Multiple Partial Discharge Source Localization in Power Cables Through Power Spectral Separation and Time-Domain Reflectometry

G. Robles, M. Shafiq and J. M. Martínez-Tarifa, “Multiple Partial Discharge Source Localization in Power Cables Through Power Spectral Separation and Time-Domain Reflectometry,” in IEEE Transactions on Instrumentation and Measurement. doi: 10.1109/TIM.2019.2896553

Open access post-print version (ie final draft post-refereeing) available (Copyright 2019 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works).

Early access available at


Abstract— Insulated power cables are becoming increasingly popular in today’s developing distribution and transportion networks. However, due to aging, deterioration, and various operational and environmental stresses, insulation defects may appear and so the cable needs to be monitored in a timely manner to avoid unexpected failures. Many of these defects are responsible for partial discharge (PD) activity. The localization of the sources of these discharges is a highly decisive facet in the condition-based monitoring of power cables. The techniques for the localization of single-PD defects in insulated power cables are well presented in the current bibliography. However, when several simultaneous PD sources are active, the localization of the sources becomes quite complex. This paper develops an efficient technique for the separation and localization of multiple PD sources in a medium voltage cable. The experimental results are obtained with single-end-based measurements using a high-frequency current transformer in a laboratory environment. The data processing based on the spectral characteristics of the signals is carried out by using the power ratios technique in order to determine the presence of different types of PD. Once the signals are separated, the PD sources can be localized with an individualized analysis of each source through time-domain reflectometry. The proposed methodology can be very valuable to improve the location diagnostic capability of the condition-based monitoring solutions, especially for underground cables.

Keywords— Condition monitoring; partial discharges (PDs); particle swarm optimization (PSO); power cables; signal characterization; signal propagation; spectral power ratios (PRs); time-domain reflectometry (TDR).

Designing a Rogowski coil with particle swarm optimization

Guillermo Robles; Muhammad Shafiq; Juan Manuel Martínez-Tarifa, Designing a Rogowski coil with particle swarm optimization, November 2018, Proceedings of the 5th International Electronic Conference on Sensors and Applications session Physical Sensors (doi: 10.3390/ecsa-5-05721)

Open access at

Abstract—Rogowski coils are inductive sensors based on Faraday’s and Ampère’s Laws to measure currents through conductors without galvanic contact. The main advantage of Rogowski coils when compared with current transformers is the fact that the core is air so they never saturate and the upper cut-off current can be higher. These characteristics makes Rogowski coils ideal candidates to measure high amplitude pulsed currents. On the contrary, there are two main drawbacks. On the one hand, the output voltage is the derivative of the primary current so it has to be integrated to measure the original signal; and, on the other hand, the transfer function is resonant due to the capacitance and the self-inductance of the coil. The solution is the use of a passive integration with a terminating resistor at the output of the sensor that splits the two complex poles and gives a constant transfer function for a determined bandwidth. The downside is a loss of sensitivity. Since it is possible to calculate the electrical parameters of the coil based on its geometrical dimensions, the geometry can be  adapted to design sensors for different applications depending on the time characteristics of the input current. This paper proposes the design of Rogowski coils based on their geometric characteristics maximizing the gain-bandwidth product using particle swarm optimization and adapting the coil to the specific requirements of the application.

Keywords—Rogowski coils; particle swarm optimization; gain-bandwidth product; current
measurement; magnetic field measurement.


20 puestos de especialización para jóvenes ingenieros y físicos aplicados en la 4ª convocatoria del Spanish Traineeship Programme, CIEMAT-CERN

En los próximos días, se abrirá la cuarta convocatoria del Spanish Traineeship Programme, FTEC-2018, un programa de especialización tecnológica en el CERN, Ginebra, Suiza, destinado a jóvenes ingenieros y físicos aplicados.

La convocatoria tiene como objetivo incrementar la presencia de investigadores y técnicos españoles en el CERN, así como consolidar un colectivo de ingenieros y físicos especializados en tecnologías de los grandes aceleradores de partículas, detectores e infraestructuras asociadas, con la finalidad de una futura incorporación a la industria e instituciones del sector.

Podéis encontrar más información en:


Localización de Fuentes de Descargas Parciales en Instalaciones Eléctricas


UNIVERSIDAD DE CANTABRIA – 27 de abril de 2017

José Manuel Fresno, Guillermo Robles, y Juan Manuel Martínez-Tarifa.  E-Mails:, y

Departamento de Ingeniería Eléctrica. Universidad Carlos III de Madrid, Avda. Universidad, 30, 28911, Leganés, Madrid, España

Enlace al póster.


  • La medida de descargas parciales (DP) permite llevar a cabo un mantenimiento predictivo en instalaciones eléctricas.
  • Las DP emiten una radiación electromagnética que puede ser medida con antenas para la localización de la fuente sin interrumpir el servicio de la instalación.


  • Actualmente, se usan al menos cuatro antenas situadas en distintos puntos para la localización de la fuente de DP.
  • Calculando la diferencia de los tiempos de llegada \tau_{ij} de la emisión a las antenas, y minimizando la función objetivo F se puede estimar la posición \hat{P}_s de la fuente de DP.


  • Se puede localizar fuentes de DP con sólo dos antenas siguiendo el procedimiento propuesto en este póster:
  • Para calcular la dirección (azimut y elevación) de la fuente de DP se deben orientar las antenas maximizando \tau_{12} y tomar datos en varias posiciones.Imagen4.pngImagen3
  • La distancia entre antenas se mantiene contante e igual a 2 m. Como la velocidad de propagación es c=3\times10^8 m/s, el máximo \tau_{12} es TDoA=2/c =6,67 µs.
  • La posición de la fuente de DP se define como la intersección de las direcciones calculadas en las posiciones donde se realizan las medidas.





  • Sistema de adquisición de señales de dos canales basado en una FPGA con un ADC de bajo coste.
  • Antenas monopolo omnidireccionales adaptadas para medir en la banda de frecuencias de las DP.


  • La nueva metodología permite localizar fuentes de DP con un sistema de adquisición de dos canales en lugar de cuatro.
  • La reducción de canales de adquisición reduce el precio y el peso del sistema de adquisición.


  • Es posible localizar fuentes de DP con un sistema de adquisición de dos canales.
  • Ubicando este equipo y las dos antenas en un vehículo aéreo no tripulado, se podría mejorar la exactitud de las medidas y por tanto de la localización.

Characterization of Peltier cells for energy harvesting applications (III)

As demonstrated in the former post, the equivalent voltage source of the cell depends on the temperature difference of the surfaces and takes a value of V_o = 0.0245 \cdot \Delta T and the internal series resistor is R_s = 2.24~\Omega. Therefore, there would be different power outputs considering the resistor load and the temperature difference. The next plot shows the delivered power to a set of loads and four temperature differences \Delta T =[5,~10,~15,~20] degree Celsius. The maximum power given by the cell is delivered to a load that equals the internal resistor, R_s=R_L and takes a value of:

P_{max}=\frac{V_o^2}{4R_L}&s=1 W


If a difference of temperatures of 20 ºC is achieved the voltage at the load would be 245 mV and it would draw a current of 109.4 mA, the maximum power would reach 26.8 mW when connecting a load of 2.24~\Omega. Of course, all these data are hypothetical since the assumptions are in the most optimistic side considering that R_s=R_L. Even under these circumstances a voltage booster would be needed to increase the voltage to a level according to the requirements of the MCU. For instance, the ultralow power STM32L432 ARM Cortex M4 requires at least a power supply of 1.71 V. There are two options to increase the voltage, using voltage multipliers or using DC-DC converters.

Voltage multipliers

These circuits use a combination of diodes and capacitors that allows to duplicate the voltage at the input in every stage. A common setup is the Cockcroft-Walton configuration as the shown in this paper to multiply the voltage obtained from events that create pulses that can reach peaks of 1 V or more. In the case of one or two Peltier cells connected in series, this circuit is out of the question since the Schottky diodes with the lowest forward voltage drop are close to 250 mV so they would consume the voltage provided by the cell or cells.


Voltage boosters

Voltage boosters or DC-DC step-up converters would be the most feasible solution. The working principle is easy. The inductance L is charged closing switch S storing a magnetic field. This field will maintain the current flowing towards load R when S is opened. Since the inductance is giving energy to the load the voltage at L is effectively reversed and added to v_i(t) increasing the voltage at the output, v_o(t). The switching should be done fast to avoid a total magnetic discharge of the coil when S is open and a total depletion of capacitor C when S is closed. The diode D prevents the capacitor from discharging through S.Booster.png

This idea has been implemented in integrated circuits (IC) that scavenge small quantities of energy from the source, in our case the Pletier cell, to drive the switch and are able to increase the voltage at the output upto 3.3 V or 5 V depending on the MCU connected. Some examples of these IC and their behavior under real conditions are shown in the next post.



Characterization of Peltier cells for energy harvesting applications (II)

With the setup described in the last post, the temperature of one of the surfaces of the cells can be controlled with an electrical current, the other surface is cooled passively with a heatsink. The resistors and the two temperature sensors are connected to the same voltage source taking advantage of the wide range of voltages supported by the LM35. The outputs of the LM35 are connected directly to two multimeters to measure the differences of temperature achieved between the two surfaces. Now, the process is easy: the resistors are heated with different currents and the output voltages of the cell and the temperatures are registered. This will solve the voltage source of the equivalent electric circuit of the cell in open circuit or the Thévenin voltage. The results are represented in the next Figure showing a perfect linearity between temperature and voltage.


I_s = [0.49~0.59~0.69~0.79] A
V_s = [12.3~14.6~17.3~20] V

V_o = [62.4~140.5~252.5~414~580~775] mV

T_h = [24.8~30.2~38.4~55.0~66.5~82.7] degree Celsius
T_c = [22.5~24.5~28.0~38.0~42.9~51.0] degree Celsius

Where, I_s and V_s are the current and voltage applied by the power supply to the resistors, respectively; V_o is the output of the cell and T_h and T_c are the temperatures of the hot and cold sides of the cell, respectively. The plot shows that V_o = 0.0245 \cdot \Delta T with the slope in V / ºC.

The next test will determine the equivalent series resistance, R_s, of the cell loading it with a known resistor, R_L = 13.5~\Omega. In this case, the current given by the cell provokes a voltage drop in the internal resistor R_s so the voltage applied to R_L, V_L, is smaller than the open circuit voltage, V_o > V_L.  A new set of measurements is conducted injecting the same current to the heating resistors to determine this voltage and, then, the internal resistance of the cell:

V_L = [51.9~121.1~ 211.7~ 344.1~ 486~ 639] mV

T_h = [27.6~ 33.2~ 40.7~ 53.0~ 65.0~ 79.3] degree Celsius
T_c = [25.5~ 27.7~ 30.7~ 36.7~ 41.9~ 48.6] degree Celsius


The blue plot represents the voltage in open circuit, V_o, and the red plot the voltage at the load, V_L. Dividing this voltage by the load resistor yields the current given by the cell, I_L. Therefore, the internal resistor is calculated applying Ohm’s Law knowing that the voltage drop is V_o - V_L and the current is I_L. This is done for several points along the experimental results in the plot giving a constant value for the resistor, R_s = 2.24~\Omega. The same process is repeated for another cell of the same type and the result for R_s differ from the first cell giving R_s = 4.75~\Omega. Even when the former internal resistor is double the latter one, the results have been re-checked and confirmed and are in agreement to other results found in the literature.

Once the cell has been characterized, it is possible to determine the current that it will give to a known load. The next step is to devise a method to store the energy delivered by the cell or to explore the possibility of boosting the voltage to drive an MCU (microcontroller unit) directly. This will be explained in the next post.

Characterization of Peltier cells for energy harvesting applications (I)

Módulo Peltier, 32.8W, 6A, 8.8VPeltier cells are usually applied to cool surfaces when connected to an electric power supply but they can also convert differences of temperature between their sides into a voltage, known as the Seebeck effect. Therefore, it is possible to have a voltage at the ends of the wires of a Peltier cell by applying heat to one of the sides and attaching a heatsink to the other side. In terms of energy harvesting, the heat should come from a residual source such as an electrical or mechanical machine or, simply, the sun using Fresnel lenses. The cooling of the other side should be passive to minimize the energy consumption, hence the use of a heatsink. It is important to know how much energy can be obtained with a single cell as a function of the difference of temperatures, for this reason, the characterization of the cell must be the first step in the design of applications scavenging energy.

20170324_162157.jpgThe mounting scheme is shown in the side Figure where the heatsink is clearly seen on top of the cell. This is hidden by some pieces of thermal insulating foam but the two wires are visible. Finally, an aluminium plate has been adhered to the cell with a thermal conductive bonding paste.

The aim of this work is to obtain the Thévenin equivalent of an Adaptive ETH-071-14-15 Peltier cell but the process is valid for any other type of cell. The datasheet details the working curves when the cell is used as load but nothing is said about its characteristics when used as a source. The characterization requires the application of a known difference of temperatures, ΔT, and this is achieved injecting current to a pair of ceramic resistors  of 47 Ω in parallel attached to the aluminium plate with silicone.


20170324_161932_001Two LM35 temperature sensors read the temperature of the aluminium plate and the heatsink. Notice that, once the plate and the heatsink are attached to the cell, its sides are no longer reachable so the sensors have to be connected to the closest surfaces to the cell. There will be an uncertainty in the measurements but we can asume that it is negligible or, at least, that it affects the two sensors in the same manner so the difference of temperatures is the same as in the surfaces of the cell. The mounted cell is enclosed in methacrylate box coated with thermal insulating panels ensuring that, once a constant difference of temperatures, ΔT, is achieved, it is maintained along the experiment so the output voltage can be directly related to the selected ΔT.

Next: Measurements.