• Aucun résultat trouvé

Perturbations of Electric and Magnetic Fields Due to the Presence of Materials in TEM Cells

N/A
N/A
Protected

Academic year: 2021

Partager "Perturbations of Electric and Magnetic Fields Due to the Presence of Materials in TEM Cells"

Copied!
12
0
0

Texte intégral

(1)

HAL Id: hal-02335034

https://hal.archives-ouvertes.fr/hal-02335034

Submitted on 17 Jul 2020

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

Perturbations of Electric and Magnetic Fields Due to

the Presence of Materials in TEM Cells

Mohsen Koohestani, Mohamed Ramdani, Philippe Besnier, Jean-Luc Levant,

Richard Perdriau

To cite this version:

Mohsen Koohestani, Mohamed Ramdani, Philippe Besnier, Jean-Luc Levant, Richard Perdriau.

Per-turbations of Electric and Magnetic Fields Due to the Presence of Materials in TEM Cells. IEEE

Transactions on Electromagnetic Compatibility, Institute of Electrical and Electronics Engineers, 2020,

62 (4), pp.997-1006. �10.1109/TEMC.2019.2928215�. �hal-02335034�

(2)

ACCEPTED MANUSCRIPT

Perturbations of Electric and Magnetic Fields Due

to the Presence of Materials in TEM Cells

Mohsen Koohestani, Senior Member, IEEE, Mohamed Ramdani, Senior Member, IEEE, Philippe Besnier,

Senior Member, IEEE,

Jean-Luc Levant, and Richard Perdriau, Senior Member, IEEE

Abstract—This paper investigates in detail the impact of electric (E) and magnetic (H) field distributions inside TEM cells, below and above the corresponding dominant modes of the cells, in conditions differing from free space, i.e. in presence of representative ferromagnetic and dielectric materials. Simulations and measurements were performed from hundreds of MHz to GHz using IEC 61967-2 (closed) and open TEM cells. Whatever the frequency and EUT location, unlike dielectric material which only locally changes the norms (∣∣.∣∣) of E (and H depending on permittivity) at its location, the presence of ferromagnetic material inside the cell changes both ∣∣E∣∣ and ∣∣H∣∣ distributions: locally below the dominant mode frequency and globally at the entire bottom of the cell above that frequency. This demonstrates that a local distortion of ∣∣H∣∣-field, due to ferromagnetic material, has a stronger influence than a local distortion of ∣∣E∣∣-field, disregarding frequency, location and magnetic losses. Moreover, the demand to use a TEM cell below the dominant mode frequency, as mentioned in the IEC 61967-2 and 661967-21361967-2-61967-2 standards, may not be relevant, provided that both EM field inhomogeneities are considered, and the presence of an IC package in the cell is accounted for in the equivalent ∣∣E∣∣-field level around the pins in immunity testing.

Keywords—closed and open TEM cells, ∣∣E∣∣ and ∣∣H∣∣ field distribution, ferromagnetic and dielectric materials

I. INTRODUCTION

E

LECTROMAGNETIC IMMUNITY (EMI) and

electro-magnetic compatibility (EMC) characterization of inte-grated circuits (ICs) are of great importance as the high-speed signals not only cause signal integrity issues, but also lead to radiation and interference problems [1]–[3]. Particularly, with the fast pace of ICs technologies to reduce their sizes and/or

Manuscript received April 10, 2019.

M. Koohestani, M. Ramdani, and R. Perdriau are with ESEO, Department Electrical and Control Engineering, RF-EMC research group, Angers 49107, France, and also with the Institute of Electronics and Telecommunications of Rennes (IETR), Rennes 35042, France (e-mail: {mohsen.koohestani, mo-hamed.ramdani, richard.perdriau}@eseo.fr).

P. Besnier is with the Institute of Electronics and Telecommunications of Rennes, Rennes 35042, France (e-mail: philippe.besnier@insa-rennes.fr)

J.-L. Levant is with Microchip (previously ATMEL) Nantes, 44300 Nantes, France (e-mail: jeanluc.levant@microchip.com)

at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TEMC.20XX.XXXXXXX.

power consumption, they are more susceptible to electromag-netic (EM) disturbances and, consequently, providing good EMC characteristics is much more challenging [1], [4]. This makes the difference among IC manufacturing competitors.

Every IC package (including the pins) has conducted and radiated immunity levels, and its malfunction is subject to exposure to electric (E) or magnetic (H) fields. Depending on the ICs internal architecture and pin couplings, EM disturbance can affect the IC performance leading to exceed the immunity levels to E and H (e.g. all microcontrollers (µCs) are sensitive to both E and H with different frequency behaviors) [5]. Therefore, determining those EMC parameters helps engineers to evaluate the IC having an idea of its immunity. In parallel, ICs must not disturb other devices located on the same PCB or in the same system: consequently, their E and H emission levels must also be assessed. The international electrotechnical commission (IEC) has set standards for ICs emission (IEC 61967-2 [6]) and immunity (IEC 62132-2 [7]) testing.

Since their introduction, transverse electromagnetic (TEM) cells have been widely used for EMC/EMI emission/immunity testing of active devices to indicate the equipment under test (EUT) radiation with respect to its orientation and to frequency [8], [9]. A TEM cell approximates a uniform and linearly polarized plane wave. Generating uniform fields of known strength is important for accurate and repeatable susceptibility measurements. A TEM cell as the far-field free space condition emulator is assumed to have a uniform field distribution with the wave impedance of approximately 120π Ω (i.e. ∣∣E∣∣ to ∣∣H∣∣ norms ratio is 377 Ω). However, there exist several questions to be addressed.

(Q1) To what extent a TEM propagation and a wave impedance of 120π can be assumed? (Q2) Does the EUT induce higher order modes and, in this case, how does their location influence the test? (Q3) What about the impact in IEC 61967-2 TEM cells (referred to as “closed” throughout the paper) where the EUT cannot be moved?

Electronic circuits are usually printed on dielectric materials with certain permittivity, and sometimes they are covered with ferromagnetic materials (like ferrite sheets) to improve their radiated immunity and/or emission. Moreover, in most cases, a piece of paper (having a dielectric constant of ∼ 3 [10]) is used as an insulator when placing the EUT at the bottom of open TEM cells. (Q4) What are the impacts of those materials on the EM field (both distribution and level) inside the TEM cell where the EUTs are located?

Furthermore, during our recent experiments using an open TEM cell, we noticed that the presence of our hand to change

(3)

ACCEPTED MANUSCRIPT

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. XX, NO. X, JANUARY 20XX 2

the orientation of the PCB hardly disturbs the received power spectrum measurements (Fig. 1a). (Q5) Does this mean that a dielectric material has no influence (but local) on the E and H distributions inside the cell?

In literature, there are several works reporting information on the EM field distribution inside TEM cells [11]–[21]. However, in addition to the above-mentioned questions (Q1 to Q5), (Q6) the question of E- and particularly H-field distributions as a function of frequency has not been addressed so far in TEM cells analysis, especially in presence or absence of different materials.

Insight into both E and H distribution mechanisms in TEM cell contributes to the understanding of the rationale behind the causes of susceptibility problems when they occur. In addition to that, (Q7) there is no clear reason why the center position in the cell, chosen for emission/immunity testing (according to IEC standards), would implicitly represent the optimal scenario for emission and immunity at any frequency. This information can allow the IC designer to better predict the emission/immunity of their ICs. Therefore, we would like to add our contribution to the in-depth study of EM fields inside TEM cells.

The main goal of this study is to examine these open challenges by reporting and clarifying the role of E and H dis-tributions in TEM cells. Indeed, the correlation between those distributions and EMC/EMI properties of EUTs is of great importance as it contributes to the advancement of knowledge in the field through better understanding of EMC/EMI in emission/immunity testing scenarios. Simulations and mea-surements were conducted to validate the study.

The paper is organized as follows. Section II describes the materials used, the methodology employed, and the simulation and experimental setup configurations. Section III details the analysis of the numerical results, while Section IV addresses the experimental results of the proposed technique. Concluding contributions of this study are provided in Section V.

II. MATERIALS ANDMETHODS

This section presents the research material and methodology to study and highlight the EM field behavior in TEM cells. A. Simulation Setup and Method

A standard Tekbox open TEM cell (Model: TBTC1, DC up to 3 GHz [22]) as the far-field free space condition emulator was numerically modeled using Ansys HFSS commercial full-wave electromagnetic simulation package [23] (Fig. 1a to Fig. 1d). All the yellow-colored metallic parts in the numerical

models were considered to be copper (σ = 5.8×107 S/m). A

rigorous iterative algorithm based on mixed basis functions was used to refine the mesh, noting that mixed order efficiency is comparable or better than all single order basis functions [23].

To study the influence of the human hand on the field distributions inside a TEM cell, a high-resolution (1 mm) dispersive anatomical geometry of a human hand was used in simulations. The model was taken from the predefined human body exteriors in the HFSS component library [23]. It is

1

TEM cell concept: highlighting relative field strength between electric and magnetic

M. Koohestani, R. Perdriau, and M. Ramdani Since its introduction, transverse electromagnetic (TEM) cell has been widely used for far-field EMC/EMI immunity testing of RF devices. However, the relative strength between the electric and magnetic field in the cell is unknown. This paper clarifies, for the first time, whether …

Introduction: Apart from the layout and housing design, the characteristics of the integrated circuits (ICs) used play a key role for the EMC characteristics of devices. Reducing the size of the structure, operating voltages and operating points makes the ICs much more sensitive.

For manufacturers of ICs, good EMC characteristics for their products mean advantages over their competitors. The objective is thus to determine those parameters which are decisive for EMC immunity and emissions and allow engineers to draw conclusions for chip design. Each IC has characteristic immunity levels with regard to conducted and radiated interference. These are its EMC parameters. The IC pins have conducted immunity levels, which can be measured with corresponding probes from the probe sets.

The IC as a whole has radiated immunity levels. Disturbance fields can affect the IC from the outside and exceed these levels of immunity to magnetic fields and electric fields. These immunity levels are independent of each other. Probes that generate suitable and defined fields are needed to determine the field immunity levels. Generation fields of known and uniform strength is important for accurate and repeatable susceptibility measurements. Power frequency uniform magnetic field is created by a Helmholtz coil arrangement. It can produce an uniform magnetic field of known strength over a volume necessary to perform experiments

ATA: IC malfunction is subject to the exposure to E or H fields.

The rate of change of magnetic flux induces a voltage in any length of PCB traces so, all electrical circuits and chips are sensitive to magnetism in the broadest sense.

Models are needed that not only predict the field coupling from the ESD gun to the IC, but that are fast enough that these variations can be explored. Accurate models can allow the IC designer to better predict the immunity of their ICs and to better understand the causes of susceptibility problems when they occur.

Proposed technique and design:

a b c

Fig. 1 Simulation setup inside a standard TEM cell

a PCB without coil b PCB with a resonating coil c PCB with a non-resonating coil

a

b

c

Fig. 2 E, H, and E/H ratio inside TEM cell for an input power of 1 W

a PCB without coil b PCB with a resonating coil c PCB with a non-resonating coil

Side-view x y z Side-view x y z Side-view x y z (a) Mohsen KOOHESTANI 1.4 cm 1.4 cm 1 0 .8 c m (b) Mohsen KOOHESTANI 1.4 cm 1.4 cm 1 0 .8 c m 1 0 c m 10 cm 3 cm 3 c m Feed point Front-view Back-view (c) 9 .9 c m 1 0 .8 c m (d) 3

Scenario S11 at resonance frequency (dB) Calculated input power (W)

0.65 GHz 0.86 GHz 0.65 GHz 0.86 GHz

TEM cell - 38.68 - 35.31 1.0001 1.0003

TEM cell loaded with a cylinder Ferrite (εr = 12, µ = 1000) - 6.05 - 11.61 1.33 1.074

TEM cell loaded with a cylinder Ferrite (εr = 1, µ = 1000) - 6.05 -11.62 1.33 1.074

TEM cell in presence of an anatomical human hand - 21.85 - 22.33 1.006 1.0058

TEM cell loaded with a rectangular Ferrite - 30.99 - 32.94 1.0008 1.0005

TEM cell loaded with a rectangular Plexiglass - 39.76 - 35.22 1.0001 1.0003

(e)

Fig. 1. Simulation and measurement scenarios: (a) an open TEM cell in

presence of a dispersive anatomical human hand; (b) open TEM cell with a 2.45 GHz patch antenna in its center; (c) patch antenna; (d) closed TEM cell; (e) open TEM cell loaded by a small piece of ferrite (to be replaced then by a Plexiglas of similar size) while measuring radiated emission of a SAMR21 µC (replacing the Kiwi Millenium transmitter) inside a Faraday cage.

homogeneous with an averaged permittivity and conductivity of hand tissues with the electrical properties taken from

Gabriel’s data base [24], i.e. εr = 39.2 and σ = 1.8 S/m

at 2.45 GHz. In order to assess the impact of ferromag-netic and dielectric materials on the EM field relations in-side the TEM cell, the latter was loaded by small pieces

of ferrite (εr = 12, µr = 1000, σ = 0.01 S/m) and Plexiglas

(εr = 3.4, µr = 1) of 76.2 mm × 28.6 mm × 6.35 mm

di-mensions (this is the size of a standard rectangular-shaped piece of ferrite with a recommended frequency range below 300 MHz [26]. A part of this study (section III-A) was conducted using ferrites of different permittivities of 12 and

(4)

ACCEPTED MANUSCRIPT

1 as well as virtual dielectric material with εr = 1000 while

keeping their permeability unchanged.

As far as emission simulation is concerned, a narrow-band patch antenna with a 2.45 GHz operating frequency was placed 1.4 cm (for the SMA connector not to touch the cell) above the bottom of the open TEM cell (Fig. 1b). The antenna consists of a 10 cm × 10 cm PCB, defined in the standard [6], with a 3 cm × 3 cm square metal patch coaxially-fed 8 mm off the center of the patch (Fig. 1c) to resonate at the frequency of interest. It was simulated on a low cost

FR4 epoxy substrate (εr = 4.4, thickness = 0.8 mm, and

tanδ = 0.02). The other side of the PCB was coated with metal serving as ground plane. A lumped port was used as the RF power source in HFSS. The input power was changed from 1 W to 4.2 W in order to compensate for the mismatch at the actual simulation frequency (2.475 GHz) corresponding to the emission frequency of the real device (SAMR21 µC) used for measurement in the next subsection. Moreover, ferrite and Plexiglas were located at the bottom corner of the cell close to the input source to be able to easily move the EUT along the cell axis y (in the measurement process) and to place it at the fields maximum peak and minimum null positions.

In addition, to investigate how the presence of the ferromag-netic and dielectric materials impact the field distributions in real IEC 62132-2 immunity testing inside closed TEM cells, a standard cell (FCC-TEM-JM2, DC up to 1.6 GHz [25]) was also modeled (Fig. 1d). A PCB similar to the one used for emission testing (Fig. 1c) was also considered, but with a 1 cm × 1 cm patch covered by FR4 material representing an IC with its package, and no feed point.

Finally, since the claimed upper frequency of commercial TEM cells often exceeds the 150 kHz to 1 GHz range of IEC 61967-2/62132-2 standards disregarding the propagating modes, simulations and measurements were also considered above 1 GHz. In fact, manufacturers usually define the upper frequency of a cell only based on the VSWR value; however, that frequency is generally not known to be well above the dominant mode frequency of the cell. This leads to a possible use outside the plane wave frequency range. The current study makes it possible to assess the relevance of the instructions mentioned in the standards where the TEM cell should be always used below the dominant mode frequency.

B. Measurement Setup and Method

The standard Tekbox open TEM cell [22] was employed for the purpose of this study (Fig. 1e). Two different measure-ments were conducted in order to investigate simulations in realistic conditions:

● within the useful frequency range of the used ferrite

● in the 2.4 GHz band in order to assess the cell behavior

at a higher frequency

All measurements were conducted inside a Faraday cage. Note that the immunity of an EUT can only be assessed by an indirect measurement based on a relevant criterion, since it is not possible to use a field probe within the cell without further disturbing the existing field. As is generally known, it is extremely difficult to identify an EUT which would be

subject to a malfunction at a given frequency, as well as the associated criterion. Therefore, emission measurements were chosen to adequately support our findings from simulations.

The first measurement campaign used a “Kiwi Millenium ver. 4.5” RF transmitter used by Centre National d’Etudes Spatiales (CNES) in France for communication with sounding balloons [27]. The center frequency was 138.5 MHz, which is within the frequency range of the ferrite. The RF transmitter was powered by a 9 V DC supply and introduced at the bottom of the open TEM cell with a paper sheet as an insulator. The emission spectrum was then measured thanks to a Keysight N9020A vector spectrum analyzer (VSA) in plain analyzer mode and peak hold detection, connected to the septum of the TEM cell with a one-meter 18 GHz cable. The other port of the TEM cell was terminated with a 50 Ω load. Three different configurations were tested:

● no other material than the EUT was introduced in the cell

● a thin piece of Plexiglas was introduced in the cell at the

access near the port connected to the VSA

● a piece of ferrite of the same size as Plexiglas was

introduced at the same location

The second campaign was based on the use of a SAMR21 (Microchip-Atmel) µC with a built-in 2.4 GHz RF transceiver [28]. The µC was located on a commercial Xplained Pro evaluation board with a built-in ceramic antenna (Fig. 1e). Moreover, it was programmed with a piece of software sending out frames according to the IEEE 802.15.4 protocol, with a center frequency equal to 2.475 GHz (i.e. channel 25), and powered by a 5 V DC supply. The locations of peaks and nulls in the TEM cell response had been previously characterized thanks to simulation and verified with measurement. A small piece of paper was used as an insulator (Fig. 1e). Then, the EUT was placed at one peak followed by one null; the former, being the worst-case scenario, would assess the true effect of the materials. The same three tests (without any material, with Plexiglas, with ferrite) were repeated with the µC. Finally, to assess the impact of the paper (which is indeed an insulator

with εr ∼ 3 [10]), measurements were conducted with and

without (w/o) paper. In that case, for a fair comparison and for the ease of setup, the EUT was placed at the bottom left corner of the cell.

III. NUMERICALRESULTS ANDDISCUSSION

This section deals with the guiding idea behind the work followed by the results of the empirical study for its further investigation and validation. The initial study comes from immunity testing. Hence, norms of ∣∣E∣∣- and ∣∣H∣∣-fields were simulated for an input power of 1 W with a 50 Ω load at the TEM cell ports. Conversely, the validation of the idea was carried out thanks to emission testing, where both ports of the cell were terminated with 50 Ω while the radiator (a patch antenna) was fed by a 1 W source.

In both cases, for a fair comparison of EM field distributions and to compensate for the mismatch, the results were then normalized to have the same matching level for the considered scenarios at the corresponding resonant frequencies. The EM field distributions were monitored at locations where EUTs are

(5)

ACCEPTED MANUSCRIPT

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. XX, NO. X, JANUARY 20XX 4

usually placed in TEM cells (i.e. at the top and bottom of the closed and open cells, respectively). According to simulations, only a TEM mode propagates below 1 GHz and 0.75 GHz in the closed and open TEM cells, respectively, while higher order modes coexist beyond those frequencies; a similar result was reported in [14].

In the immunity simulations, two adjacent resonance fre-quencies of the cells were considered, one below and another beyond their corresponding dominant mode frequencies (i.e. 0.9 and 1.4 GHz for the closed cell, and 0.53 and 0.88 GHz for the open one), in order to be able to find out the impact of frequency and presence of higher order modes. On the contrary, the emission simulations were carried out at the operating frequency of the chosen radiator (i.e. 2.475 GHz). A. Basic Idea

The guiding idea behind the paper stems from the fact that during a recent experiment using an open TEM cell, while

changing the PCB orientation, we noticed that the presence of our hand did not disturb the measurements to characterize the EUTs emission frequencies. This made us think about how EM fields could influence the TEM cells behavior.

A similar exposure scenario was considered with the human hand in proximity to an open TEM cell in the absence of the EUT, to see how the EM field behaves (Fig. 1a). To be able to make a comparison of the results in presence and absence of the hand, we firstly simulated both ∣∣E∣∣- and ∣∣H∣∣-field distributions inside a standard open TEM cell (TBTC1 [22], Fig. 2a). The latter was then loaded by the presence of a human hand; results are shown in Fig. 2b. Note that:

● since field distributions do not change according to any xy

plane, E- and H-fields which are always in z and x directions, respectively, are plotted at the cell’s bottom level only; any change of distribution in this plane helps distinguishing the results from one scenario to another

● the colors representing the field distribution in logarithmic

Mohsen KOOHESTANI E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (a) Mohsen KOOHESTANI E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (b) Mohsen KOOHESTANI E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (c) Mohsen KOOHESTANI E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (d) Mohsen KOOHESTANI E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (e)

Fig. 2. Simulated ∣∣E∣∣- and ∣∣H∣∣-field distributions inside an open TEM cell for 1 W input power: (a) unloaded, (b) in presence of a human hand; (c)

loaded by a small piece of ferrite (εr= 12, µr= 1000); (d) loaded by a small piece of ferrite (εr= 1, µr= 1000); (e) loaded by a small piece of Plexiglas

(6)

ACCEPTED MANUSCRIPT

scale go from dark blue (low current density) to green to yellow to red (high current density)

As observed, in line with our expectations, the hand pres-ence locally disturbs the ∣∣E∣∣ distribution whereas it has a small effect on ∣∣H∣∣ at the hand location when compared to those obtained in the unloaded cell. This may imply that inside open TEM cells ∣∣H∣∣-field coupling plays a stronger role compared to ∣∣E∣∣, as the hand presence (as a dielectric) has only a local effect on ∣∣E∣∣.

To further investigate this, we placed a small piece of a

ferrite material (εr = 12, µr = 1000) inside the open TEM cell

at the location where the hand was previously present to see how it impacts the EM fields; results are shown in Fig. 2c. As it can be seen, presenting the ferrite not only disturbs the ∣∣H∣∣-but also the ∣∣E∣∣-field distri∣∣H∣∣-butions in the cell. Unlike at fre-quencies lower than the dominant mode frequency where the presence of the ferrite only locally perturbs both ∣∣E∣∣ and ∣∣H∣∣, at higher frequencies it changes their distributions at the whole bottom of the cell. In order to check whether the disturbance in

the ∣∣E∣∣-field distribution is related to the permittivity of ferrite, although not physically possible, we considered in simulation a

case where the ferrite permittivity is set to εr= 1 while keeping

its permeability unchanged (µr = 1000); results are shown in

Fig. 2d. As observed, the ∣∣E∣∣- and ∣∣H∣∣-field distributions are almost insensitive to the permittivity modification in the con-sidered frequencies. It is worth mentioning that the influence of ferrite dimensions (increasing thickness from 6 to 20 mm and area from 76.2 mm × 28.6 mm to 105 mm × 58 mm) was checked through simulations and found to follow the same observations on the EM field distributions inside the cell.

For a fair comparison, a “virtual” dielectric material of

εr = 1000 permittivity and unit permeability was also

sim-ulated (Fig. 2e), demonstrating that it has only a very local influence on both ∣∣E∣∣ and ∣∣H∣∣. After all, in order to signifi-cantly perturb the ∣∣E∣∣-field, the thickness of the Plexiglas was increased from 6 to 20 mm observing hardly any difference (results are not shown here due to minor changes).

To conclude, these results show that, independently of

Mohsen KOOHESTANI F er ri te P le x ig la s F S E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz F er ri te P le x ig la s F S @ 2 .4 7 6 G H z (a) Mohsen KOOHESTANI F er ri te P le x ig la s F S E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz F er ri te P le x ig la s F S @ 2 .4 7 6 G H z (b) Mohsen KOOHESTANI F er ri te P le x ig la s F S E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz F er ri te P le x ig la s F S @ 2 .4 7 6 G H z P C B (c) Mohsen KOOHESTANI F er ri te P le x ig la s F S E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz P C B E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz (d) H @ 0.92 GHz H @ 1.4 GHz Mohsen KOOHESTANI F er ri te P le x ig la s F S E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz P C B E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz E @ 0.92 GHz E @ 1.4 GHz H @ 0.92 GHz H @ 1.4 GHz (e)

Fig. 3. Simulated ∣∣E∣∣- and ∣∣H∣∣-field distributions inside a closed TEM cell for an input power of 1 W at its two first resonance frequencies: (a) unloaded; (b) loaded with ferrite; (c) loaded with Plexiglas; (d) loaded with the IEC PCB and a 1 cm × 1 cm IC in the center; (e) similar to (d) with the IC shifted towards its side along the cell main axis.

(7)

ACCEPTED MANUSCRIPT

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. XX, NO. X, JANUARY 20XX 6

frequency and even in presence of higher order modes, loading an open TEM cell with a dielectric (such as the human hand) can only locally modify the ∣∣E∣∣ distribution (and ∣∣H∣∣

for a very high εr value) whereas a ferromagnetic material

(like ferrite) can alter both ∣∣E∣∣ and ∣∣H∣∣ distributions, locally (below the dominant mode frequency) or in the cell bottom where EUTs are usually placed (above that frequency). This is convincing enough to draw a conclusion about the relative ∣∣E∣∣- and ∣∣H∣∣-field influences inside open TEM cells, i.e. “a local and significant ∣∣H∣∣-field distortion has a stronger impact than a local and significant ∣∣E∣∣-field distortion. In other words, the local ∣∣H∣∣-field distortion couples to the TEM cell in such a way that the wave propagation is dramatically altered”.

In order to gain a deeper understanding, since there is supposed to be a linearly-polarized plane wave inside the TEM cell, and the wave experiences different media in presence of the considered materials, the power reflection and transmission mechanisms are different. We considered the power reflection (Γ) and transmission (T) coefficients based on the intrinsic impedance of the air-Plexiglas and air-ferrite interfaces; for air

(εr = 1, σ = 0) - Plexiglas (εr = 3.4, σ = 0) interface Γ and T

are 8.8% and 91.2%, whereas for the air ferrite (εr= 12, σ =

0.01 S/m, µr= 1000) Γ significantly increases to 64.4% while

T decreases to 35.6%. This justifies the stronger distortion in the cell due to the ferromagnetic rather than the dielectric materials. As far as Plexiglas is concerned, the traveling wave passes almost entirely in the block across the whole frequency range of the cell leading to minimal disturbances (only locally around the block). Conversely, with a ferrite, at frequencies above the dominant mode, not only the fundamental wave is perturbed due to the high amount of reflected power but also all higher order modes, which leads to a global distortion in the cell. More generally speaking, a ferrite offers larger (normal) electric field and (tangential) magnetic field discontinuities at the boundaries, inducing secondary sources.

For the sake of completeness and to support that main finding, further simulations in presence of ferromagnetic and dielectric materials were performed:

● to see whether it is valid inside closed TEM cells

● to assess the influence of an IC mounted on a PCB on

field distribution

● to compare the results with practical measurements

For a fair comparison, small rectangular-shaped pieces of ferrite and Plexiglas of the same dimensions (76.2 mm × 28.6 mm × 6.35 mm), shown in Fig. 1e, were used in simulations and measurements.

Fig. 3 exhibits the ∣∣E∣∣- and ∣∣H∣∣-field distributions inside a closed TEM cell (FCC-TEM-JM2 [25]). Similar observations to those with an open cell can be made:

● whatever the frequency and the propagating modes, unlike

ferrite that modifies both ∣∣E∣∣ and ∣∣H∣∣ distributions when compared to an unloaded cell, Plexiglas only changes ∣∣E∣∣

● therefore, an ∣∣H∣∣-field distortion is more significant than

a ∣∣E∣∣-field distortion inside closed cells

What is interesting is that the locations of both ∣∣E∣∣ and ∣∣H∣∣ maxima vary with frequency, resulting in field inhomogeneity even just below the dominant mode. Therefore, in TEM cell

measurements where a non-dielectric EUT (such as a piece of wire as a transmission line over the PCB intended for field evaluation purposes) is located at the center of the PCB, practical results would not always correspond to the most important disturbance within the cell (that can be seen at 0.92 and 1.4 GHz for ∣∣E∣∣ and ∣∣H∣∣, respectively). This implies that, depending on frequency, caution has to be taken for testing using closed TEM cells. Moreover, it can be stated that the presence of an IC on the IEC 62132-2 standard PCB [7] changes dramatically the field distribution. In particular, it reduces the field magnitudes around the IC pins and even more in the IC package (Fig. 3d). What can also be noticed is that field magnitudes at IC location are the same regardless of its position on the PCB, despite the fields being non-uniform within the cell (Fig. 3e). The latter corroborates the fact that measurements could overestimate IC immunity at pin locations by about 30% (145 V/m instead of 210 V/m, Fig. 3d) compared to the maximum field available in the TEM cell. It is also worth mentioning that, according to simulations, the wave impedance inside both unloaded closed and open TEM cells where the EUT is usually placed is partially beyond 377 Ω, up to 511 Ω; a fact corroborated in [20].

Note that to have an idea of how relocating the materials inside the open cell influences the EM field distributions at its resonance frequencies as well as the same frequencies used in emission measurements (138.5 MHz and 2.475 GHz), results are depicted and discussed in the Appendix.

B. Proof of Concept

In order to assess the open TEM cell behavior in emission testing with a setup comparable to the one intended to be

Mohsen KOOHESTANI F er ri te P le x ig la s F S (a) Mohsen KOOHESTANI F er ri te P le x ig la s F S (b) Mohsen KOOHESTANI F er ri te P le x ig la s F S (c)

Fig. 4. Simulated ∣∣E∣∣- (left) and ∣∣H∣∣-field (right) distributions inside an

open TEM cell in presence of a patch antenna radiating at 2.475 GHz: (a) unloaded, (b) loaded with ferrite; (c) loaded with Plexiglas. Arrows indicate the locations of field distribution discrepancies among all cases.

(8)

ACCEPTED MANUSCRIPT

(a) (b) (c)

(d) (e) (f)

(g) (h) (i)

Fig. 5. Measured radiated emission using an open TEM cell of (a-c) a Kiwi Millenium µC at 138.5 MHz; (d-f) and (g-i) a SAMR21 µC at 2.475 GHz with a EUT placed at a peak and a null location, respectively: (left) unloaded; (middle) loaded with ferrite; (right) loaded with Plexiglas.

used in measurements later, simulations were carried out at 2.475 GHz with the patch antenna as the radiator. Fig. 4 shows both ∣∣E∣∣- and ∣∣H∣∣-field distributions. As observed, in those simulations, totally similar conclusions to those drawn in immunity simulations can be made, except that no reduction around the IC pins can be noticed.

IV. EXPERIMENTALRESULTS ANDDISCUSSION

The emissions of both ”Kiwi Millenium” transmitter and SAMR21 µC were measured using an open TEM cell where the EUTs were placed firstly at a peak and then at a null location; results are shown in Fig. 5a to 5i and at the same time summarized in Table I in terms of frequency and maximum received power levels for improved readability. In order to characterize those two locations, the EUT was slightly moved along the cell central y-axis while monitoring the level of the signal. Note that, at 2.475 GHz, due to the fast wave traveling at the chosen frequency, 1 mm steps and a 1 cm range

were enough to ensure finding the maximum and minimum power levels.

As can be seen in Fig. 5c, at 138.5 MHz where only the TEM mode is present, the presence of dielectric material has hardly any influence on the power levels collected by the cell (−17.9 dBm) and, consequently, the field distribution in the TEM cell. Conversely, a 1.2 dB decrease in power level was observed with the ferromagnetic material (Fig. 5b).

Moreover, it can be seen in Fig. 5 that those results are also valid at 2.475 GHz (0.8 and 1.5 dB at the null and peak, respectively, although the ferrite is supposed not be operational at that frequency). Note that the decrease was even more pronounced with the ferrite getting closer to the EUT (up to 6 dB for 1 cm close proximity), whereas a minor change was noticed with the Plexiglas at the same location. This is due to the fact that Plexiglas has a local influence on EM field distributions, whereas the ferrite changes the distributions in the entire bottom of the cell. These results are in line with the

(9)

ACCEPTED MANUSCRIPT

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. XX, NO. X, JANUARY 20XX 8

TABLE I

MTable I Measured power levels (in dBm) for the considered scenarios EASURED POWER LEVELS(IN DBM)FOR THE CONSIDERED SCENARIOS

Scenarios Unloaded cell Cell loaded with Ferrite Cell loaded with Plexiglas 138.5 MHz –17.9 –19.1 –17.9 2.475 GHz, peak –28.9 –30.4 –29.1 2.475 GHz, null –36.6 –37.5 –36.1 REVIEWER COMMENT: - REFERENCES

* In the mean time, Ref.[3] is published in the IEEE EMC Trans.: vol.60, no.5, pp.1249-1253, Oct.2018.

REPLY OF AUTHORS:

Ref. [3] has now been revised as follows:

F. Pythoud, “Proficiency testing in EMC radiated immunity,” IEEE Trans. Electromag. Compatibility, vol. 60, no. 5, pp. 1249-1253, 2018.

Thank you again for your detailed relevant and useful comments. We tried to take them into account very carefully and have updated the paper accordingly.

---

REPLIES TO REVIEWER 3 COMMENTS: REVIEWER COMMENT:

This paper presents the analysis of the E-field and H-field in the TEM cells as a function of frequency and various materials added inside. The findings presented in the paper are interesting but not communicated in a clear manner. Furthermore, the novelty of the presented results is questionable. The paper will have significantly improved quality if the findings related to the change in the field magnitude (due to the added materials) are applied to the test case in order to demonstrate usefulness, i.e. increase in radiated immunity.

REPLY OF AUTHORS:

Regarding the "questionable novelty" of the work, the minimum is to provide us with at least one similar study. We believe that we are well aware of previous works reported in literature but definitely accept if wrong. As you suggested, we do also believe that the findings mentioned in the paper are useful for the community. It is worth mentioning that the impact of adding dielectric materials over an IC (not in the TEM cell itself) has already been published by us (reference [20] in the paper) as a way to improve the radiated immunity of the IC. Note that it was demonstrated only for a given IC at a frequency below the dominant mode of the cell. The scope of the current study is highly unrelated to our previous work, highlighting different findings as mentioned in the conclusion of the current paper.

Regarding the test case, thanks to the reviewer, we can also take the opportunity to mention that the immunity of an EUT can only be assessed by an indirect measurement based on a relevant criterion, since it is not possible to use a field probe within the cell without further disturbing the existing field. As generally known, it is extremely difficult to identify an EUT which would be subject to a malfunction at a given frequency, as well as

conclusions extracted from simulations at frequencies higher than the cell’s dominant mode frequency.

Furthermore, measurements with and w/o paper were con-ducted; results are shown in Fig. 6. As can be seen, the presence of the paper has hardly any influence on the power level. This is in fact due to the ∣∣E∣∣-field being locally perturbed only in the close proximity of the EUT, which does not impact the measurements in emission testing. It can be expected from the simulation results that it would also have a tiny impact on immunity testing due to the low permittivity value of the paper.

Finally, in order to understand the impact of the considered materials on the cell input impedance and to have an idea of the material losses at the chosen frequencies, the scattering parameters were measured using a Keysight 8753ES vector network analyzer; results are shown in Fig. 7. As observed, the presence of the materials does not detune the cell input reflection coefficient. Moreover, ferrite losses are negligible

at 138.5 MHz (the S21 parameter is unchanged, Fig. 7) and,

therefore, can not explain by themselves the difference in the measured power levels (Fig. 5). Even though there is a 0.8 dB (1.5 dB) difference in power levels at the null (peak)

at 2.475 GHz, the difference in S21 parameter is lower at that

frequency (0.65 dB), reminding that it is outside of the ferrite operating range. The results were perfectly reproducible. They are in line with our expectations and clearly in agreement with all simulations performed previously.

V. CONCLUSION

A study of electromagnetic field distributions inside closed and open TEM cells was presented to figure out whether the presence of ferromagnetic and dielectric materials influences the field distributions, below and above the corresponding dominant modes of each cell. Simulations and measurements were performed at different frequencies (i.e. hundreds of MHz to GHz) to assess the relative influence of the electric and magnetic fields generated or received by the cell when loaded by representative ferrite and Plexiglas materials.

First of all, just below the dominant mode and not only above, the inhomogeneity of both fields should be taken into account in order to assess the immunity level of an EUT. It was demonstrated that it is actually feasible to use the cells between the cutoff frequency and the upper frequency, provided that inhomogeneity is considered in the field levels. Note that, in the case of the open cell, it is possible to relocate the EUT at a peak in order to enhance field magnitude for a given input power. That result demonstrates that VSWR cannot be the only criterion limiting the operating range of a TEM

(a)

(b)

Fig. 6. Radiated emission measurements at 2.475 GHz with a SAMR21 µC at the bottom corner of an unloaded open TEM cell: (a) with and (b) w/o paper.

0 0.5 1 1.5 2 2.5 3 Frequency, GHz -60 -40 -20 0 |S11 |, dB Unloaded Loaded by Ferrite Loaded by Plexiglas 0 0.5 1 1.5 2 2.5 3 Frequency, GHz -8 -6 -4 -2 0 |S21 |, dB Unloaded Loaded by Ferrite Loaded by Plexiglas X: 0.1397 Y: -0.09603 X: 2.475 Y: -4.502 X: 2.475 Y: -3.854

Fig. 7. Measured reflection and transmission coefficients of the open TEM cell without any material, with ferrite, and with Plexiglas.

cell: the dominant mode frequency must be determined, and measurements, while being feasible above that frequency, must be carefully rescaled according to field levels.

Then, experiments with materials resulted in the following: independently of frequency and EUT location, unlike the dielectric which only locally changes ∣∣E∣∣ (and ∣∣H∣∣ depending on permittivity) at the material location, the presence of a ferromagnetic material inside the cell changes both ∣∣E∣∣ and ∣∣H∣∣ distributions: locally below the dominant mode frequency and globally at the entire bottom of the cell above that frequency. This clearly demonstrates that a local distortion of ∣∣H∣∣-field, due to a ferromagnetic material, has a stronger influence than a local distortion of ∣∣E∣∣-field inside both closed and open TEM cells, disregarding frequency, location and magnetic losses.

(10)

ACCEPTED MANUSCRIPT

Mohsen KOOHESTANI E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (a) Mohsen KOOHESTANI E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (b) Mohsen KOOHESTANI

For 2.475 Num. Division to HFSS 14 gozashtam for unloaded and loaded with Plexiglas to avoid scalling issues

E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (c) Mohsen KOOHESTANI

For 2.475 Num. Division to HFSS 14 gozashtam for unloaded and loaded with Plexiglas to avoid scalling issues

E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (d) Mohsen KOOHESTANI

For 2.475 Num. Division to HFSS 14 gozashtam for unloaded and loaded with Plexiglas to avoid scalling issues

E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 138.5 MHz E @ 2.475 GHz H @ 138.5 MHz H @ 2.475 GHz E @ 0.53 GHz E @ 0.88 GHz H @ 0.53 GHz H @ 0.88 GHz (e)

Fig. 8. Simulated ∣∣E∣∣- and ∣∣H∣∣-field distributions inside an open TEM cell for an input power of 1 W at different frequencies: (a) and (d) loaded with ferrite; (b) and (e) loaded with Plexiglas; (c) unloaded.

dominant mode frequency, as mentioned in the IEC 61967-2 and 62132-2 standards, may not be relevant, provided that both EM field inhomogeneities are considered, and the presence of an IC package in the cell is accounted for in the equivalent ∣∣E∣∣-field level obtained around the pins in immunity testing for a given input power whatever the frequency.

APPENDIX

Fig. 8 shows the EM field distributions inside an open TEM cell when loaded by ferrite and Plexiglas placed near the cell input port at the cell first two resonances and the emission frequencies of both EUTs (138.5 MHz and 2.475 GHz). As can be seen, similar observations and explanations to those given in Section III-A for the materials located in the TEM cell center can be made.

ACKNOWLEDGMENT

The authors would like to cordially thank Prof. A. Alves Moreira for the fruitful discussions. We also thank V. Arboux

for his assistance with prototype construction.

REFERENCES

[1] S. Ben Dhia, M. Ramdani, E. Sicard, “Electromagnetic Compatibility of Integrated Circuits: Techniques for low emission and susceptibility,” ISBN 978-0-387-26601-5, Berlin: Springer, 2006.

[2] V. S. Reddy, P. Kralicek, and J. Hansen, “A novel segmentation approach for modeling of radiated emission and immunity test setups,” IEEE Trans. Electromag. Compatibility, vol. 59, no. 6, pp. 1781-1790, 2017. [3] F. Pythoud,“Proficiency testing in EMC radiated immunity,” IEEE Trans.

Electromag. Compatibility, vol. 60, no. 5, pp. 1249-1253, 2018. [4] M. Ramdani, E. Sicard, A. Boyer, S. Ben Dhia, J. Whalen, T. Hubing, M.

Coenen, and O. Wada, “The electromagnetic compatibility of integrated circuits - past, present, and future,” IEEE Trans. Electromag. Compati-bility, vol. 51, no. 1, pp. 78-100, 2009.

[5] Sh.-Y. Yuan, Y.-L. Wu, R. Perdriau, and Sh.-S. Liao, “Detection of electromagnetic interference in microcontrollers using the instability of an embedded phase-lock loop,” IEEE Trans. Electromag. Compatibility, vol. 55, no. 2, pp. 299-306, 2013.

[6] IEC 61967-2: Integrated circuits, measurement of electromagnetic emis-sions, 150 kHz to 1 GHz, Measurement of radiated emissions - TEM cell and wideband TEM cell method, 2005.

(11)

ACCEPTED MANUSCRIPT

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. XX, NO. X, JANUARY 20XX 10

[7] IEC 62132-2: Integrated circuits, measurement of electromagnetic immu-nity, 150 kHz to 1 GHz, Measurement of radiated immunity - TEM cell and wideband TEM cell method, 2010.

[8] M. L. Crawford, J. L. Workman, “Using a TEM cell for EMC measure-ments of electronic equipment,” US Department of Commerce, National Bureau of Standards, 1979.

[9] M. Kanda, R. David Orr, “Generation of standard electromagnetic fields in a TEM cell,” US Department of Commerce, National Bureau of Standards, 1988.

[10] D. E. Anagnostou, Organic paper-based antennas. “Innovation in Wear-able and Flexible Antennas”, chapter 2, WIT Press, 2014.

[11] M. L. Crawford, “Generation of standard EM fields using TEM trans-mission cells,” IEEE Trans. Electromag. Compatibility, vol. EMC-16, no. 4, pp. 189-195, 1974.

[12] K. Malaric, “Electromagnetic compatibility analysis using TEM cells,” OSEE Online Symposium for electronic engineers, no. 14976, 2001. [13] P. Kay, “Field distribution and over-moding in a TEM cell,” 17th Int.

Zurich Symp. Electromag.c Compatibility, Singapore, pp. 457-460, 2006. [14] A. El Abbazi, “Etude et r´ealisation d’une nouvelle cellule TEM `a support rotatif pour des mesures CEM des circuits int´egr´es: application du mod´ele ICEM,” Ph.D. dissertation, university of Rennes 1, 2006. [Online] https://tel.archives-ouvertes.fr/tel-00143891.

[15] N. Boriraksantikul, “A TEM CELL design to study electromagnetic radiation exposure from cellular phones,” MS.C. dissertation, university of Missouri, 2008.

[16] P. F. Lopez, A. Ramanujan, Y. V. Gilabert, C. Arcambal, A. Louis and B. Mazari, “A radiated emission model compatible to a commercial electro-magnetic simulation tool,” 20th International Zurich Symp. Electromag. Compatibility, Zurich, Switzerland, pp. 369-372, 2009.

[17] C. Iftode, S. Miclaus, “Design and validation of a TEM cell used for radiofrequency dosimetric studies,” Progress Electromag. Research, vol. 132, pp. 369-388, 2012.

[18] K. Kwak, J. Kim, J. Kim, “Near-field validation of dipole-moment model extracted from GTEM cell measurements and application to a real application processor,” IEEE Trans. Electromag. Compatibility, vol. 60, no. 2, pp. 423-434, 2018.

[19] A. De Vita, R. Gaffoglio, B. Sacco, “Characterization of an open GTEM cell with the COMSOL multiphysics software,” proc. COMSOL Conf., Lausanne, Switzerland, 2018.

[20] M. Koohestani, R. Perdriau, J.-L. Levant, and M. Ramdani, “A novel passive cost-effective technique to improve radiated immunity on PCBs,” IEEE Trans. Electromag. Compatibility, early access, 2019.

[21] K. Satyaprasad, P. Siva Kumar, N. Deepika Rani, “Design and validation of transverse electromagnetic (TEM) cell for measurement of pulsed transients,” Lecture Notes in Electrical Eng., vol 521, Springer, 2019. [22] Tekbox TBTC1 TEM cell for EMC pre-compliance testing, Rekirsch

Elektronik Inc. [Online] http://www.rekirsch.at/Tekbox+TBTC1.htm [23] Ansys Inc., “High Frequency Structure Simulator (HFSS),” 2018,

Elec-tromagnetics Suite v19.1, Pittsburgh (PA), USA.

[24] S. Gabriel, R. W. Lau, and C. Gabriel, “The dielectric properties of biological tissues: II. Measurements in the frequency range 10 Hz to 20 GHz,” Phys. Med. Biol., vol. 41, pp. 2251-2269, 1996.

[25] FCC-TEM-JM2 TEM cell, Fischer Custom Communications Inc. [On-line] https://www.fischercc.com/products/fcc-tem-jm2

[26] Fair-Rite products corp. [Online] https://www.fair-rite.com/product/flat-cable-emi-suppression-cores-2643164051/

[27] Kiwi Millenium version 4.5, Centre National d’Etudes Spatiales (CNES) [Online] https://www.planete-sciences.org/espace/IMG/pdf/manuel-kiwi-millenium.pdf

[28] ATMEL SAMR21 low-power smart ARM-based wireless microcon-troller. [Online] http://ww1.microchip.com/downloads/en/devicedoc/sam-r21-datasheet.pdf/

Mohsen Koohestani (S’13−M’15−SM’19) received the PhD degree (with honor) in electromagnetics

from the ´Ecole polytechnique f´ed´erale de Lausanne

(EPFL), Lausanne, Switzerland, and the Universi-dade de Lisboa (ULISBOA), Lisbon, Portugal, in 2014.

From 2014 to 2018, he was a Post Doc-toral Fellow with the Institut d’Electronique et de T´el´ecommunications de Rennes (IETR), Universit´e de Rennes 1, working mainly on biomedical appli-cations of wireless power transfer systems (WPTs). He is currently an Associate Professor at ESEO, Angers, France, in the field of RF electromagnetic. He is also an associate researcher at IETR. He has authored over 40 peer-reviewed scientific papers. His main field of research includes novel design and analysis of antennas and microwave components, design and characterization of human body models, dosimetry and exposure assessment, and WPT and energy harvesting.

He is a member of the France IEC standardization working group in the SC47A subcommittee on the EMC for integrated circuits.

Mohamed Ramdani (M’01−SM’07) was born in Oujda, Morocco. He received the PhD Degree in Microelectronics from Universit´e Paul Sabatier, Toulouse, France, in 1989, and the “Habilitation Diriger des Recherches” (Accreditation Degree) from Universit´e de Rennes 1, France, in 2004.

From 1991 to 2008, he was an Associate Professor at ESEO, Angers, France, in the fields of microelec-tronics and microwave elecmicroelec-tronics. In 2008, he was promoted to Full Professor. His research interests are EMC of integrated circuits and integrated circuit design. He has authored several book chapters in EMC of ICs, and many peer-reviewed scientific papers. Prof. Ramdani is a member of AFNOR (French section of the IEC). He served as General Chair of EMC Europe 2017, which was held in Angers, France. He was elected Vice-Chair of the IEEE France Section in 2016.

0123567289153 588 762 8932 2 57623371236 59 7 7232 5953 63 7271718 72  52 572 

 !"#$%&'()(''*+*,,+('*&%-+*(&'-*&.*./%(//(&+(/ 00

1234%5667889:;4%95<=:5<>)4?9@@=<A:B<A<C5<8A>DE5<FG9FF9=<@9<A G=>A@6=E8=<>78D9<HI9EAF:JKLLLMN!OP#L"Q NRS!TO#URSV! #:W=@42X: <=42:YY4Z0[\Z[]:=W4[^^24

1Z3;4<>E9A7:+4_=<A:)4=8`7AD:4'AG=7@9<:5<>a4b,4,5EG5<D9AE: B%7@D98=<>78D=EEA>78D9=< DA8C<9`7A 6=EG=>A@9<H 8=GG=<bG=>A 87EEA<DF=<85c@Ac7<>@AF5DC9HC6EA`7A<8d6=E57D=G=D9WA5YY@985D9=<F:J KLLLMN!OP#L"Q NRS!TO#URSV! #:W=@4Z^:<=40:YY40eZ\0]2:)Ac4[^^]4 1X3/4C5c5<A:,4AF<9AE:5<>%4_@9<H@AE:BfDA<F9=<=6DCADE5<FG9FF9=<

@9<ADCA=Ed5YY@985D9=<I9DCG=>9gA>A<C5<8A>YAEb7<9Db@A<HDCY5E5GAb DAEF:JhNRT#L"Q NRS!TO#iP#:W=@4j[:YY4[Ze\[e^:[^0j4

1e3/4C5c5<A:,4AF<9AE:5<>%4_@9<H@AE:B<C5<8A>DE5<FG9FF9=<@9<A DCA=Ed$)EA`7A<8db>AYA<>A<D@9<AY5E5GADAEF5<>DCA9E9<FAED9=<9<58@5Fb F985@DE5<FG9FF9=<@9<AA`75D9=<F=@WAE:J9<hNRQ#KO #klSV#L"Q NRS!TO# URSV! #:[^0j:YY4j[X\jj04

1]3)4.58C9>95<>/4*m58CA<m=:L"Q NRS!TO nQon"pKO N!Q nROqnr MN!OPSnPPnROsnOPtoNRS U"!PPnQ!"MrRNl Ruoi!pn! nROLvvQ P4 FC7EFD:-4_4$w(*,EAFF:[^^]4

1x3a449DF8C5<>/4?4*m58CA<m=:B=GY@AfbW5@7A>DE5<FG9FF9=<b@9<A Y5E5GADAEF5<>DCA9EEA@5D9=<D=DCAE5>95D9=<EAF9FD5<8A:JKLLLMN!OP# L"Q NRS!TO#URSV! #:W=@42X:<=4j:YY42ee\2]e:7H4[^^24 10^3,4'AH57`7A5<>4yA>>5G:B.AG5EmF=<DCADE5<FG9FF9=<b@9<A5YYE=58C

D=>ADAEG9<9<HDCA87EEA<D9<>78A>=<5c=WAbHE=7<>85c@AF:JKLLLMN!OP# L"Q NRS!TO#URSV! #:W=@4j^:<=40:YY4ee\]^:)Ac40x]]4

1003/44,9H<5E95<>'4A@@5<:B(<8=EY=E5D9<HWAED985@E9FAEF9<DCADE5<FG9Fb F9=<@9<AA`75D9=<FI9DCAfDAE<5@F=7E8AF:J9<hNRQ#KO #klSV#L"Q NRS!TO# URSV! #:[^^2:W=@4j:YY4xe2\xex4

10[34'4%AdAE:z! Nn{|O!"lPnP!Op|VV"npsnO!N|"T}N!4,C9@5>A@YC95: ,:-/$/(%:[^^^4

10j3,4'AH57`7A5<>4yA>>5G:URSV! n}n"n ~~L"Q NRS!TO~ n€tP UROQV P‚!P|€{|VV"nQ! nROPƒ4,5E9F:)E5<8A$+5W=9F9AE:[^^e4 10234.4,57@:|O!"lPnPRvz€"nQROp€Q RNMN!OPSnPPnROsnOP4=c=mA<:a:

-/$w9@Ad:0xx24

„…†‡ˆ‰Š‹‡Œ‡ˆ‰I5Fc=E<9<5H5:@HAE95:9< 0x]j4AEA8A9WA>DCA'-Ž4<H45<>%4<H4 >AHEAAFc=DC9<5AE=<57D98F5<>6E=GDCAAE=<57D98F (<FD9D7DA=6@9>5:@9>5:@HAE95:9<[^^Z5<>[^^]: EAFYA8D9WA@d:DCA%4/84>AHEAA9<YCdF98FŽ8=GY@Af <5D7E5@5<>9<>7FDE95@FdFDAGF6E=G DCA-<9WAEF9Dd =6.A<<AF0:.A<<AF:)E5<8A:9<[^0^:5<>DCA,C4'4 >AHEAA9<A@A8DE=<98F5<>DA@A8=GG7<985D9=<F6E=G (<FD9D7D<5D9=<5@>AFF89A<8AF5YY@9`7AAF>A.A<<AF: .A<<AF:9<[^024

ADCA<‘=9<A>,5F85@(<FD9D7DA:@AEG=<Db)AEE5<>: )E5<8A5F5,=FD>=8D=E5@.AFA5E8CAE:ICAEACA>AWA@=YA>DE5<FG9FF9=<@9<AG=>b A@F6=E8=GY@Af<ADI=EmF:E5<>=GC5E<AFFG=>A@F:YAEb7<9Db@A<HDCY5E5GADAEF 8=GY7D5D9=<7F9<H%=%’% GADC=>F:5<>Y5ED989Y5DA>9<7<8AED59<Dd:EAb @95c9@9Dd:5<>FA<F9D9W9Dd5FFAFFGA<D6=E8=GY@Af%YE=c@AGF4/9<8AYE9@ [^0X:CAC5FcAA<I9DCDCA<5@dD985@.“''AY5EDGA<D=6/56A<H9<AAEb 9<H/AEW98AF“*A8C<=@=H9AF@D>4:+5W5@:5<5>5:ICAEACA9F87EEA<D@dFAEW9<H 5F5.AFA5E8C5<>'AWA@=YGA<D/89A<D9FD5<>9F9<W=@WA>9<DCAEAF=@7D9=<=6 8=GY@Af%(5<>HE=7<>9<HYE=c@AGF49FG59<EAFA5E8C9<DAEAFDF9<8@7>A A@A8DE=G5H<AD988=GY5D9c9@9Dd5<>9<DAE6AEA<8A:A@A8DE=G5H<AD98G=>A@9<H5<> F9G7@5D9=<ŽDE5<FG9FF9=<@9<ADCA=Ed:5<DA<<5DCA=Ed:AD84:7<8AED59<DdG=>A@9<H 6=EA@A8DE=G5H<AD985YY@985D9=<FŽE9FmF:EA@95c9@9Dd:5<>FA<F9D9W9Dd5<5@dF9F:5<> EA8A<D@dHE=7<>9<H5<>A5EDC9<H5<5@dF9F6=EW5E9=7F5YY@985D9=<F4

”‹•–•——‰ ˜‰™ˆ•‰š Ž%›^2\/%›0^ EA8A9WA> DCA >9Y@œ=GA>›9<HA<9A7E>AHEAA6E=GDCA8=@A-<9WAEF9b D59EA>›(<HA<9A7EF>A+9@@A:+9@@A:)E5<8A:9<0xx^:5<> DCA,C4'4>AHEAA9<A@A8DE=<98F6E=GDCA-<9WAEF9Dd =6+9@@A:+9@@A:9<0xxj4

)=@@=I9<H5=<AdA5EYAE9=>5D&.:%A7>=<: 5F5<FF9FD5<D/89A<D9FD9<DCA%>9W9F9=<:CAI5F I9DCDCA+5c=E5D=Ed=6.5>9=,E=Y5H5D9=<5<>@A8b DE=<98F:-<9WAEF9Dd=6+9@@A:5F5.AFA5E8CAE5DDCA A<DEA5D9=<5@>A@5.A8CAE8CA/89A<D9g`7A6E=G 0xx2D=0xxe4)E=G 0xxeD=[^^[:CAI5FDCA'9b EA8D=E=6A<DEA>›D7>AFAD>A.A8CAE8CAFA<,E=DA8D9=<@A8DE=G5H<AD9`7A: 5<=<YE=gD=EH5<95D9=<6=EEAFA5E8C:AfYAED9FA:5<>DE59<9<H9<%:5<>EAb @5DA>58D9W9D9AF:c5FA>9<+5W5@:)E5<8A4A8=b6=7<>A>*_% 9<0xx]:5 YE9W5DA8=GY5<dFYA895@9A>9<D7E<bmAdFdFDAGF6=E%GA5F7EAGA<DF458m D=./9<[^^[:CAC5FcAA<F9<8ADCA<I9DCDCA(<FD9D7DA=6@A8DE=<98F5<> *A@A8=GG7<985D9=<F=6.A<<AFŽ(*.:.A<<AF4AI5F5YY=9<DA>5F5/A<9=E .AFA5E8CAE5D./9<[^0j:5<>F9<8A[^0[:C5FcAA<DCA=bA5>=6DCA<b DA<<5F5<>%98E=I5WA'AW98AF'AY5EDGA<D:(*.49FEAFA5E8C58D9W9D9AF5EA G59<@d>A>985DA>D=9<DAE6AEA<8A5<5@dF9F=<85c@AC5E<AFFAFŽ9<8@7>9<HA@A8DE=b G5H<AD98D=Y=@=Hd:EAWAEcAE5D9=<8C5GcAEF:<A5EbgA@>YE=c9<H:5<>EA8A<D@dD= DCA5<5@dF9F=67<8AED59<DdYE=Y5H5D9=<9<%G=>A@9<H4

‡šž…Ÿ–•ˆ –‰šŽ%›x]I5Fc=E<9<y7E98C:/I9DAEb @5<>:9<0xXj4AEA8A9WA>DCA<H9<AAE›F>AHEAA9< 8=GY7DAEF89A<8A6E=G(:+9@@A:)E5<8A:9<0x]x: DCA'Ž%4/4>AHEAA9<57D=G5D98F’E=c=D98F:5<> DCA,C4'4>AHEAA9<A@A8DE=<98F:9<0x]x5<>0xx[: EAFYA8D9WA@d:c=DC6E=GDCA-<9WAEF9Dd=6+9@@A:+9@@A4 ADCA<‘=9<A>DCA)EA<8C5D9=<5@(<FD9D7DA6=E *E5<FY=ED5<>/56ADd.AFA5E8CŽ(.*/:?9@@A<A7WA >›F8`:)E5<8A5F5.AFA5E8CAE:ICAEACAI5FDCA (<bC5EHA=6DCA.“'58D9W9D9AF9<%=6HE=7<> DE5<FY=ED5D9=<FdFDAGF4(<[^^[:CA‘=9<A>DCA'Ab WA@=YGA<D'9W9F9=<:;E=7YA,/:?A@9db?9@@58=7c@5d:)E5<8A:ICAEACAI5F F788AFF9WA@dDCA(<bC5EHA=6DCA%>AF9H<58D9W9D9AF:DCA%’5<DA<<5 F9G7@5D9=<58D9W9D9AF:5<>g<5@@dDCA%67@@WAC98@AW5@9>5D9=<58D9W9D9AF4(< [^00:CAG=WA>D=DCA.AFA5E8C'9W9F9=<ICAEACA9F87EEA<D@d5<%fYAED 5<>EAFY=<F9c@A=6DCA%’5<DA<<5EAFA5E8C58D9W9D9AF49FEAFA5E8C9<DAEAFDF 9<8@7>AA@A8DE=G5H<AD989<DAE6AEA<8AF=<,F:cAC5W9=E=6A@A8DE=<988=GY=b <A<DF9<A@A8DE=G5H<AD98A<W9E=<GA<DF:8=7Y@9<HD=I9EAFDE78D7EAF:DAFDGADCb =>F:DAFD6589@9D9AF:%G=>A@9<H5<>F9G7@5D9=<=657D=G=D9WAA@A8DE98Y=Ib AEDE59<F:G7@D98=<>78D=EDE5<FG9FF9=<@9<AF:<AIG5DAE95@F:5<>FYA89g8WAC98@A 5<DA<<5F4

Philippe Besnier (M’04−SM’10) received the

diplˆome d’ing´enieur degree from ´ecole universitaire d’ing´enieurs de Lillle (EUDIL), Lille, France, in 1990 and the Ph. D. degree in electronics from the university of Lille in 1993. Following a one year period at ONERA, Meudon as an assistant scientist in the EMC division, he was with the laboratory of radio-propagation and electronics (LRPE), Univer-sity of Lille, as a researcher (charg´e de recherche) at the Centre National de la Recherche Scientifique (CNRS) from 1994 to 1997. From 1997 to 2002, Philippe Besnier was the director of Centre d’Etudes et de Recherches en Protection Electromagn´etique (CERPEM): a non-for-profit organization for research, expertise and training in EMC and related activities, based in Laval, France. He also co-founded TEKCEM in 1998, a small business company specialized in turn-key systems for EMC measurements. Back to CNRS in 2002, he has been since then with the Institute of Electronics and Telecommunications of Rennes (IETR). Philippe Besnier was appointed as CNRS senior researcher (directeur de recherche au CNRS) in 2013. He was co-head of the “antennas and microwave devices research department of IETR between 2012 and 2016. He headed the WAVES (electromagnetic waves in complex media) team during the first semester of 2017. Since July 2017, he is now a deputy director of IETR. His research activities are mainly dedicated to interference analysis on cable harnesses (including electromagnetic topology), theory and application of reverberation chambers, near-field probing and uncertainty quantification in EMC modeling.

(12)

ACCEPTED MANUSCRIPT

Jean-Luc Levant received the PhD degree from Institute National des Sciences Appliques (INSA), Rennes, France, in 2007. He is currently working in Microchip as a senior EMC expert. His main research interests include ICs modeling, power and signal integrity, and electrical fast transient analyses. He is a member of the IEC standardization work-ing group in the SC47A subcommittee on the EMC for integrated circuits.

Richard Perdriau (M’01−SM’07) was born in Angers, France, in 1971. He received the Engineer-ing Degree in Electronics and Computer Science from ESEO, Angers, in 1992, the PhD Degree in Applied Sciences from Universit´e Catholique de Louvain, Louvain-la-Neuve, Belgium, in 2004, and the “Habilitation Diriger des Recherches” (Accred-itation Degree) from Universit de Rennes 1, France, in 2012.

From 1992 to 2012, he was an Associate Profes-sor at ESEO, in the fields of microelectronics and embedded systems. In 2013, he was promoted to Full Professor. His research interests are EMC of integrated circuits, mixed-signal hardware description languages and integrated circuit design. He served as the Vice-Chair and Technical Program Co-Chair of EMC Europe 2017.

Figure

Fig. 1 Simulation setup inside a standard TEM cell  a PCB without coil
Fig. 2. Simulated ∣∣E∣∣- and ∣∣H∣∣-field distributions inside an open TEM cell for 1 W input power: (a) unloaded, (b) in presence of a human hand; (c) loaded by a small piece of ferrite (ε r = 12, µ r = 1000); (d) loaded by a small piece of ferrite (ε r =
Fig. 3. Simulated ∣∣E∣∣- and ∣∣H∣∣-field distributions inside a closed TEM cell for an input power of 1 W at its two first resonance frequencies: (a) unloaded;
Fig. 3 exhibits the ∣∣E∣∣- and ∣∣H∣∣-field distributions inside a closed TEM cell (FCC-TEM-JM2 [25])
+4

Références

Documents relatifs

For a dielectric that is contacted by means of a conducting liquid, we have observed the intrinsic Abraham force, which we find to be proportional to ∂ t ( E × B ), thereby

We prove that if the initial magnetic field decays sufficiently fast, then the plasma flow behaves as a solution of the free nonstationnary Navier–Stokes equations when |x| → +∞,

In contrast, the presence of a magnetic field perpendicular to the shell collision axis (mhd1r) flattens the shock expansion law to about 0.4.. A test simulation similar to mhd1t,

The very good agreement between experimental and calculated losses observed in figures 3a and 3b stresses the fact that for the first time we can predict correctly the dependence

2014 The relation between the order parameters and the static dielectric constant of the smectic phases.. is investigated here, in the cases of one or more dipoles

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

The corresponding pair production rate in the presence of constant electric and magnetic fields was calculated in [22] using the exact solution of Dirac equation.. In [23] the

Nous avons ensuite pour- suivi cette ´etude en comparant la dynamique stochastique d’un moment magn´etique `a un mod`ele d´eterministe sur les moments de sa distribution