for Calibration of
Nonintrusive Load Meters With Load Identification Ability," IEEE
Trans Instru. Meas., vol. 64, no. 6, pp. 1350-1354, June 2015.
[7] Del Prete, S., Delle Femine, A., Gallo, D., Landi, C., Luiso, M.,
“Implementation of a distributed Stand Alone Merging Unit”, (2018)
Journal of Physics: Conference Series, 1065 (5), DOI: 10.1088/1742-
6596/1065/5/052042
[8] G. Crotti, A. D. Femine, D. Gallo, D. Giordano, C. Landi and M. Luiso,
"Measurement of the Absolute Phase Error of Digitizers," in IEEE
Transactions on Instrumentation and Measurement. doi:
10.1109/TIM.2018.2888919
181
[9] D. Porcarelli, D. Brunelli and L. Bennini, "Clamp-and-Forget: A self-
sustainable non-invasive wireless sensor node for smart metering
applications," ELSEVIER Microelectronics Journal., vol. 45, no. 12,
pp. 1671–1678, Dec. 2014.
[10] G. D. Antona, P. Pennacchi, C. Pensieri and C. Rolla, "Turboalternator
shaft voltage measurements," in Proc. IEEE International Workshop
on Applied Measurements for Power Systems (AMPS), Aachen, Sept.
26-28, 2012, pp. 1-4.
[11] C. Xu, J. G. Liu, Q. Zhang, C. Xu and Y. Yang, "Investigation of the
thermal drift of open-loop Hall Effect current sensor and its
improvement," in Proc. IEEE International Workshop on Applied
Measurements for Power Systems (AMPS), Aachen, Sept. 23-25, 2015,
pp. 19-24.
[12] E. Mohns, G. Roeissle, S. Fricke and F. Pauling, "An AC Current
Transformer Standard Measuring System for Power Frequencies,"
IEEE Trans Instru. Meas., vol. 66, no. 6, pp. 1433-1440, June 2017.
[13] Crotti, G., Delle Femine, A., Gallo, D., Giordano, D., Landi, C.,
Letizia, P.S., Luiso, M., “Calibration of Current Transformers in
distorted conditions”, (2018) Journal of Physics: Conference Series,
1065 (5), DOI: 10.1088/1742-6596/1065/5/052033
[14] A. J. Collin, A. Delle Femine, D. Gallo, R. Langella and M. Luiso,
"Compensation of Current Transformers' Non-Linearities by Means of
Frequency Coupling Matrices," 2018 IEEE 9th International
Workshop on Applied Measurements for Power Systems (AMPS),
Bologna, 2018, pp. 1-6. doi: 10.1109/AMPS.2018.8494893
[15] G. Crotti, D. Giordano, A. Delle Femine, D. Gallo, C. Landi and M.
Luiso, "A Testbed for Static and Dynamic Characterization of DC
Voltage and Current Transducers," 2018 IEEE 9th International
Workshop on Applied Measurements for Power Systems (AMPS),
Bologna, 2018, pp. 1-6. doi: 10.1109/AMPS.2018.8494873
[16] K. Zhu, W. K. Lee and P. W. T. Pong, "Non-Contact Capacitive-
Coupling-Based and Magnetic-Field-Sensing-Assisted Technique for
Monitoring Voltage of Overhead Power Transmission Lines," IEEE
Sens. J., vol. 17, no. 4, pp. 1069-1083, Feb.15, 2017.
[17] K. L. Chen, Y. Guo and X. Ma, "Contactless voltage sensor for
overhead transmission lines," IET Generation, Transmission and
Distribution, vol. 12, no. 4, pp. 957-966, Feb. 2018.
[18] S. Kang, S. Yang and H. Kim, "Non-intrusive voltage measurement of
ac power lines for smart grid system based on electric field energy
harvesting," IET, Electronics Letters, vol. 53, no. 3, pp. 181-183, Feb.
2017.
[19] Shenil P.S. and B. George, "An Auto-Balancing Scheme for Non-
Contact AC Voltage Measurement," 2018 IEEE 9th International
Workshop on Applied Measurements for Power Systems (AMPS),
Bologna, 2018, pp. 1-5. doi: 10.1109/AMPS.2018.8494863.
[20] L. Costanzo, A. Lo Schiavo, M. Vitelli, “Power Extracted From
Piezoelectric Harvesters Driven by Non-Sinusoidal Vibrations”, IEEE
Trans. on Circuits and Systems – I: Reg. Papers, vol. 66, n. 3, pp. 1291
– 1303, March 2019.
[21] A. Buonomo, A. Lo Schiavo, “Analysis and Design of Dual-Mode
CMOS LC-VCOs”, IEEE Trans. on Circuits and Systems – I: Reg.
Papers, vol. 62, n. 7, pp. 1845 – 1853, July 2015.
[22] M. A. Haberman, E. M. Spinelli, "Noncontact AC Voltage
Measurements: Error and Noise Analysis," in IEEE Transactions on
Instrumentation and Measurement, vol. 67, no. 8, pp. 1946-1953, Aug.
2018.