URSI2002 Ch19
URSI2002 Ch19
URSI2002 Ch19
net/publication/230853336
CITATIONS READS
47 1,927
1 author:
Mats Bäckström
SAAB Group
89 PUBLICATIONS 1,502 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Mats Bäckström on 03 February 2018.
1. ABSTRACT
This paper provides a review of the theory and applications of reverberation chambers. The
focus is on EMC applications, but the determination of antenna efficiency is also considered.
Benefits and shortages are described. Basic features and modes of operation are elucidated, such
as the statistical properties of the chamber, the quality factor, and different means to create the
necessary samples in a test. Due to its crucial significance in evaluating the result of a test, the
relationship between anechoic and reverberation chamber testing is discussed. Also, the problems
of defining and operating at the lowest usable frequency are treated. Applications, in terms of
radiated susceptibility testing, measurement of shielding effectiveness and of antenna efficiency,
are dealt with. Finally, a brief look at unorthodox chambers and at new ideas is included.
2. INTRODUCTION
The reverberation chamber (RC), also called the mode-stirred chamber, is a highly
conductive, electrically large, shielded cavity, normally equipped with one or several stirrers
(tuners) to provide a statistically uniform and statistically isotropic field. The reverberation
chamber is used to conduct electromagnetic measurements on electronic equipment. The basic
setup consists of a transmitting antenna, used to excite the field; a receiving antenna, to calibrate
the field; and the equipment under test (EUT). The measurements comprise radiated susceptibility
(RS) testing, also denoted as immunity testing, as well as emission measurements and the
determination of shielding effectiveness. The lowest usable frequency is mainly determined by the
size of the chamber, the size and shape of the stirrer(s), and the quality factor, Q, of the cavity. The
reverberation chamber has, during the last decade, become a well-established tool for
electromagnetic compatibility (EMC) testing and measurements. Today, there are several EMC
standards that allow the use of reverberation chambers. Also, especially during recent years, a
rapidly growing interest has been shown for using the reverberation chamber to measure the
429
430 Mats Bäckström, Olof Lundén, and Per-Simon Kildal
radiation efficiency of electrically small antennas. A picture of the large reverberation chamber at
FOI – on that occasion, equipped with two mode-stirrers – is shown in figure 1.
Figure 1. The large reverberation chamber (2.5 × 3.0 × 5.1 m3), at FOI, Sweden.
• High test field levels for moderate input power, of the order of kV/m for 100 W input
• Short testing time, i.e., cost-effective testing
• No rotation of the equipment under test (EUT) is needed
• The location of the EUT is not critical
• Good repeatability between tests and chambers
• Good simulation of the electromagnetic environment for an EUT that operates in a resonant
cavity, such as a large avionics bay
• Absorbers are normally not needed
• Statistical properties are well known, thus the measurement uncertainties can be calculated
• Provides electromagnetic isolation from the external environment
• High lowest usable frequency (LUF), around a few hundred MHz for a 5 × 3 × 3 m3 chamber
• Problems using short pulses in testing, due to the chamber’s high quality factor
• Loss of information regarding polarization and directional properties of the EUT.
• Difficult to relate test results to free space conditions, when necessary.
The history of using reverberation chambers for EMC applications dates back more than 30
years. A driving force was the poor repeatability and measurement accuracy in shielded-room
testing, usually performed in the EMC community. The possibility of using an over-moded
shielded room to improve the repeatability in emission [Mendez, 1968] and shielding [Jarva,
1970] tests was identified, and evaluated with encouraging results. A standardized methodology,
using a reverberation chamber for measuring shielding effectiveness of cables and enclosures, was
presented in [MIL-STD 1377, 1971]. Some years later, Cummings [1975] reported results on
immunity testing in a reverberation chamber. The following year, Corona et al. [1976] presented a
19. Reverberation Chambers for EMC Susceptibility and Emission Analyses 431
method for measuring the total absorption of lossy materials and radiated emissions from
equipment. At this stage, comprehensive theoretical models of the reverberation chamber were
lacking. Such models, including the highly useful statistical descriptions of the field properties,
were developed in the 1980s and 1990s.
3. REVERBERATION-CHAMBER CHARACTERISTICS
2 2 2
c m n p
f mnp = + + , (1)
2 a b d
where c = 3 × 108 m/s is the speed of light (this assumes ε r = µr = 1 ), and m, n, and p are integers
(at least two shall differ from zero). For a ≤ b ≤ d , the lowest resonance frequency is given by
f 011 . As an example, the large chamber at FOI has the dimensions 2.5 × 3.0 × 5.1 m3, which gives
f 011 = 58 MHz. Note that if all m, n, and p ≠ 0 , there is a degeneracy between the TE mnp and
the TM mnp modes [Harrington, 1961]. If two or more sides of the chamber are equal, additional
degeneracy occurs.
3
8π f f 1
N= V − (a + b + d ) + , (2)
3 c c 2
where V is the volume. The first term is equal to the well-known asymptotic expression given by
Weyl, which is valid for arbitrary shapes of the chamber (see [Liu et al., 1983]).
The mode density is of particular interest for a reverberation chamber. With the exception of
frequencies close to the lower resonance frequencies, using Equation (2) the mode density is given
approximately by
dN 8π V 2
= 3 f . (3)
df c
8π Vf 3
Ns = . (4)
c 3Q
expectation values for maximum-to-mean values [Ladbury and Koepke, 1999]. In susceptibility
testing, the statistics for the maximum value are of obvious interest (see Freyer et al. [1998]).
In practice, the conditions referred to above do not seem to be very severe. The large
chamber at FOI, for example, has been shown to follow the assumed statistical function from –
depending on the stirrer geometry – around 0.5 GHz and up [Lundén and Bäckström, 2000]. At
0.5 GHz, the value of N s is around one; at 1 GHz, it is about four; and at 18 GHz, it is around
1500. Of course, even when N s is equal to one, each new independent stirrer position will create
a new field distribution, i.e., a new mode. It should be noted that the measured statistics for the
maximum field, i.e., the upper tail of the statistical distributions, show a less-than-good agreement
with the theoretical distributions (see, e.g., Wellander et al. [2001]).
Besides moving the stirrer (mechanical stirring), the necessary sampling can also be obtained
by probing the field at uncorrelated space locations (keeping the chamber geometry fixed). This is
utilized in platform stirring [Rosengren and Kildal, 2001] (see Section 4.4). Also, the same
statistical functions are valid if the frequency is stepped between uncorrelated frequencies
(keeping the probe position and the chamber geometry fixed). This is true provided the Q of the
chamber is a slowly varying function. This method is called frequency stirring, and is also called
frequency sweeping. As is pointed out in [Lehman, 1993], the use of this method for equipment
testing requires that the response of the equipment be a slowly varying function of frequency.
When applicable, frequency stirring could be of great value when performing examinations of (for
example) equipment bays where, for practical reasons, it is not possible to insert a stirrer. Note the
difference between frequency stirring and electronic mode stirring: the latter uses simultaneous
excitation of the frequencies within a certain bandwidth (cf. Section 3.4).
In order to create independent field conditions and, thereby, independent samples, the stirrer
has to create a new complex chamber geometry for each new position. This means that the
boundary conditions have to be changed to a sufficient extent. Wu and Chang [1989] claim that
the crucial factor for a stirrer to be efficient is the amount of frequency shift for an eigenmode the
stirrer can induce as a function of rotational angle. By experience, to accomplish this the stirrer has
to be electrically large, at least in one dimension. Another requirement is that the ratio between the
chamber volume and the volume swept by the stirrer shall not be too small ([Ladbury et al.,
1997]). In [IEC 61000-4-21, 2002], it is prescribed that in one dimension the stirrer shall be at
least three-quarters of the smallest chamber dimension. The benefit of changing the boundary
conditions as much as possible is illustrated by the fact that several stirrers provide a more
efficient stirring [Corona et al., 1998]. In using two stirrers, for example, these shall be moved
independently, or sequentially, that is each angular position of the first tuner shall be combined by
all positions of the other stirrer [Lundén and Bäckström, 2000]. Finally, in order to achieve as
many independent stirrer positions as possible, the tuner shall be asymmetrically shaped to avoid
repetitive field patterns over one tuner rotation (cf. Section 3.9.1).
sin k r1 − r2
ρ ( r1 , r2 ) = , (5)
k r1 − r2
f 2 − f1 f
ρ ( f1 , f 2 ) ≈ 1 − Q for f 2 − f1 ≤ 1 , and zero elsewhere. (6)
f1 Q
An alternative method that has been proposed for equipment testing in a reverberation
chamber is electronic mode stirring, sometimes also called frequency stirring [Loughry, 1991;
Hill, 1994]. The idea is related to the method of frequency stirring mentioned above. However, in
electronic mode stirring, the frequencies within a certain bandwidth are excited simultaneously.
Loughry [1991] accomplished this by up-converting bandlimited white Gaussian noise (BLWGN)
to microwave frequencies. By use of this method, a real-time, uniform, and isotropic field is
claimed to be achieved, using relatively narrow, 5 MHz, bandwidths. As mentioned in [Loughry,
1991; Hill, 1994], a potential problem in equipment testing, and in the characterization of
shielding properties, will arise if the equipment has coupling resonances with bandwidths that are
smaller than the agility bandwidth required to achieve a well-behaved electronic mode stirring.
This is presumably one reason why the interest in this method has decreased in recent years.
An important parameter in immunity testing is the time constant of the chamber, since it
determines the minimum pulse length that can be used in pulse-modulation testing. The time
constant, τ , is given by
Q
τ= . (7)
2π f
If the chamber time constant is too large compared to the pulse length, the chamber has to be
loaded to reduce the Q value (see, e.g., [IEC 61000-4-21, 2002]).
16π 2V Pr
Q= 3
, (8)
λ Pt
19. Reverberation Chambers for EMC Susceptibility and Emission Analyses 435
where Pt is the net power transmitted into the chamber, and Pr is the power measured by the
receiving antenna, compensated for antenna loss and free-space impedance mismatch. The
brackets denote the average over all stirrer positions. An alternative method to determine Pr Pt
from a spectral analysis of the field in the chamber is given in [Corona et al., 1998].
Finally, it should be noted that Q can also be determined from Equation (7) by a time-domain
measurement of τ . An advantage with such a measurement is that it is less affected by antenna
efficiency and impedance mismatch [Hill, 1998b].
The Q value can also be estimated from theory. By use of a local plane-wave method, Dunn
[1990] derived the following expression for Q due to losses in the highly conducting walls:
3 V 1
Q= , (9)
2 S µrδ s 1 + 3λ S
32 V
where S is the total surface area of the walls, V is the volume, δ s is the skin depth, and µr is the
relative magnetic permeability. A related expression has been derived using reflection coefficients
[Hill, 1998b]. Equation (9) has also been derived for a rectangular geometry using averages of Q
for modes close to the excitation frequency [Liu et al., 1983]. In calculating Q, it is usually
assumed that µr ≈ 1 , which is true for most construction materials at high frequencies [Loughry,
1991]. Equation (9) is derived assuming all losses are due to losses in the walls. In reality, there
are additional losses due to antenna loading, aperture leakage, and objects in the chamber.
Theoretical expressions for the different contributions to the total Q, validated by measurements
on an aluminum cavity, are given in [Hill et al., 1994]. Measurements on a shielded room showed
[Bäckström and Lundén, 1996] that losses due to aperture leakage were negligible, and that losses
due to the well-matched receiving and transmitting antennas showed up only at lower frequencies
(see also [Ladbury et al., 1997]). Measured Q values for shielded rooms often differ appreciably,
even at high frequencies, from those given by Equation (9). Measured Q values for the FOI
chambers are (at frequencies above a few GHz) around five times lower than what is expected
from Equation (9), using µr ≈ 1 and standard values for the conductivity [Bäckström and Lundén,
1996; 1999]. Similar results were also reported in [Crawford and Koepke; 1986] and by Lehman et
al. [1997]. The reason for the unexpectedly large difference is unclear, but it may partly be
explained as due to the presence of the stirrers, cables, and electrical fittings, and to wall joints and
the surface roughness of the walls (see also [Corona, 1980]).
One of the advantages of a reverberation chamber is the high field strengths that can be
obtained in radiated susceptibility testing for moderate levels of injected power.
In immunity testing, the test field strength is defined in terms of the maximum value of a
suitable field parameter, commonly the magnitude of the total electric field or of a rectangular
component thereof. Data for field coupling to probes mounted inside shielded enclosures and to
wires, measured in reverberation chambers, have been shown to follow a Chi-square distribution
with two degrees of freedom [Jansson and Bäckström, 1999; Silfverskiöld et al., 2002]. In other
words, they behave as ordinary antennas. Therefore, it seems appropriate ([Ladbury and Koepke,
436 Mats Bäckström, Olof Lundén, and Per-Simon Kildal
1999]) to define the test field in terms of the maximum of a rectangular component of the electric
field.
In Ladbury and Koepke [1999], expressions were given that yield expected values of
maximum field levels, given the value of Pr measured by the receiving antenna. As an example,
we consider the expected maximum of a rectangular component of the electric field, E R,Max, N ,
when N independent stirrer positions are used. Elaboration of Equation (10) in [Ladbury and
Koepke, 1999], in combination with Equation (8) and assuming a constant Pt , gives
Qλ Z 0 Pt Γ ( N ) N
E R,Max, N = ↑
24V Γ ( N + 1/ 2 ) N
( ER ) , (10)
The problem in relating the outcome of a susceptibility test, carried out in the complex
environment of a reverberation chamber, to free-space conditions was identified and addressed
early ([Bean and Hall [1978]). The statistical isotropy of a reverberation chamber means that an
antenna will receive signals from many directions, which means that the average received signal
will be independent of the directional properties of the antenna [Corona, 1980]. Thus, the
ensemble average of the receiving cross section, σ p , is, for all antennas and irrespective of
their directivity properties, given by ([Hill, 1998a,b; Warne and Lee, 2001])
λ2 λ2
σp = D e p q= eq , (11)
4π 8π
where D is the directivity (sometimes denoted the directive gain), e is the antenna efficiency due
to losses, p is the polarization-matching factor, and q is the impedance-mismatch factor. It holds
that D = 1 and p = 1 2 in an isotropic environment. The combined effect of e and q is
commonly referred to as the radiation efficiency [Kildal, 2000, Chapter 2.5.1]. Equation (11)
illustrates the loss of information regarding polarization and directional properties that was
mentioned in Section 2.
19. Reverberation Chambers for EMC Susceptibility and Emission Analyses 437
The fact that the variations in directivity and polarization are averaged out in a reverberation
chamber – illustrated by Equation (11) – indicates that an immunity test carried out in a
reverberation chamber might be less severe than a test carried out in an anechoic chamber (AC), at
least if the anechoic-chamber test comprises the worst angle of incidence and polarization. By
combining Equations (10) and (11) in [Ladbury and Koepke, 1999], and equating
E AC = E R,Max,RC , it can be shown that that the ratio between the maximum received power
in the anechoic chamber and in the reverberation chamber is
PAC ,Max 2
≈ Dmax , (12)
PRC ,Max 3
where denotes the expectation value. Thus, the expected difference between the maximum
stress on the EUT in the two chambers is roughly equal to the maximum directivity of the EUT.
Since an EUT can be regarded as a lossy antenna, it is assumed that Equation (12) can also be
applied to an EUT in a radiated susceptibility test, as long as terminals with linear loads can be
identified ([Hill, 1998a]).
Estimates of Dmax for typical shielded enclosures have been carried out by comparing the
measured coupling to internal field probes in the two kinds of chambers [Jansson and Bäckström,
1999]. In the measurements, the power density in the anechoic chamber was equated to the scalar
power density – defined in Equation (16), below – in the reverberation chamber. For such a case, it
can be shown [Ladbury and Koepke, 1999] that the ratio between the maximum received power in
the anechoic chamber ( D = Dmax and p = 1 ) and the average received power in the
reverberation chamber is equal to 2Dmax . The measurements [Jansson and Bäckström, 1999]
yielded estimates of Dmax of (typically) 10 dB, for frequencies of a few GHz and above. The
measured data fit quite well to theoretical predictions developed by Koepke et al. [2000] (see also
[Bäckström et al., 2001]). The maximum dimension of the investigated enclosures ranged from a
few decimeters to about one meter.
Equation (12) points out the risk for under-testing in a reverberation chamber, compared to a
test in an anechoic chamber. However, apart from in-band radiated-susceptibility testing of
antenna systems, it should be noted that the maximum stress in the anechoic chamber is, in
practice, never attained. The reason is that it would either require knowledge beforehand of the
worst angle of incidence and polarization, or that a huge number of aspect angles and polarizations
have to be used in the test [Bäckström et al., 2001]. In [Freyer and Bäckström, 2001], the outcome
of a typical anechoic-chamber test – using four aspect angles and two polarizations – was
compared to a reverberation-chamber test using 12 stirrer positions. The result was expressed in
terms of an error bias for each test chamber, defined as the ratio of a measured maximum response
to the true maximum plane-wave response. It was shown that if the field strength in the anechoic
chamber was equated to the maximum of a rectangular component in the reverberation chamber,
the expected error biases became similar in the two chambers. Thus, provided the statistics of the
real environment are adequately simulated by the test procedure in the anechoic chamber, the test
in the reverberation chamber is entirely appropriate.
It should be noted that the comparison above, between anechoic and reverberation chamber
testing, is based on the simplified assumption that the failure of a piece of equipment can be
438 Mats Bäckström, Olof Lundén, and Per-Simon Kildal
related to the stress on a single component. In reality, an equipment failure may require the
simultaneous interference of several components. More investigations, including radiated
susceptibility testing, are clearly needed to further evaluate the relationship between reverberation-
chamber and plane-wave testing.
Reverberation chambers are also used for emission measurements. The quantity measured is
the total radiated power. If the EUT radiates the constant power Pt ,EUT , this can be obtained by
use of Equation (8), provided Pr has been measured [Hill, 1998b]:
16π 2V
Pt , EUT = Pr . (13)
λ 3Q
average received power, Pr ,ref . By doing this, the coefficient on the right-hand side of
Equation (13) is determined, and Pt ,EUT is given by
Pt ,ref
Pt , EUT = Pr . (14)
Pr ,ref
The accuracy in an emission measurement depends on the number of samples in the test
[Kostas and Boverie, 1991]. As an example, 12 independent samples yield an uncertainty (95%
confidence level) of roughly ±2.8 dB, while 100 independent samples yields an uncertainty of
about ±0.9 dB, and 1000 samples yields an uncertainty of ±0.27 dB.
3.9 MISCELLANEOUS
The ability of a stirrer to give many independent test samples is usually taken as a criterion
for stirrer efficiency. To determine the maximum number of independent stirrer positions at a
given frequency, the correlation coefficient is calculated as a function of the stirrer interval (the
stirrer step angle) from measured Pr Pt data. As a rule of thumb, it is typically required that the
measured correlation coefficient shall not exceed e −1 ≈ 0.37 . An example of a correlation
measurement in the large chamber at FOI, using one stirrer [Lundén and Bäckström, 2000], is
shown in Figure 2. It is clear from the figure that, at the low-frequency end, the cross-shaped
stirrer yields, as expected, strongly correlated data close to 0°, but also around 90°, 180°, and
270°. The latter results are due to the fact that the intentional rotational asymmetry, given by the
folded shape of the vanes, is too small compared to the wavelength.
19. Reverberation Chambers for EMC Susceptibility and Emission Analyses 439
Figure 2. The measured correlation coefficient versus the frequency and stirrer interval (also see Plate 7).
The correlation test may alternatively be performed as a goodness-of-fit test. In that case, at a
given confidence level, the maximum-allowed value of the measured correlation coefficient will
depend on the number of samples [Lundén and Bäckström, 2000]. It should be pointed out that
uncorrelated data does not, in general, imply statistical independence: cf. the discussion in
[Arnaut, 2000; Wellander et al., 2001]. An exception is the case when the data are normally
distributed. Thus, although the use of the correlation coefficient to evaluate the stirrer efficiency
seems to work well, for the time being it represents an ad-hoc approach.
If many independent samples are needed in a test, it may be necessary to use two stirrers (cf.
Section 3.3). Other possibilities are to move around the EUT in the chamber, platform stirring, or
to use frequency stirring (cf. Sections 3.2 and 4.4).
The well-defined statistics for an ideally working reverberation chamber make it possible to
perform a goodness-of-fit test or a Kolmogorov-Smirnov test as a criterion for acceptance (see
[Corona et al., 1996]). An indication of how well the expected statistics are followed can be
achieved by comparing the standard deviation of the measured transmission-loss data to what is
expected from theory [Jansson and Bäckström, 1999; Ladbury and Goldsmith, 2000]. Often, the
cumulative distribution is simply compared to the expected one (see, e.g., [Freyer et al., 1996]).
Also, the moment theorem can be employed to confirm, or even to determine, the statistical
distribution [Corona et al., 1996]. In [Capetta et al., 1998], Lyapounov exponents (chaos theory)
were used to evaluate reverberation properties. Finally, Arnaut and West [1998] proposed an
440 Mats Bäckström, Olof Lundén, and Per-Simon Kildal
evaluation method based on coefficients for field anisotropy and inhomogeneity; see also Annex J
in [IEC 61000-4-2, 2002].
Reverberation chambers are often used below those frequencies for which the chamber
behaves as an ideal chamber. In that case, one has to abandon the type of methods described in the
previous section, and one must develop more ad-hoc criteria to determine the lowest-usable
frequency (LUF). As an example, if a certain (statistical) uniformity is stipulated, one has to
accept a larger number of stirrer positions at lower frequencies (cf. [IEC 61000-4-2, 2002]).
Besides uniformity, there are also other criteria, such as requiring that the stirring ratio – the ratio
between the maximum and minimum power received by the calibration antenna – shall be more
than 20 dB (see, e.g., Ladbury and Goldsmith [2000]). In applications based on the use of average
values, e.g., measurements of shielding effectiveness or antenna radiation efficiency, the relevant
criterion might be a certain measurement accuracy. A low uncertainty can also be achieved for
small chambers by a combination of different methods of stirring: see Section 4.4, below. The
current disparity between existing criteria was illustrated by Musso et al. [2001], who compared
seven different criteria, including those from four standards (of which three were drafts). The
chamber used in the study had a lowest resonance frequency of 30 MHz. Following the standards,
values of LUF ranged from 92 MHz to 550 MHz, hardly an acceptable situation.
3.9.4 Diffusers
In acoustics, the quality of rooms is sometimes improved by use of scattering elements: so-
called diffusers. The incident wave is scattered in all directions, resulting in a diffuse propagation
of the sound. In a reverberation chamber, the purpose of using diffusers is to locally increase the
mode density at low frequencies, thereby decreasing the lowest-usable frequency [Clegg et al.,
1996; Petirsch, 1999; Höijer et al., 2000]. In Clegg [1996], a measure was introduced to predict
the efficiency of a grating. A related way to decrease the LUF could, according to Dunn [1990], be
to use nonrectangular shapes of the chamber. So far, the decrease of the LUF by the use of
diffusers seems to be rather marginal.
4. APPLICATIONS
Today there are several EMC standards that allow the use of reverberation chambers
[Hatfield, 2000], for example EUROCAE ED-14D/RTCA DO160D, MIL-STD 461E, and IEC
61000-4-21 (draft). These standards comprise susceptibility testing as well as emission
measurements, and measurements of shielding effectiveness of equipment and components. Some
aspects of radiated susceptibility testing and coupling measurements are treated below. Also,
emission testing will be treated in terms of measurements of the radiation efficiency of terminal
antennas, an application that has recently gained considerable attention. A typical reverberation-
chamber facility is shown in Figure 3.
19. Reverberation Chambers for EMC Susceptibility and Emission Analyses 441
The basis of radiated susceptibility testing has already been treated in Sections 3.6 and 3.7.
Here, we give some comments of a more practical nature.
The field calibration is carried out using either electrically short field sensors probing the
field independently in three orthogonal directions, or by simply using a linearly polarized
receiving antenna, such as a horn antenna. Besides providing the calibration factor, i.e., the
relationship between the test field strength and the input power, the calibration procedure can also
comprise an evaluation of the field uniformity. In a test, it is necessary to control the loading of
the chamber due to the presence of the EUT, and, if necessary, to adjust the calibration factor
accordingly. Furthermore, it is vital to avoid direct coupling between the transmitting and
calibration antenna and, in an immunity test, between the transmitting antenna and the EUT.
In radiated susceptibility testing, a difference is commonly made between mode tuning and
mode stirring. In mode tuning, the stirrer is stepped. The dwell time at each stirrer position should
be sufficiently long for the EUT to respond properly. Mode stirring, on the other hand, means that
the stirrer is moved continuously. This method is faster, and may well be used, for example, in
measurements of shielding effectiveness, provided the response times of the field probes are
sufficiently short. MIL-STD 461E and RTCA DO160D (change notice one, November 2000)
allow only mode tuning, while IEC 61000-4-21 also allows mode stirring, provided the EUT’s
response time is taken into account. In some cases, it has been argued that mode stirring better
represents the conditions the EUT will meet in real life, for example, when avionics inside an
approaching aircraft are irradiated by an airport radar, or when a piece of equipment faces a
complex urban environment. A problem in using mode stirring is to determine the number of
samples used in the test, as well as in the chamber calibration.
chamber be shorter than the pulse length of the modulated waveform. To achieve this, the Q value
of the chamber may have to be decreased, which means that the test field strength is reduced for a
given level of input power. A criterion used in several standards is that the chamber’s time
constant should not be greater than 0.4 times the pulse length.
The reverberation chamber has proven to be a very suitable tool for the testing of equipment
located in electrically large equipment bays. The reason is that the field conditions in electrically
large bays are similar to those in a reverberation chamber, despite the fact that the Q is usually
rather low in an equipment bay [Johnson et al., 1998]. In many cases, however, equipment
designed to face a plane-wave threat is tested in a reverberation chamber. In such cases, it is of
vital importance to be able to relate the outcome of such a test to plane-wave excitation. This
subject was treated in Section 3.7.
S 2π V
SE = inc = , (15)
S sc σ a λQ
where Sinc is the power density of the incident field. S sc is the scalar power density, defined by
[Hill, 1998a,b]
8π
S sc = Pr , (16)
λ2
where Pr is the power received by the (matched and lossless) receiving antenna in the
reverberation chamber . The transmission cross section of the aperture, σ a , is defined by
Pt = σ a Sinc , (17)
where Pt is the power transmitted through the aperture. From Equation (17), it follows that the
dimension of σ a is given in square meters. In general, the transmission cross section will depend
on the polarization and angle of incidence of the plane wave exciting the aperture.
The brackets around SE in Equation (15) points to the fact that the scalar power density,
S sc , is an ensemble property. In this case, the average is to be taken over all spatial points within
the enclosure. Assuming the properties of the shielded enclosure to be similar to that of a
reverberation chamber, use can be made of the statistical theory, together with Equation (15), to
predict the maximum stress on cables and components located inside the enclosure. The nature of
the statistical distributions of such an aperture-driven over-moded cavity has been recently
discussed. Lehman [2001] claimed that for excitation by electrically small apertures, the square of
a rectangular field component is distributed according to a Chi-square distribution with two
degrees of freedom, while for electrically large apertures, it is distributed according to a Rice
distribution. Holland and St. John [2001], on the other hand, claimed that the distribution for
excitation by small apertures is a convolution between a Chi-square distribution with two degrees
of freedom and one with six degrees of freedom.
In some cases, shielded enclosures may be located inside an electrically large equipment bay
with electromagnetic field properties similar to that of a reverberation chamber, cf. Section 4.1. In
such a case, the enclosure’s aperture cross sections are preferably measured using nested
reverberation chambers: see [Bäckström and Lundén, 1996]. In that case, Sinc in Equation (15) is
the scalar power density of the isotropic field exciting the aperture. In other words, the measured
σ a corresponds, in that case, to an isotropic average due to plane waves arriving from all possible
angles of incidence and will all polarizations, from the half space irradiating the aperture.
Methods to measure the transmission cross section in anechoic and reverberation chambers
have been developed [Quine, 1994; Bäckström and Lundén, 1996; Martin et al., 1997]. In both
cases, the aperture is backed by a reverberation chamber. The power transmitted through the
aperture, Pt in Equation (17), is measured in the same way as is done in emission testing, cf.
Section 3.8.
444 Mats Bäckström, Olof Lundén, and Per-Simon Kildal
Figure 4. The transmission cross section. The left plot is for a circular hole, 20 mm diameter, with plane-
wave excitation at normal incidence. The solid line represents the measured data; the dashed line is the
theoretical result. The right plot is for a riveted seam. The solid line represents the measured result for plane-
wave excitation at normal incidence; the dashed line represents a measurement in a reverberation chamber.
Two examples of measured cross sections are shown in Figure 4. When measured in the
anechoic chamber, the aperture under test was backed by a small reverberation chamber, 1.19
× 1.05 × 0.87 m3 (see Martin et al. [1997]). For these measurements, frequency stirring was used.
This was implemented by the use of moving averages. The number of equidistant frequency points
was 7001 between 0.5 and 18 GHz, and the moving average was taken over 72 points. For the
20 mm circular hole (the left plot in Figure 4), a comparison with theory was made [Kristensson
and Waterman, 1982]. In the right plot in Figure 4, a comparison was made between plane-wave
excitation and isotropic excitation of an aircraft-like riveted seam. For the isotropic excitation, 20
independent stirrer positions were used in each reverberation chamber (see [Bäckström and
Lundén, 1996]). The 290 mm-long seam consisted of two aluminum plates, riveted together by a T
element. The rivet-to-rivet distance was 30 mm. There was a 0.1 mm distance between the T
element and the plates, essentially consisting of (non-conductive) paint. The plane-wave excitation
was for normal incidence, with the electric field orthogonal to the seam. The plot shows that
plane-wave excitation may exceed the isotropic σ a by more than 15 dB.
Beside the methods for measuring the transmission cross section, methods to numerically
calculate σ a have also been developed. Moreover, a special approach, denoted the semi-
empirical approach, has been developed to make use of measurements of transmission cross
section as input data in numerical models intended to calculate the shielding properties of
electrically large, complex structures [Martin and Bäckström, 1999]. By using this method,
apertures that cannot be resolved due to limitations in computer memory, or the detailed structure
or material properties of which are not fully known, can be taken into account.
a uniform multipath environment, i.e., that the plane waves come in from directions uniformly
distributed in space, provided the chamber dimensions are sufficiently large. In a uniform
multipath environment, the quality of a phone antenna is characterized by its radiation efficiency
(cf. Section 3.7). In [Rosengren et al., 2001], how the radiation efficiency of a small antenna can
be measured with very good accuracy – in a chamber that was 0.8 × 1.0× 1.0 m3 in size, in the
GSM 900 MHz band – was described. The measurements made use of platform stirring and
25 MHz frequency stirring (cf. Section 3.2). The best accuracy was obtained when the antenna
was located close to a cylinder filled with a liquid with the same electric characteristics as human
brain cells. The same group has also measured different phone antennas in different locations
close to a phantom head of the type used when measuring specific absorption rate (SAR). The
reverberation chamber with a phantom head and mobile phone is shown in Figure 5. By
comparing SAR values and radiation-efficiency values, it was found that for a given antenna type,
an increase in radiation efficiency corresponded to a reduction in SAR. It was also shown that it is
possible to extract the deterministic free-space reflection coefficient of an antenna from the
statistical reflection coefficient, measured in the reverberation chamber. The same chamber was
used to measure the total radiated power of mobile phones in both GSM 900 and GSM 1800
bands. The conclusion was that the reverberation chamber can be used to measure both radiation
efficiency and free-space impedance, and that it was a much more convenient alternative than an
anechoic chamber.
The range of variation between operating reverberation chambers is rather large, both
regarding size and the stirrer geometry. There are, however, some chambers, or proposed
chambers, that differ markedly from the “standard” chamber.
The comparatively high LUF of a reverberation chamber is sometimes pointed out as a severe
limitation in using reverberation chambers as a versatile tool for EMC testing. A solution in terms
of a combination of a TEM (transverse-electromagnetic) chamber and a reverberation chamber
was proposed by Crawford et al. [1990]. By using this, the frequency range can be extended down
to 10 kHz. This combined chamber has three regions of operation: a TEM region up to the first
multimode cutoff frequency, a transition region, and a reverberation region. A fundamental
problem seems to be to define the field conditions in the transition region. A related idea has been
put forward by Perini and Cohen [2000], in which wires are placed along the length of the cavity,
supporting a TEM wave. By changing the phase or amplitude of the excited wires, it is claimed
that field stirring is achieved.
In some cases, it might be difficult to bring the EUT to a test facility. To remove this
problem, Leferink et al. [2000] constructed a reverberation chamber for in-situ testing. The
chamber is made of vibrating flexible walls, by which the stirring is accomplished. The walls are
made of a flexible conducting material. The chamber is connected to a rigid structure by flexible
rubber strings. A limitation of the method is that mode tuning cannot be performed.
The electromagnetic environment inside a city may – for example, in the case of mobile
telephone communications – be regarded as a mixture of stochastic and deterministic components.
Such an environment might be simulated by introducing a deterministic signal, i.e., an unstirred
component, in the otherwise statistically isotropic environment of a reverberation chamber. A
theoretical model for this kind of test environment has been presented and experimentally
validated by Corona et al. [2000]. A similar idea is to use an elongated chamber to create a non-
isotropic environment [Rosengren et al., 2001]
During the last decade, the reverberation chamber has become a widely used tool in EMC
testing and measurement. It has also been found to be a valuable tool for the characterization of
the radiation efficiency of small antennas. Some of its greatest assets, in relation to traditional
EMC testing, are good repeatability and its well-known statistical properties, allowing the
uncertainty in a test to be easily calculated. Although rather thoroughly investigated in recent
years, a remaining issue is to better understand the relationship to plane-wave testing, and another
issue is to improve the evaluation of stirrer efficiency. Also, it would be beneficial to formulate,
and agree on, unambiguous criteria that characterize a properly working chamber, especially its
lowest usable frequency. The problem here is probably to find clear and distinct quantitative
criteria that are, at the same time, not too rigorous. This is because the reverberation chamber has
been found to be useful, although it does not perfectly follow the assumed theoretical statistical
distributions. In the future, we will probably see an increased use of numerical methods for the
design and analysis of reverberation chambers, even at several GHz.
19. Reverberation Chambers for EMC Susceptibility and Emission Analyses 447
6. ACKNOWLEDGEMENTS
The authors acknowledge the valuable help from G. J. Freyer, M. O. Hatfield, J. Ladbury, T.
H. Lehman, L. Pettersson, and M. Höijer. This work was financially supported by the Swedish
Armed Forces and the Swedish Defence Material Administration.
7. REFERENCES
L. R. Arnaut and P. D. West [1998], “Evaluation of the NPL Untuned Stadium Reverberation
Chamber Using Mechanical and Electronic Stirring Techniques,” NPL Technical Report CEM 11,
National Physics Laboratory, Teddington, Middlesex TW11 OLW, United Kingdom.
M. Bäckström, J. Lorén, G. Eriksson, and H-J Åsander [2001], “Microwave Coupling into a
Generic Object. Properties of Measured angular Receiving Pattern and its Significance for
Testing,” in Proceedings of the 2001 IEEE International Symposium on Electromagnetic
Compatibility, Montreal, Canada, pp. 1227-1232.
P. Corona, G Latmiral, E. Paolini, and E. Piccioli [1976]. “Use of a Reverberating Enclosure for
Measurements of Radiated Power in the Microwave Range,” IEEE Transactions on
Electromagnetic Compatibility, EMC-18, 2, pp. 54-59.
448 Mats Bäckström, Olof Lundén, and Per-Simon Kildal
P. Corona, G. Ferrara, and M. Migliaccio [1998], “A Spectral Approach for the Determination of
the Reverberating Chamber Quality Factor,” IEEE Transactions on Electromagnetic
Compatibility, EMC-40, 2, pp. 145-153.
D. A. Hill [1994], “Electronic Mode Stirring for Reverberation Chambers,” IEEE Transactions on
Electromagnetic Compatibility, EMC-36, 4, pp. 294-299.
D. A. Hill [1995], “Spatial Correlation Function for Fields in a Reverberation Chamber,” IEEE
Transactions on Electromagnetic Compatibility, EMC-37, 1, p. 138.
D. A. Hill [1998a], “Plane Wave Integral Representation for Fields in Reverberation Chambers,”
IEEE Transactions on Electromagnetic Compatibility, EMC-40, 3, pp. 209-216.
F. Hoeppe, P-N. Gineste, and B. Demoulin [2001], “Numerical Modelling for Mode-Stirred
Reverberation Chambers,” in Proceedings of EMC Zurich’01, Zurich, Switzerland, pp. 635-640.
R. Holland and R. H. St. John [2001], “Statistical EM Field Models in an Externally Illuminated,
Overmoded Cavity,” IEEE Transactions on Electromagnetic Compatibility, EMC-43, 1, pp. 56-
66.
IEC 61000-4-21, “Electromagnetic Compatibility (EMC) – Part 4-21: Testing and Measurement
Techniques – Reverberation Chambers,” Draft, March, 2002.
W. Jarva [1970]. “Shielding Tests for Cables and Small Enclosures in the 1- to 10-GHz Range,”
IEEE Transactions on Electromagnetic Compatibility, EMC-12, 1, pp. 12-24.
L. Jansson and M. Bäckström [1999], “Directivity of Equipment and Its Effect on Testing in
Mode-Stirred and Anechoic Chamber,” in Proceedings of 1999 IEEE International Symposium on
Electromagnetic Compatibility, Seattle, USA, pp. 17-22.
G. Koepke, D. Hill, and J. Ladbury [2000], “Directivity of the Test Device in EMC
Measurements,” in Proceedings of 2000 IEEE International Symposium on Electromagnetic
Compatibility, Washington DC, USA, pp. 535-539.
J. G. Kostas and B. Boverie [1991], “Statistical Model for a Mode-Stirred Chamber,” IEEE
Transactions on Electromagnetic Compatibility, EMC-33, 4, pp. 366-370.
G. Kristensson and P. C. Waterman [1982], “The T Matrix for Acoustic and Electromagnetic
Scattering by Circular Disks,” J. Acoust. Soc. Am., 72, 5, pp. 1612-1625.
F. Leferink, J-C. Boudenot, and W. van Etten [2000], “Experimental Results Obtained in the
Vibrating Intrinsic Reverberation Chamber,” in Proceedings of 2000 IEEE International
Symposium on Electromagnetic Compatibility, Washington DC, USA, pp. 639-644.
B. H. Liu, D. C. Chang, and M. T. Ma [1983], “Eigenmodes and the Composite Quality Factor of
a Reverberation Chamber,” NBS Technical Note 1066, National Institute of Standards and
Technology, Boulder, Colorado 80303-3328, USA.
MIL-STD 1377 [1971], “Effectiveness of Cable, Connector, and Weapon Enclosure Shielding and
Filters in Precluding Hazards of Electromagnetic Radiation to Ordnance; Measurement of,”
Department of Defense, USA.
A. K. Mitra and T.F. Trost [1997], “ Statistical Simulations and Measurements Inside a
Microwave Reverberation Chamber,” in Proceedings of the 1997 IEEE International Symposium
on Electromagnetic Compatibility, Austin, USA, pp. 48-53.
L. Musso, J. Bossé, V. Berat, and F.G. Canavero [2001], “Critical Study of Calibration
Techniques for a Reverberation Chamber,” in Proceedings of EMC Zurich ‘01, Zurich,
Switzerland, pp. 659-662.
J. Perini, L. S. Cohen [2000], “An alternative Way to Stir the Fields in a Mode-Stirred Chamber,”
in Proceedings of 2000 IEEE International Symposium on Electromagnetic Compatibility,
Washington DC, USA, pp. 633-637.
J. P. Quine [1994], “Distortion of Radiation Pattern for Leakage Power Transmitted Through
Attenuation Cover Panels and Shielding Gasket – Need for Reverberation Chamber Measurement
of Total Leakage Power,” in Proceedings of 1994 IEEE International Symposium on
Electromagnetic Compatibility, Chicago, USA, pp. 285-290.
452 Mats Bäckström, Olof Lundén, and Per-Simon Kildal
K. Rosengren, P-S. Kildal, C. Carlsson, and J. Carlsson [2001], “Characterization of Antennas for
Mobile and Wireless Terminals in Reverberation Chambers: Improved Accuracy by Platform
Stirring,” Microwave and Optical Technology Letters, 30, 20, pp. 391-397.
K. Rosengren and P-S. Kildal [2001], “Study of Distributions of Modes and Plane Waves in
Reverberation Chambers for Characterization of Antennas in Multipath Environment,” Microwave
and Optical Technology Letters, 30, 20, pp. 386-391.