Experimental Results Obtained in the Vibrating Intrinsic Reverberation Chamber Frank Leferink Hollandse Signaalapparaten B.V. Research, Development & Technology Environmental Test Laboratory P.O. Box 42, 7550 GD Hengelo The Netherlands leferink@signaal.nl University of Twente Faculty of Electrical Engineering Department of Telecommunication Engineering P.O. Box 217, 7500 AE Enschede The Netherlands leferink@cs.utwente.nl Abstract: Measurements have been performed in a Vibrating Intrinsic Reverberation Chamber (VIRC). This chamber has varying angles between wall, floor and ceiling. Inside the VIRC a diffuse, statistically uniform electromagnetic field is created without the use of a mechanical rotating mode stirrer. In comparison to other reverberation chambers the VIRC displays an improved low frequency behavior, enabling faster, cost effective testing and allows for in-situ measurements. Several measurement results obtained in the VIRC are presented in this paper, such as the change in resonance frequenccy, the stirring ratio and the power density function. Introduction A reverberation chamber generally consists of a rectangular test room with metal walls and a mode stirrer, usually in the form of a large paddle, near the ceiling of the chamber. The EUT is placed in the chamber and exposed to an electromagnetic field while the stirrer slowly revolves. The average response of the EUT to the field is found by integrating the response over the time period of one revolution of the stirrer. The metal walls of the chamber allow a large field to be built up inside the chamber. The EUT is therefore exposed to a high field level consisting of several different polarizations [1,2,3,4]. The electromagnetic field inside a reverberation chamber is not time invariant as in most classic test methods utilizing free space set-ups, but provides a periodic electromagnetic environment, which is • randomly polarized, i.e. the phase between all waves is random, • spatialy uniform, i.e. the energy density in the chamber is uniform everywhere and • isotropic, i.e. the energy flow in all directions is the same. It has been shown that in a Mode Stirred Chamber (MSC) at least 200 modes should be available [3] in order to fulfill these requirements. In [6] several mode modification techniques, such as irregular walls, a moving wall and the Vibrating Intrinsic Reverberation Chamber (VIRC) have been discussed. The VIRC has vibrating walls in order to change the boundary conditions with large amplitude and short time variation. The VIRC is decribed in the next paragraph. Several tests have been carried out using the VIRC. The test set-ups and results are discussed showing the advantages of the VIRC compared to other techniques. Jean-Claude Boudenot Wim van Etten Thomson-CSF Comsys Hardening, Safe Instumentation and Systems Unit BP 82, 92704 Colombes France jean-claude.boudenot@ comsys.thomson-csf.com University of Twente Faculty of Electrical Engineering Department of Telecommunication Engineering P.O. Box 217, 7500 AE Enschede The Netherlands etten@cs.utwente.nl The Vibrating Intrinsic Reverberation Chamber The VIRC is a reverberation chamber with walls made of flexible conducting material. It is mounted in a rigid structure and connected to that structure via flexible rubber strings, as shown in Figure 1, and drawn schematically in Figure 2. By moving one or more ridges or one or more walls the modal behavior of the field inside the chamber is changed, and thus the resonance frequencies are changed. Because this frequency shift is much larger compared to what is possible with a conventional mode stirrer, the frequency range of the chamber is extended to lower frequencies compared to a reverberation chambers with equal dimensions. Note the natural corrugation of the flexible walls in Figure 1, which is beneficial for the spatial uniformity too. Another advantage is that the flexible chamber can be erected inside a standard anechoic chamber where the EUT has been installed for standard EMI tests, thus reducing measurement time and cost. Furthermore the VIRC does not need extra space inside the laboratory: it can be folded and put away fast. The most important advantage of the flexible structure of the VIRC is that it can be installed in-situ. Figure 1: The Vibrating Intrinsic Reverberation Chamber hanging in strings y • h x l z w • Figure 2: The VIRC: a flexible tent with irregularly shaped walls and varied angles of the wall-floor-ceiling interface Experimental characterisation of a reverberation chamber Numerous experiments could be carried out to characterize a reverberation chamber. Some of them are: • Resonance frequency variation By measuring the field strength as function of the frequency for several (fixed) boundary conditions, i.e. VIRC positions, the resonance frequencies and change in resonance frequencies can be observed. The change in resonance frequency provides an indication of the ‘stirring efficiency’ of the VIRC • Voltage Standing Wave Ratio By measuring the incident and reflected power as function of the frequency the voltage standing wave ratio (VSWR) can be obtained. The VSWR indicates at what frequency the maximum energy transfer can be achieved. • Stirring Ratio By recording the field strength as a function of time during a periodic or sufficiently random change in boundary conditions the so-called stirring ratio (SR) is obtained. This figure is defined as the ratio between the maximum and minimum field strength at a specific position and frequency: SR( f ) = • Emax Emin The stirring ratio indicates the effectiveness of a stirring mechanism. If the stirring ratio would be small then the boundary conditions are changed only slightly, and, as a result, the spatial uniformity could be low. In [8] a stirring ratio of at least 20 dB is considered as sufficient. The SR is an assess of the stirring efficiency in terms of field strength amplitude, which is an amplitude modulation. The change in resonance frequency due to a change in boundary condition is also an assess of the stirring efficiency, but now in terms of frequency modulation. Power Density Function The experimental power density function (PDF) is obtained by normalizing the measured absolute values of the electric field strength (in three directions) by their mean, and sorting the samples of total field strength from • the minimum field strength to the maximum field strength. By comparison of the experimental PDF with the theoretical PDF the quality of the chamber can be found: if the experimental PDF is equal to the theoretical PDF then the accuracy of a prediction is very high. Cumulative Density Function The cumulative probability density function (CDF) is obtained by integrating the PDF. Via the CDF a comparison of PDFs for different frequencies, or a comparison of chambers is facilitated. Spatial Field Uniformity Another important parameter is the spatial field uniformity (SFU). The SFU gives the ability to generate an isotropic, randomly polarised field which is stochastic equal in the whole volume of the chamber, exept near the walls. The SFU has not been measured yet. Quality Factor The chamber quality factor Q determines the efficiency of the chamber in relation to the power needed for generating a specific field strength or power density. The quality factor Q is an important parameter. It should be very high so that the generated field strength at a specific net input power is high, and thus a high efficiency. On the other hand, a low quality factor results in a higher mode overlap. The quality factor has not been measured yet. Measurement results The measurements were carried out using an experimental VIRC, as shown in Figure 1. The size of the VIRC was approximately 1.5 x 1.7 x 1.9 m. The VIRC was not vibrating for the measurement of the resonance frequency variation and the VSWR. For the other measurements (SR, PDF and CDF) the VIRC was randomly excited by three independent persons creating an amplitude variation of at least 10% of the length, width and heigth of the VIRC. Because the field strength inside the VIRC changes rapidly compared to the field inside the mode stirred chamber the standard field strength sensors, as used for EMI testing, are too slow. Therefore a very fast 3-axis sensor was used for the field strength measurements: the ET2003 [7]. The advantages of this sensor are the • low loading of the chamber, thus not decreasing the quality factor of the chamber, • small dimensions (60x60x60 mm), • large bandwidth with flat frequency response (10 kHz - 2.5 GHz), • large dynamic range (nearly 80 dB) and • short risetime (200 ps) making this sensor very useful for reverberation chamber testing. Resonance frequency variation The field strength in the three orthogonal directions was measured inside the VIRC as function of the frequency using the set-up as drawn in Figure 3. MHz and 200 MHz has been drawn in Figure 6. The resonance frequency variation in the VIRC is quite large, upto 10 MHz for the f011 mode, while the resonance frequency variation in the MSC is only a few MHz. Field strength [dBV/m/dBm] 20 optical coaxial ESS measuring receiver BR2000-3 optical receiver f101 f110 10 f011 0 Etotal 2 -10 Etotal 3 Etotal 4 -20 50 100 150 200 Frequency [MHz] Field strength [dBV/m/dBm] An electromagnetic field was generated inside the VIRC by a log-spiral antenna (150 MHz-1 GHz). The antenna was activated by the output of the tracking generator of the Rohde&Schwarz ESS measuring receiver. The field strength inside the VIRC has been measured using the Thomson-CSF ET2003 three-dimensional E-field sensor. The output signal of the sensor was fed to the optical receiver via an optical fiber. The output voltage level of the optical receiver was measured by the ESS measuring receiver for the three axes of the field. The total measured electric field strength between 900 MHz and 1 GHz has been drawn in Figure 4. During the measurements the VIRC was at two fixed positions (position 3 and 4), i.e. not vibrating. In this frequency range, 900 MHz – 1 GHz, numerous resonance frequencies are available and a distinctive resonance frequency can thus not be observed. 10 Etotal 3 Etotal 4 Figure 5: Total electric field strength as function of the frequency for VIRC position 2, 3 and 4 60 Field strength [dBuV/m] Figure 3: Measurement set-up for field modification 50 40 30 MSC, Stirrer pos. 1 MSC, Stirrer pos. 2 MSC, Stirrer pos. 3 (stirrer moved) 20 10 50 100 150 200 Frequency [MHz] Figure 6: Total electric field strength as function of the frequency for a MSC with 3 stirrer positions Voltage standing wave ratio 0 -10 900 950 1000 Frequency [MHz] Figure 4: Total electric field strength for VIRC position 3 and 4 between 900 MHz and 1 GHz In Figure 5 the total electric field strength between 50 MHz and 200 MHz has been drawn for the situation that the VIRC was at three fixed positions (position 2, 3 and 4). The resonance frequencies are evident. The resonance frequency variation due to the changed VIRC positions is also noticeable. A similar measurement has been carried out in a MSC with dimensions of approximately 2x2x2 m and for three stirrer positions. The total electric field strength between 50 The standing wave ratio (SWR) is defined as the ratio between the incident and reflected power at a specific position and frequency. If the measurement impedances are equal (50 Ω) then the voltage SWR (VSWR) in terms of incident Ui and reflected voltage Ur is: U VSWR = r Ui If the VSWR equals 1 then all incident energy is absorbed in the reverberation chamber. If the VSWR is high then a large portion of the energy is reflected. The VSWR for VIRC position 3 has been drawn in Figure 7, using the measurement set-up as drawn in Figure 3. From the VSWR we can calculate the net power injected into the VIRC. This net power has been drawn in Figure 8. The electric field strength for VIRC position 3 has been drawn in Figure 9. By comparison of Figure 8 and 9 the resemblance between net power and resonance frequencies is high, as could be expected. 5 11, 12 and 13 respectively. The measurement period was 4 s, with 8000 samples and thus 0.5 ms/sample. It can be observed in the figures that the SR for 150 MHz was less than 20dB. For 400MHz and 1 GHz the SR satisfies the rule SR>20dB, as stated in [8]. VSWR 4 3 2 10 50 1 00 150 2 00 F re q u en c y [M H z ] Figure 7: VSWR for VIRC position 3 Net Power [dBm] -1 8 -2 0 Field strength [dBV/m] 1 0 -1 0 -2 0 Ex Ez -2 2 -3 0 0 2000 4000 6000 8000 S am p le -2 4 50 100 150 Figure 11: Field strength as function of time or stirring ratio, for 150 MHz 200 F re q u e n c y [M H z ] 10 70 60 50 V IR C V IR C V IR C E to tal 40 Ex 3 Ey 3 Ez 3 3 Field strength [dBV/m] Figure 8: Net power for VIRC position 3 80 Field strength [dBuV/m] Ey E tot 30 0 -10 -20 Ex Ey Ez Etot -30 50 10 0 150 2 00 0 2000 F re qu en c y [M H z ] 4000 6000 8000 Sample Figure 12: Field strength as function of time, 400 MHz Figure 9: Electric field strength for VIRC position 3 10 The stirring ratio has been obtained by measuring the field strength as a function of the time at a fixed frequency. In the measurement period the VIRC was randomly excited by three independent persons creating an amplitude variation of at least 10% in width, heigth and length of the VIRC. The test setup has been drawn in Figure 10. The electromagnetic field inside the VIRC was generated by a R&S SMS-2 generator, amplified by a Kalmus amplifier and injected by the log-spiral antenna. The power level was adjusted to 30 dBm incident power. The field strength inside the VIRC has been measured using the Thomson-CSF ET2003 three-dimensional E-field sensor. The output signal of the sensor was fed via an optical fiber to the optical receiver. The output voltage levels, for every axis (x,y,z) one channel, was detected by a logarithmic detector. This detector makes use of an AD 8313 integrated circuit, which has a bandwidth of 2.5 GHz, a dynamic range of 70 dB, an amplitude flatness of ±1 dB, a noise level of less than –73 dBm and a response time of 40 ns. The stirring ratio, or the field strength as function of the time, for 150 MHz, 400 MHz and 1 GHz has been drawn in Figures Field strength [dBV/m] Stirring ratio 0 -10 -20 Ex Ez Ey E to t -30 0 2000 4000 6000 8000 Sam ple Figure 13: Field strength as function of time, 1 GHz Probability density function The electric field strength has been measured in the three ortogonal directions while the boundary conditions were changing; i.e. the VIRC was vibrating. The movements were random: three persons were independently moving the VIRC. The test set-up was as drawn in Figure 10. Measurements have been performed using a sample time of 1 ms for a signal frequency of 150 MHz decreasing to 0.1 ms for a signal frequency of 1 GHz. For every signal frequency (150 MHz, 500 450 SMS-2 M ea sured , V IR C , M P 1 , 30 0 M H z Number of Samples 400 amplifier dir. coupler splitter T h e o re ti c a l P D F , C h i (6 d o f) 350 300 250 200 150 100 50 optical 0 0 0 .5 1 1 .5 2 2 .5 3 3 .5 4 F ie ld S tr e n g th [V /m ] coaxial 5-channel logarithmic detector Figure 16: PDF for 300 MHz BR2000-3 optical receiver 400 350 AR1100 5-ch. analysing recorder T h e o r e ti c a l P D F , C h i( 6 d o f ) 250 200 150 100 Figure 10: Test set-up for measuring the field strength inside the VIRC as function of change in boundary conditions 200 MHz, 400 MHz, 600 MHz, 800 MHz and 1 GHz) 4000 samples have been recorded. The samples were recorded simultaneously for the incident and reflected power and the field strength for every axis, giving a total of 5 channels with 4000 samples. To obtain the probability density function the measured absolute value of the electric field strengths in the three directions were normalized. The samples of the total electric field strength have been sorted from the minimum field strength to the maximum field strength. This yields the experimental probability density function (PDF). The PDFs are shown for the frequencies 150 MHz, 200 MHz, 300 MHz, 400 MHz, 600 MHz, 800 MHz and 1 GHz in Figures 14 through 20. 50 0 0 0 .5 1 1 .5 2 2 .5 3 3 .5 4 F ie ld S tre n gth [ V /m ] Figure 17: PDF for 400 MHz 350 300 M e a s u re d , V I R C , M P 1 , 6 0 0 M H z Number of Samples T h e o r e ti c a l P D F , C h i( 6 d o f ) 250 200 150 100 50 0 0 0 .5 1 1 .5 2 2 .5 3 3 .5 4 F ie ld S tre n gth [ V /m ] Figure 18: PDF for 600 MHz 300 250 250 M ea sured , V IR C , M P 1 , 1 5 0M H z T h e o r e ti c a l P D F , C h i ( 6 d o f ) M ea sured , 8 0 0M H z, V IR C 1 T h e o r e ti c a l P D F , C h i ( 6 d o f ) 200 Number of Samples Number of Samples M e a s u re d , V I R C , M P 1 , 4 0 0 M H z 300 Number of Samples power meter 200 150 100 150 100 50 50 0 0 0 .5 1 1 .5 2 2 .5 3 0 3 .5 0 0 .5 F ie ld S tre n g th [ V /m ] 1 1 .5 2 2 .5 F ie ld S tre n g th [V /m ] Figure 14: PDF for 150 MHz Figure 19: PDF for 800 MHz 200 400 180 M easu red , 1G H z , V IR C 1 T h e o r e t ic a l P D F , C h i (6 d o f) T h e o r e ti c a l P D F , C h i( 6 d o f ) 300 140 Number of Samples Number of Samples 350 M e asure d, V IR C , M P 1 , 2 0 0M H z 160 120 100 80 60 250 200 150 40 100 20 50 0 0 0 0 .5 1 1 .5 2 2 .5 F ie ld S tre n gth [ V /m ] Figure 15: PDF for 200 MHz 3 3 .5 0 0 .5 1 F ie ld S tre n g th [ V /m ] Figure 20: PDF for 1 GHz 1 .5 The theory of the stochastic behaviour of reverberation chambers is described in detail in [4]. In [4, 9] it is shown that the magnitude of the electric field strength in any direction is Χ (Chi) distributed with two degrees of freedom, or Rayleigh distributed. This is based on the assumption that there is no direct path between the transmitting antenna and receiving sensor. If there is a direct path between the antenna and sensor then the Rayleigh distribution has to be modified resulting in the Rayleigh-Rice distribution. This distribution results in a longer tail for the total electric field strength distribution. From the comparison between measured and theoretical PDF in figures 14 to 20 we can observe that the measured PDF has a longer tail than the theoretical PDF. Because the dimension of the VIRC was small with respect to the transmitting antenna, the longer tail could be caused by a direct transmission path between transmitting antenna and field strength sensor. Cumulative probability density function The cumulative probability density function (CDF) for the measurement frequencies 150 MHz, 200 MHz, 300 MHz, 400 MHz, 600 MHz, 800 MHz and 1 GHz are shown in Figure 21. To estimate the quality of the chamber the experimental data has been compared to the theoretical curve. A similar measurement has been carried out in a MSC. The CDFs are drawn in Figure 22. The CDF of 200 MHz deviates considerably while the CDF of 150 MHz is nearly equal to the theoretical curve. 100% 200 MHz Cumulative Probability 80% Theoretical CDF CHI (6DOF) VIRC, MP1, 150MHz 60% Conclusion A new type of reverberation chamber to create a spatial uniform and isotropic electromagnetic field is the Vibrating Intrinsic Reverberation Chamber (VIRC). Several measurements have been performed within the VIRC. It was shown that the VIRC can even change the lowest resonance frequency by more than 10 MHz, whereas the conventional mode stirred chamber (MSC) resonance frequency changes only a few MHz. The stirring ratio was more than 40 dB for low frequencies, which proves that the VIRC is capable to change the boundary conditions adequately. The field strength at several frequencies has been measured as function of the time and the samples of the total electric field strength have been sorted from the minimum field strength to the maximum field strength, which yields the experimental probability density function. The corresponding cumulative probability density function (CDF) has been compared to the theoretical CDF and shows good resemblance. VIRC, MP1, 200MHz 1 GHz References VIRC, MP1, 300MHz 40% [1] VIRC, MP1, 400MHz VIRC, MP1, 600MHz 20% VIRC, MP1, 800MHz 150 MHz VIRC, MP1, 1000MHz 0% 0 0.5 1 1.5 2 Field Strength w.r.t. Mean Figure 21: CDF for the VIRC 100% Cumulative Probability Furthermore the CDFs for higher frequencies are lower than the theoretical CDF, for both the VIRC and the MSC. This could be caused by the direct ray path and corresponding Rayleigh-Rice distribution as described in the previous paragraph . A comparison of the VIRC and MSC CDFs, as shown in Figure 21 and 22, shows that the VIRC CDFs are approaching the theoretical CDF better than the MSC CDF. Therefore we can conclude that an estimation of the field strength in the VIRC is much more reliable than an estimation of the field strength in this MSC. We have to admit that the MSC used for the experiments is not optimized as reverberation chamber, and a properly operated MSC should result in a better CDF. 200 MHz 80% 1 GHz Theory, Chi(6dof) MSC 1, 200MHz MSC 1, 400MHz MSC 1, 600MHz MSC 1, 800MHz MSC 1, 1000MHz 60% 40% Theoretical CDF 20% 0% 0 0.5 1 Field Strength w.r.t. Mean Figure 22: CDF for a MSC 1.5 2 P. Corona, G. Latmiral, E. Paolini, L. Piccioli, Use of a reverberating enclosure for measurements of radiated power in the microwave range, IEEE Transactions on Electromagnetic Compatibility, vol.EMC-18, no.2, May 1976, p.54-9 [2] M.L. Crawford, G.H. Koepke, Operational Considerations of a Reverberation Chamber for EMC immunity Measurements - Some Experimental Results, IEEE Symposium on EMC, 1984 [3] M.L. Crawford, G.H. Koepke, Design, Evaluation and Use of a Reverberation Chamber for Performing Electromagnetic Susceptibility/Vulnerability Measurements, NBS Technical Note 1092, April 1986. [4] D.A. Hill, Electromagnetic theory of reverberation chambers, National Institute of Standards and Technology Technical note 1506, dec. 1998 [5] D.I. Wu, D.C. Chang, The Effect of an Electrically Large Stirrer in a Mode-Stirred Chamber, IEEE Transactions on Electromagnetic Compatibility, May 1989, pp. 164-169 [6] F.B.J. Leferink, W. van Etten, The Vibrating Intrinsic Reverberation Chamber, IEEE Transactions on Electromagnetic Compatibility, 2000 [7] Mélopée-ET2003, 3-axis Electric Field Sensor, Thomson-CSF [8] M. Petirsch, A. Schwab, Optimizing Shielded Chambers Utilizing Acoustic Analogies, IEEE Symposium on EMC, 1997, pp. 154-158. [9] J.G. Kostas, B. Boverie, Statistical model for a mode stirred chamber, IEEE Transactions on Electromagnetic Compatibility, vol. 33, no. 4, nov. 1991, pp. 366-370 [10] C.W. Helstrom, Probability and Stochastic Processes for Engineers, 2ed., MacMillan, 1984