The AUDACE Collaborative Project

The AUDACE Collaborative Project: A joint innovation Program between Thales, Valeo, SMEs & Research Labs on Mechatronic Devices Reliability Hichame Maanane1 , Philippe Pougnet2, Jean Philippe Roux3, Jean Luc Lefebvre4, Moncef Kadi5, Olivier Latry6, Farid Temcamani7 , Jean Loup Alvarez 8, Francis Taulelle9, Philippe Eudeline1 1: THALES AIR SYSTEMS, ZI du Mont Jarret 76520 Ymare 2: VALEO, 14 avenue des Béguines, 95800 Cergy 3: CEVAA, Technopôle du Madrillet, 76801 Saint Etienne du Rouvray 4: PRESTO ENGINEERING, 2 rue de la Girafe, 14000 Caen 5: ESIGELEC/IRSEEM, Technopôle du Madrillet, 76801 Saint Etienne du Rouvray 6: NORMANDIE UNIVERSITE/GPM: Technopôle du Madrillet, 76801 Saint Etienne du Rouvray 7: ENSEA/ECIME: 6 avenue du Ponceau, 95014 Cergy 8: MB ELECTRONIQUE: 106 rue des Frères FARMAN, 78533 Buc 9: TECTOPSPIN, Institut Lavoisier, Université de Versailles Saint-Quentin en Yvelines, 78035 Versailles Abstract: This paper highlights important results of the four year collaborative project called "AUDACE". Through its six work package groups (WP), this project establishes a methodology based on experimentation and modelling with the ultimate goal to improve and optimize the design of mechatronic products to ensure their robustness/reliability under their operating conditions.

accelerated) benches. Implementation or experimental validation of dedicated "AUDACE" benches are made with two types of mechatronic products (from automotive and aerospace fields). Unfortunately, we cannot present in detail all the results obtained in this project. However, we invite you to visit the website of the project for more information (see Appendix).

Keywords: Mechatronics, reliability, robustness, automotive, aerospace

2. Innovative benches for characterization

1. Introduction

2.1 1D/3D Scanning Laser Doppler Velocimetry for mechanical and thermo-mechanical measurements from board level to component level

Embedded mechatronics is essential for the competitiveness of world leading companies in the automotive and aerospace industries. This new approach, combining mechanics and electronics showed failure phenomena that have not been analyzed in depth and are therefore not sufficiently under control. The objectives of this project are twofold:  Identify and classify these failure mechanisms to extract key ones, to analyze and model them in order to ensure in one hand the robust design of mechatronic products shipped, and secondly to make the necessary improvements to meet the growing expectations of reliability and durability required now in the two application fields (Automotive and Aerospace).  Find and develop ways that will validate the robust design of the new complex mechatronic objects. Ultimately, this project will allow OEM and automakers, as well as main actors of the aerospace sector, to achieve the levels of quality and competitiveness which are essential to ensure the success of innovations resulting from mechatronics. The paper will focus on some important experimental achievements, namely development of innovative characterization and testing (accelerated and highly

An innovative approach to investigate the dynamic behaviour of an Electronic Board was established by CEVAA using Laser-Doppler Vibrometer (LDV) technology. The principle of LDV is based on optical Laser interference, requiring two coherent light beams.

Figure 1: Principle of Laser Doppler Velocimetry measurement (Doppler Effect)

Figure 2: Laser Doppler Velocimetry test bench (1D configuration)

Page 1/8

The time signal of the observed response gives a precise indication of the velocity of the vibrating surface. Time data acquisition is performed for a large number of measurement points (up to 10 000 3D locations for a single component) and then FFT (Fast Fourier Transform) is applied in order to identify the frequency response of the whole component. Post processing and data analysis allows us to observe dynamic responses of the components and local deformations (in plane and out of plane deformations) under mechanical (vibration), electrical and thermal stresses (time domain and frequency domain with phase information). This new multi-physics experimental approach gives very precise and accurate information of the dynamic behavior of sensitive components which can explain failure and reliability problems. In addition to this operational measurement, Frequency Response Function (FRF) measurements can be performed under artificial excitation (piezo-electric actuator) and can bring all the useful information for experimental modal analysis (knowledge of Eigen frequencies, deformations and damping factors) of the whole component. This modal base can then be used for Finite Element (F.E) model tuning and durability optimization of the components and board design.

the metal cavity and has presented a new approach to estimate the radiated field in the cavity from measurements made in the absence of the cavity [1]. This method was validated for all frequencies except resonant frequencies. 2.2.1 Cavity effect on electromagnetic cartographies of a simple RF circuit To validate the proposed approach, a simple RF microstrip line is considered. The electromagnetic field profiles are calculated, using a 3D electromagnetic simulator HFSS, on a plane located 2 mm above the microstrip line. The electromagnetic behaviour is studied in two configurations: microstrip line without cavity and microstrip line enclosed in metallic cavity. By comparing the electric field distribution for the both configurations at frequencies f1 = 3 GHz and f2=6.75 GHz in Fig.4 and Fig. 5, we noticed that at the frequency f1 there is no impact on the profiles except in the vicinity of the metal walls [1]. In contrast, at the frequency f2 which is one of resonant frequencies of the system the electromagnetic field distribution changes completely. We note the presence of a stationary wave which represents the cavity mode with an index (2, 1, 0). Therefore, no wave propagates along the line. The effect of the cavity intensifies by decreasing the size of the cavity or increasing the frequency. We found the same results for other resonant frequencies.

Figure 3: Experimental modal analysis using LDV measurements of wire bonds 2.2 EMC characterization from equipment up to component level A signal, propagating in RF electronic circuits enclosed in a metal cavity, can excite the cavity’s natural modes. In the context of RF amplifier reliability study against electromagnetic stress, a precise characterization of radiated emissions is obligatory. This characterization is often achieved by measuring the near field of the circuit without enclosing it in a cavity (the top cover of the cavity is kept open), which does not always resemble the electromagnetic behavior when the same circuit is shielded within the cavity. In the AUDACE project, IRSEEM (Research Institute of Embedded Systems) has studied the changes in the cartographies (electromagnetic fields mapping) of the electromagnetic field radiated under the influence of

Figure 4: Electric field distribution for f=3 GHz

Figure 5: Electric field distribution for f = 6.75 GHz

Page 2/8

2.2.2 Proposed approach for estimation of the EM fields in a closed cavity from open cavity cartographies Due to use of small sized cavities, making radiated field measurements in a closed cavity is extremely difficult. This motivates us to develop a novel methodology to predict the electromagnetic field radiated by a microwave circuit placed in a closed cavity from measurements made on the same circuit placed in an open cavity. The principle of the method is summarized in Fig.6. EM field mapping with Cover Open

Modeling by electric dipole

Inserting the model in HFSS

Encapsulate the model in a cavity

Mapping of the EM field in closed cover Figure 6: Schema of the procedure to build electromagnetic field in closed cover

2.3 Nuclear Magnetic Resonance for Moisture and ageing effects in electronic package Nuclear Magnetic Resonance spectroscopy measures the resonating frequencies in a strong magnetic field of a family of atoms (such as 1H, 13C, 29Si....). The spectrum exhibits resolved signals in high resolution mode. Those signals are assigned by their frequencies to atoms in a given environment, primarily chemical environment but also with physical characteristics.

Figure 7: The spectrometer is an Advance 3 Bruker NMR instrument, equipped with a 4.7 Tesla cryomagnet. The probe used is a 4 mm rotor type with Magic Angle Spinning capability. Above are the three components of the spectrometers: the superconducting magnet (top left picture), the electronic console (bottom picture), and the probe (top right picture).

Le rotor e cylindriqu diamètre quantité d rotor et de diamètre de produi puis mis haut cham

Figure 13. sample Photo de deux rotors Figure 8: The containers, the "rotors" are made of zirconia cylinders spun at speeds of 5-15 kHz. The samples are ground to powder and then Pour un échantillon inserted into the rotor.solide, les mesures

so l’échantillon à des vitesses de quelques kHz. Le rot 2.3.1 study: ageing of packaging resinsdoit être hom deCase son of axe de révolution. La poudre d’instabilité du rotor les chocs peuvent endommage An example of ageing of packaging resins is Dans la sonde, le rotor est toujours incliné p presented below. La rotation de l’échantillon sert à moyenner les int Hz) sont inférieures à la fréquence de rotatio l’interaction dipolaire. L’angle d’inclinaison est dipolaires. Le couplage dipolaire est une interactio que deux spins. Chaque dipôle crée un petit cham Cela donne lieu à un élargissement assez important 3/8 d’éliminer (la réorientation rapide Page des molécules m un liquide, milieu isotrope).

Figure 9: MOSFET with epoxy packaging The epoxy based casing used to protect D-PAK MOSFET components has been identified by 13C NMR analysis. The cause of failure has been discovered by 1H NMR as due to the breakage by water of by temperature of reticulate sites between the silica particles.

L e s a g e n ts d e ré tic u la tio n d e s ré sin e s é p o x y d e s s o n t d e s a m in e s, d e s a n h y d re s d ’a c id e o u d e s p h é n o ls [1 3 ]. D a n s le s e n ro b a g e s é le c tro n iq u e s, le s a g e n ts d e ré tic u la tio n le s p lu s h a b itu e ls so n t l’a n h yd re d ’a c id e B T D A e t le p h é n o l P h é n o l n o v o la q u e (c f. F ig . 1 ). L e d ia m in o d y p h e n yl su lfo n e (D D S , F ig . 2 0 ) e st u n a g e n t d e ré tic u la tio n a m in é trè s ré p a n d u a v e c le s ré sin e s é p o x yd e . D a n s n o s ré s in e s n o ire s , l’a g e n t d e ré tic u la tio n ! n ’e st p a s le D D S c a r se s c a rb o n e s so n t to u s d e s F i g u r e 2 0 . P o l ym èr e d e D D S [ 2] a ro m a tiq u e s o r d a n s n o s ré sin e s l’a g e n t d e ré tic u latio n p ré se n te d e s c a rb o n yle s (C = O ) c o n tra ire m e n t a u D D S . L ’a g e n t d e ré tic u la tio n e st c e rta in e m e n t d e ty p e a n h y d re d ’a c id e (à c a u s e d e s c a rb o n y le s). L ’a g e n t d e ré tic u la tio n le p lu s p ro b a b le e st le b e n z o p h é n o n e té tra c a rb o x yliq u e d ia n h y d re d ’a c id e (B T D A , c f. F ig . 2 0 ) c a r il e st a c tu e lle m e n t le p lu s u tilis é [1 4 ].

Figure 10: different epoxy types, with silica particles L e p ic à 1 6 p p m n e c o rre sp o n d p a s à D G E B A n i à B T D A . C e d é p la c e m e n t c o rre sp o n d à fillings the d e s m used é th yle s (C Hto ) d ’umatch n e c h a în e c o u rte c o m mthermal e d u p ro p a n e o u expansion d u p ro p a n o l. Il e st p o s sib le q u e c e sig n a l so it issu d ’u n m o rc e a u d e c h a în e d e D G E B A ré s u lta n t d e sc issio n s. C e p e n d a n t coefficients of the resin and of Résines epo x ydes d’enr élect c e n ’e st p a s la s e u le h y p o th è se c a r le s ré sin e s p e uo v e nbage t the ê tre c o nelectronic s titu é e s d ’u n rm o é la nnique g echip de

In both packaging situations the nature of the resin could be identified and the ageing of the moulding compound followed by measuring the variation of area under each resonance. The relation between the chemical group affected by ageing and the stress applied to the material can be correlated. The information of the environment fragility relative to stress can be given to the formulator to be improved. 3. Innovative benches for accelerated and highly accelerated testing 3.1 SUPER HAT from equipment up to board level SUPER HAT (Highly Accelerated Testing) is an innovative solution for checking and improving the robustness of a mechatronic product under combined harsh environmental conditions (thermal, mechanical and moisture). The features of the SUPER HAT bench are the following:   

3

p o ly m è re s, le p ic o b s e rv é à 1 6 p p m p e u t p a r e x e m p le p ro v e n ir d e s m é th y le s d e l’é p o x y d e

ré so l n o v o lair q u er(Eadiat O C N , F ig . io 1 ). Ln a fig13 u re C 2 1 illu u tio n d eat s p icts r c o ibut n fo rm é m eio n t à ns doon .co uble { 1streHl’a}ttribles s h y p o th è se s. (c f. A n n e x e /ta b le d e s d é p la c e m e n ts c h im iq u e s ).



Expose the Device under Test (DUT) to 90 % relative humidity at 95 °C. Fast temperature ramp down from 95°C to -40°C to freeze potential humidity in contact with electronics thus revealing sealing defects. Combine these stresses with random vibrations (10 Hz to 5 kHz 70 gRMS) to precipitate defects. Drive and synchronize the QUALMARK chamber and the ESPEC humidity chamber to optimize the humidity test profile in the environment of the DUT.

F i g u r e 2 1 . Sp ectr e R M N 1 3 C d e l a r ési n e n o i r e fo n cti o n n el l e, o b ten u p a r C P M A S à u n e vi tesse d e r o ta ti o n d e 9 k H z.

Figure 11: 13C CPMAS NMR spectrum and ! L e v ie illisse m e n t h yg ro th e rm a le n ’a p a s e n tra în é d e d iffé re n c e s n o ta b le s e n tre le assignment of each atomic carbon to its sp e c tre d e la ré sin e d é fa illa n te e t c e lu i d e la ré sin e fo n c tio n n e lle , so itenvironment le s sc iss io n s n e so n t p a s p ro d u ite s e n q u a n tité su ffisa n te (lim ite d e d é te c tio n ) so it le s ra ie s so n t tro p la rg e s p o u r corresponding resonance. p e rm e ttre d e re m a rq u e r d e s d iffé re n c e s (m a n q u e d e ré so lu tio n ). L e s m e s u re s su r le s n o y a u x d e c a rb o n e 1 3 C n e p e u v e n t p a s p e rm e ttre la d é te c tio n d e la d é fa illa n c e .

. Figure 13: SUPER HAT bench synoptic 3.2 EMC robustness testing from equipment up to component level

Figure 12: 1H MAS NMR spectrum and assignmentof each atomic hydrogen environment to its resonance.

In order to study the reliability of RF High Power Amplifiers (HPA) in their real environment, a study of the behavior of AlGaN/GaN HEMTs performances under electromagnetic stress is achieved by IRSEEM [2]. The DUT has undergone several stress combinations (electromagnetic stresses, electromagnetic and RF stresses, electromagnetic

Page 4/8

and DC stresses …). A near field setup is used to apply an electromagnetic field to the DUT (fig. 4). Degradations in DC and power characteristics are observed for all stress types. These could be associated with electron trapping within the AlGaN barrier and AlGaN surface leading to depletion of the 2-DEG. Fig. 14 illustrates the near-field disturbance setup. It is composed of three sets of equipment. The first set is used to feed the CW board such as an RF signal generator to generate an input power, with a power amplifier to amplify the input signal and an external power supply to feed the transistor. The second set includes all the measurement units such as a current meter to measure drain current, a spectrum analyzer to measure output power, a directional coupler used as an attenuator and a DC analyzer to measure all DC characteristics. The third set is composed of equipments used for the generation of the electromagnetic disturbances such as an RF signal generator, a power amplifier and a magnetic probe.

hours we stop the stress and we measure the transistor characteristics. The probe was fed by a RF generator (0 dBm) at 3 GHz connected to a 40 dB power amplifier and it is located 1 mm above the micro-strip line connecting the gate. The same conditions are applied to the probe located above the micro-strip line connecting the drain, or above the package, but no degradation can be seen in these conditions. That’s why the transistor was stressed on the micro-strip line connecting the gate under the same conditions as when the probe is located 5 mm in front of the gate input. To examine the effects of the electromagnetic near-field stress on our component, the HEMT is characterized by the use of an Agilent HP-4142 Semiconductor Parameter Analyzer. Biasing mode and monitoring were conducted under computer control using commercial software IC-CAP from Agilent. The critical parameters related to these characteristics are the maximum drain current (Idmax), the threshold voltage (Vth), the peak DC transconductance (gm_peak), the drain-source resistance (Rds), the power gain and the drain efficiency (DE). 0.4

100

0.3

80

Ids (A)

60 0.1 40

Rds (Ohm)

0.2

0

20

-0.1

Figure 14: The near-field disturbance setup The studied device uses an AlGaN/GaN HEMT structure. It was mounted in a 50 Ohm application board tuned for CW operation at 3.0 GHz. It is a power amplifier that was biased at 28V, 300mA and produced 50W of linear power, 60% drain efficiency and 11 dB of gain. Fig.15 shows the test fixture used for electromagnetic stress tests.

Figure 15: Board tuned for CW of AlGaN/GaN HEMT We apply to this board a pulsed input power of 30 dBm that has a 500 μs period with a duty ratio of 50%. The drain current is equal to 1.29 A. Every two

Before stress After 2h of stress After 4h of stress 0

5

10

15

20

25

0 30

Vds

Figure 16. Ids and Rds in function of Vds before and after electromagnetic stress A variation of all DC characteristics over time was recorded. Device degradation was defined as a drop of the maximum drain current of more than 16%, it was 369.7 mA before stress: drain current reduced to 311.15 mA after stress. Drain resistance also increased by 18%: it was 76.17 Ω before stress and rose to 90.41Ω after stress, as shown in Fig.16. In fact, using the magnetic coupling between the probe and the PCB tracks, power is created at the transistor input. This power represents the electromagnetic perturbation which is superimposed on the transistor input power and leads to all these degradations. These degradations could be associated to the presence of traps in AlGaN\GaN HEMT layer. Under stress, carriers may be excited and fill up such traps. In the case of acceptor-type traps, these become negative charged after being filled and act as a second gate. In fact significant

Page 5/8

numbers of electrons are transferred from 2-DEG to the barrier and surface region and are trapped in the AlGaN layer causing a reduction in the 2-DEG carrier density due to the charge neutrality condition and behave as a virtual gate causing a decrease in the drain current. 3.3 Electrical Over Stress bench from board up to component level 3.3.1 Bench purpose and description: The architecture of the Transient Power Generator (TPG) bench is designed to force and monitor any transient voltage waveform to a DUT by a large range of frequencies, voltages and currents. The capabilities of this bench are:  Frequency bandwidth: DC to 1MHz  Voltage: up to 1200V at 100mA  Up to 160V at 8Amp  Hot testing: Up to 125°C The current is limited by a resistance mounted on the test head while the voltage pulse is applied. Also, it is monitored using a 0.1Ω serial resistor connected to the DUT. The Software application was developed under LABVIEW (See the layout of the bench in Fig. 17).

Figure 17. Photo of the TPG bench 3.3.2 Case study: In power electronics applications in automotive industry, MOSFETs are used in switching mode. Electrical Overstress (EOS) failures in transistors are generally the consequence of Over-Voltage Stress (OVS) events. The intensity of the voltage and the current in power electronic systems is now increasing. Consequently, the over-voltage events occur at higher levels thus causing failures. Unlike the Electrical Static Discharges (ESD), OVS is not necessarily caused by neighbouring systems. On the other hand, the circuit design may induce OVS events. OVS mostly relates to capacitances and

inductances in the transistor itself (Wire bonding) and in the application (connector). In the case of the H-bridge circuit in Fig.18, transistor T1 is in PWM mode and T3 is ON-state. T2 and T4 are OFF-state. The current uses the path from T1 to T3 crossing the load, when a high voltage is applied on T3's gate. The current keeps the same direction across T3 & the load when a low level is applied on the same gate. Therefore VA is negative and the body diode in T2 is in direct bias mode. The measurements during the switching of VDS and IDS on the transistor show up as over-voltage when the transistor is turned-off (Fig.19). It is induced by input and output capacitances inside the MOSFET and the bonding and wire inductances around he MOSFET.

Fig.18: H-bridge schematic

Figure 19. VDS(t) & IDS(t) on transistor with resistive charge The experiment was designed to determine the robustness of this MOSFET relative to OVS triggering signals. We synchronize the pulses on the drain with the transient from ON-state to OFF-state on the gate (Fig.20). The voltage level on the drain increases and the results show conditions to turn-on avalanche transport mechanism.

Page 6/8

Some branches could have more capabilities like extreme temperatures as occur during the operating life of the amplifier. This capability for temperature stress is illustrated in figure 21.

Figure 20. Pulse on drain and gate during test OVS testing replicated the failure in MOSFET while the avalanche phenomenon is turned-on. This test demonstrates the influence of the parasitic capacitances and inductances within the transistor. 3.4 Temperature RF pulsed accelerated test bench from board up to component level This test bench is dedicated to radiofrequency power transistors for long period testing in operating mode. A central unit pilots several devices with data monitoring:  Radiofrequency power supplies, Pin, Pout, Pref.  DC power supplies, VDS, VGS, IDS, IGS.  Temperatures of base plates under components, During ageing tests, 15 parameters could be monitored on the bench. All parameters are summarized in Table 1. This group is synchronized by a salve generator able to create different kinds of patterns of pulses or groups of patterns all synchronized exactly like in an operating product (radar type for example).

Table 1: Parameters used on the bench.

Figure 21: Capabilities of cold (-40°C) testing for the components on this bench. One important and innovative feature of the bench is the ability to perform complementary electrical measurements without having to remove any devices under test. This functionality makes it possible to do some pulse IV characteristics or S parameters. Demonstrations have been done with different kinds of technologies using this equipment. For example, this bench has been used in order to demonstrate the reliability of high power amplifiers with LDMOS 330W [3]. This study based on 5000 hours in radar operating mode, demonstrates the very good performances of this type of amplifiers with LDMOS. Another type of application is the study of hot carriers and their modelling in LDMOS technology. The paper [4] refers to the entire protocol used on the bench in order to study the threshold voltage evolution over time with a hot carrier injection in oxide region. The reliability testing demonstrated here is only one of the many possibilities of this type of bench. The failure analysis used here is the first step of a much larger study of the root causes of failure. The study was done using HEMT AlGaN/GaN components. The first step is to choose the stress mode on the bench in order to understand which physical aggression (temperature, high field, current…) is responsible for performance degradation. This type of methodology was done with 50 W-class amplifiers under saturated pulsed-RF saturation life test with enhanced drain bias voltage. The next step is electrical investigation after stress with RF, IV, and small signals. Mainly due to gate leakage, investigations showed one phenomenon due to traps and another due Schottky contact change [5].

Page 7/8

A third step is to investigate with non-destructive technics (IR, photons emission microscopy) in order to reveal the site on the die where degradations are located. After that, the destructive techniques with a focused ion beam and a high-resolution transmission microscope confirm the Schottky contact degradation [6]. The final investigation aims to find the root cause of the degradation. In this example, X-rays reveal some voids in the soldier joint between the die and the package. These voids which are larger after ageing tests are located under the transistors with degraded Schottky [7]. All of these studies demonstrate the importance of the primary analysis using the temperature RF pulsed accelerated test bench. The possibility of monitoring multiple parameters help to establish the first hypothesis on failure causes. Electrical RF, IV and small signals could bring accurate hypotheses but have to be correlated with failure analysis methodology as showed above.

Microelectronic Reliability Journal, 51 (2011) 1783– 1787 [3]

O Latry, P Dherbécourt, K Mourgues, H Maanane, JP Sipma, F Cornu, P Eudeline, and M Mas moudi : "A 5000h RF life test on 330 W RFLDMOS transistors for radars applications", Microelectronics Reliability, 50(9-11) :1574–1576, 2010.

[4]

L. Lachéze, O. Latry, P. Dherbécourt, K. Mourgues, V. Purohit, H. Maanane, J. P. Sipma, F. Cornu, and P. Eudeline : "Characterization and modeling of hot carrier injection in LDMOS for L-band radar application", Microelectronics Reliability, 51(8) :1289–1294, August 2011.

[5]

JB Fonder, O Latry, C Duperrier, M Stanislaw iak, H Maanane, P Eudeline, and F Temcamani : "Compared deep class-ab and class-b ageing on AlGaN/GaN HEMT in S-band pulsed-RF operating life", Microelectronics Reliability, 52(11) :2561– 2567, 2012.

[6]

JB Fonder, L Chevalier, C Genevois, O Latry, C Duperrier, F Temc amani, and H Maanane : "Physical analysis of schottky contact on pow er AlGaN/GaN HEMT after pulsed-RF life test", Microelectronics Reliability, 52(9-10) :2205–2209, 2012.

[7]

O Latry, JB Fonder, L Chevalier, C Genevois, C Duperrier, F Temcamani : "Fiabilité d’amplificateurs de puissance à base de GaN: Analyse de la défaillance", Telecom2013, Marrakech, Maroc, 2013.

4. Conclusion The AUDACE program has demonstrated unique and concrete achievements in the field of characterization and stress testing of mechatronic products. With these innovative tools for characterization (Electromagnetic, Thermomechanical, electrical, thermal, mechanical,…) and testing (Electromagnetic, Electrical Overstress, thermal, mechanical, moisture,…) coupled with modelling, industrial manufacturers involved in mechatronic function development for future markets (electric car, more electric aircraft, RADAR, wind farms,…) are now able to tackle the potential issues of reliability and robustness of breakthrough mechatronic functions. 5. Acknowledgement The authors acknowledge MOVEO and the French Government for technical and financial support of the AUDACE program. 6. Appendix http://audacereliability.crihan.fr/Projet_Audace.html 7. References [1]

S. Khemiri, A. Ramanujan, M. Kadi, Z. Riah, A. Louis, “ Estimation of the Electromagnetic Field Radiated by a Microw ave Circuit Encapsulated in a Rectangular Cavity”, IEEE International Sy mposium on Electromagnetic, Long Beach, Los Angeles, USA 2011

[2]

S. Khemir i, M. Kadi, A. Louis, “ Reliability Study of AlGaN/GaN HEMT under Electromagnetic stress”,

Page 8/8