IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 37, NO. 3, AUGUST 1995

The Effects of Intrinsic Test Fixture Isolation on Material Shielding Effectiveness Measurements

Using Nested Mode-Stirred Chambers

Thomas A. Loughry and Shyam H. Gurbaxani

Abstmct-This paper evaluates the effects of test fixture isolation when using nested mode-stir chambers for conducting electromagnetic shielding measurements. The nested chamber technique is used by both government and industry to evaluate the electromagnetic attenuating properties of materials as vaned as infrared sensor windows to the composites used in the hulls of new ships. Two cases are considered: 1) The test aperture provides little inherent isolation between the two chambers, and 2) the aperture provides reasonably good isolation. It is shown that in the former case, data taken in this fashion can exhibit significant deviation from the latter. A correction factor for the former case is developed.

I. INTRODUCTION Over recent years, mode-stirred reverberation chambers have

gained some popularity for measuring the shielding properties of materials. Originally developed at the Naval Surface Warfare Center (NSWC) in Dahlgren, Virginia, the technique is based on two nested chambers [l]. The material under investigation is placed over a window in the smaller chamber and thus, the amount of isolation between the two chambers can be measured. This paper shows that results obtained when the two chambers are tightly coupled may need further correction and offers both theoretical and experimental methods for doing so. Readers not acquainted with reverberation chambers are encouraged to read 111 and [2].

11. NESTED CHAMBER THEORY

A. Nested Chamber General Description A typical nested chamber setup is shown in Fig. 1. A small

reverberation chamber is placed inside a larger one. Both chambers use paddle wheels to stir the modes within each cavity. The smaller chamber has a window fixture used to mount the material of interest over an aperture and a receive antenna to measure the fields within, while the larger chamber uses a transmittal antenna to generate the reference fields. The shielding properties of the material of interest are measured by first leaving the aperture uncovered and measuring the power received inside the smaller chamber. The window is then covered with the material of interest and the received power is again measured. The ratio of these two numbers is then reported as the material's shielding effectiveness.

B. Shielding Effectiveness Before the nested chamber technique can be evaluated, the con-

cept of shielding effectiveness must be quantified. In general, the shielding effectiveness of a material or configuration of materials (such as a screen imbedded in glass) is a measure of its ability to attenuate electromagnetic energy. This ability will depend upon both its reflection and absorption properties. Energy not being reflected

Manuscript received October 31, 1994; revised February 16, 1995. T. A. Loughry is with the Phillips Laboratory/WSM, 3550 Aberdeen

Avenue, Kirtland AI%, NM 87117-5776 USA. S. H. Gurbaxani is Professor Emeritus of the University of New Mexico,

Albuquerque, NM 87131 USA. IEEE Log Number 9413154.

da

I

Fig. 1. 'Qpical nested chamber configuration.

449

Fig. 2. Shielding effectiveness.

or absorbed by the material will be transmitted from one side to the other. For a given material, the amount of energy transmitted has a complex dependence upon the angle of incidence and polarization of the incoming electromagnetic wave. It is assumed that the interest is in how the material will attenuate an isotropically impinging wave front (hence, the use of a mode-stirred chamber). Fig. 2 depicts a material being used to attenuate an isotopically impinging wavefront originating from the left. A shielding factor, <, will be defined as the total power per unit area, S,,,, incident on the left side divided by the total power per unit area, P/A,., leaving the right side. The power incident per unit area, S,,,, on the left side is related to the isotropic power density S on the left side by

Hence, the ratio of power incident per unit area to power exiting per unit area is

Now, the shielding effectiveness measured by the nested chamber technique can be analyzed in terms of C and the parameters associated with the test fixture.

0018-9375/95$04.00 0 1995 IEEE

450 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 37. NO. 3, AUGUST 1995

CHAMBER 1

P

Fig. 3. Conservation of power.

C. Effects of Chamber Isolation The measured shielding effectiveness of a given window is defined

by [ 11 as the ratio of the power density in the small chamber without the window installed to the power density in the small chamber with the window installed. It will first be assumed that the two chambers are tightly coupled with the window aperture uncovered. This condi- tion will exist when the aperture is relatively large compared to the small chamber which, in practice, is assumed to have a high quality factor without the aperture. Thus,,the power density in the small chamber with the window open, Sz , is approximately equivalent to the power density in the large chamber, SI. Later, we will allow for the loosely coupled case where the window aperture provides some inherent isolation between the two chambers. It is also assumed that the difference in the loading of the large chamber by the small chamber, with and without the window installed, can be neglected. As seen in Fig. 3, when the system has reached steady state, power entering through side one of the window, PIZ, must equate to the sum of the power exiting the window from side two, PZ 1 ; the power absorbed by the window, PA; the power lost to the enclosure walls, PE; and the power exiting through the antenna port, PANT.

Pi2 = Pzi + PA + PE + PANT (3)

Power flow PIZ and PZI can be related to the isotropic power density inside and outside the enclosures respectively by (2). The power lost to the enclosure walls and the power exiting through the antenna port can be expressed as follows [3]:

x2 1 71c2 PANT = - s2 = - - s2 2 w2 87r

where w, c, A E , SE, and are the radian frequency at which the test is conducted, free space speed of light, inside surface area of chamber 2, the skin depth of chamber 2 walls, and the conductivity of the chamber 2 walls, respectively. Thus, (3) can be written as

where ( is the absorption loss factor of the window, PA/&. If after inserting the window to be measured, Sa decreases significantly, i.e., 10 dB or better, the second term in (6) will be much larger than

one. Also, if the window absorption loss factor is much less than the combined enclosure and antenna loss factors, (6) can be written as

s6 sz 2 w2 iw ( ; A E ~ E W + - T - "> . (7) lOlog - N 1 0 l o g ~ + 1 O l o g - ~

From (7) it is clear that the ratio of S; to SZ is dependent not only on the shielding properties of the window, but the properties of the smaller chamber (net fixture loss factor) and the size of the window sample, as well. The second term in (7) can be considered an offset term and, as will be seen later, can be calculated or indirectly measured. It is also notable that the effect of the smaller chamber losses on the measurement are not constant with frequency. Thus, when comparing shielding measured in this way, the resulting shielding measured may be significantly different than that which might be measured using a different smaller chamber with different losses.

Now consider the case when the aperture can not be considered large and some inherent aperture shielding, CW, results. The ratio of internal power density in the smaller chamber without the window installed, Si, to that with the window installed, Sz, can then be expressed as

If we again assume tha! the absorption loss factor can be neglected and that the ratio SI /Sz is large (implying the aperture itself provides a large degree of isolation between the two chambers), then (8) simplifies to

(9)

Thus, in this case where the aperture itself affords a reasonable degree of isolation between the two chambers, the resulting measurement as described in [l] will be reasonably close to the shielding parameter defined in (2).

III. EXPERIMENTAL MEASUREMENTS

The nested chamber technique was used to measure the shielding of a wire mesh from 4.0-8.0 GHz. The outer chamber measured 4.95 m x 6.15 m x 2.62 m and was constructed of zinc plated steel. The inner chamber measured 0.9 m x 0.9 m x 0.74 m and was constructed of aluminum. Both a large diameter (0.55 m) aperture and a small diameter (0.071 m) aperture wyre available on the inner chamber. The isolation (ratio of SI to S,) from 4.0-6.0 GHz between the two chambers was measured for both apertures while mode stirring (Fig. 4). An aluminum plate was used to cover one aperture while the other was being measured. As can be seen, the large diameter aperture provided very little, if any, isolation between the two chambers while the small diameter aperture afforded about 10 dB of isolation over the test frequency range. Hence, from the discussion above, (9) can be reasonably applied in the case of the small aperture but not in the case of the large aperture.

Two windows were made from similar wire mesh material to fit over each aperture. The wire mesh consisted of woven 0.3 mm diameter wire with 388 openings per cm'. live plexiglass plates were fashioned slightly larger than the apertures and were used to support the wire mesh windows. Each window's shielding was measured separately by covering the unused aperture with an aluminum plate. As can be seen in Fig. 5, although the same mesh was used in each case, the results are considerably different. As pointed out in the

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 37, NO. 3, AUGUST 1995 45 1

Fig. 4. Aperture isolation. Fig. 5. Measured shielding.

previous paragraph, the results for the small aperture window are reasonably close to the desired answer because of the applicability of (9).

Iv . CORRECTING NESTED CHAMBER DATA From the discussion above, if this test technique is to give

consistent results from one test facility to another using the same nested chamber technique but different size chambers, it is clear that the shielding offset created by the test configuration itself must be accounted for when the aperture is large and offers little chamber to chamber isolation. This could be accomplished in one of three ways. First, the shielding as measured using this method in dB could be corrected by subtracting from it the calculated theoretical offset in dB. Second, a window with a known shielding factor could be used as a reference provided both the reference window and the test window are relatively good shields and reduce SZ by 10 dB or more when installed in the test setup. The difference in the measured shielding of both windows using the nested chamber technique would indicate the true delta in shielding between the test window and the reference. Finally, the smaller chamber net loss factor (offset) could be measured directly and subtracted from the measured shielding. This would require covering the window aperture with a good conducting material and measuring the power density inside for a given input power or determining the Q through some other method.

A. Theoretical Correction If the large window under test is a relatively good measured

shielding, the theoretical method of correcting the data is straight- forward and simply involves calculating the offset term in (7) for each frequency, w, at which data was taken

This value is then subtracted from the experimental data. If the window under test offers a low value of shielding (less than 10 dB), then (6) should be used instead of (7) to determine the correct value. The theoretical method of correction is not recommended because in practice power losses in a real cavity are much greater than predicted by (3x5).

B. Reference Window If the unknown window has a fairly good measured shielding

factor, say 10 dB or more, then the last method of correcting the data can be used and involves using a reference window of known

high shielding (also measured at 10 dB or more), ('R. The shielding of the reference window can be determined by using a smaller sample on a smaller aperture so that (9) applies. Since CR is independent of window size (except for thickness), a larger window matching the size of the unknown window can then be fabricated with known shielding. The fields in the small chamber with the reference window installed, SR, and with the unknown window installed, Sc., can then be measured and the shielding for the unknown window, ('u, is determined as

(11)

This may be the easiest, most accurate way to deal with large aperture windows.

S lolog CU = lolog CR.

SR

C. Using Measured Loss Factor The loss factor of the smaller chamber can be determined experi-

mentally by measuring its Q. This is accomplished by replacing the window with a conducting plate. The receive antenna is then used to transmit energy into the cavity and a small probe is used to monitor the fields. The probe must be selected so as not to significantly load the chamber. B-dot and D-dot probes work well for this setup. The Q can then be determined using the methods described in [2] or [4]. The offset is then determined.

. - _ 4 wl

OFFSET = lolog - - A, cQ

where V is the volume of the inner chamber. This method can significantly improve the data if the window provides a reasonably good measured shielding so that (7) applies and the conducting plate and probe don't load the cavity significantly.

The Q of the inner chamber, discussed in Section III, was measured and used to correct the data for the large aperture wire mesh window. Fig. 6 shows the large aperture data after correction using (1 2) and compares it to the small aperture data. As can be seen, there is now much better correlation between the two window samples.

V. CONCLUSION The effects of intrinsic chamber isolation can be accounted for

when using the nested mode-stirred chamber technique for deter- mining the shielding effectiveness of a given material as defined above. For relatively large test samples requiring large test fixture apertures that create the case of the tightly coupled chambers, the data should be corrected as described above to reduce the effects of the test fixture. If the loading properties of the test fixture closely

452 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY, VOL. 37, NO. 3, AUGUST 1995

1 1 I 14

I 1

I O

I 1

1 1

I‘

I O 4 5 i o I 5 i I i 5 1 O 1 5 I O

I I ~ q V ~ I I , I G Y l ,

Fig. 6. Large window versus small window after correction.

resemble the properties of the system the window is being used in and the shielding of the material as installed in the system is desired, no correction may be needed. It should also be noted that there may be other effects that must be considered when performing nested chamber measurements. For example, bonding of the test sample to the fixture and edge effects associated with the fact that the sample is not an infinite plane.

REFERENCES

[ 11 M. 0. Hatfield, “Shielding effectiveness measurements using mode- stirred chambers: A comparison of two approaches,” IEEE Trans. Electromagn. Compat., vol. 30, no. 3, Aug. 1988.

[2] M. L. Crawford and G. H. Koepke, “Design, evaluation, and use of a reverberation chamber for performing electromagnetic susceptibil- ity/vulnerability measurements,” NBS Tech. Note 1092, Apr. 1986.

[3] T. A. Loughry, “Frequency stirring: An alternate approach to mechanical mode-stirring for the conduct of electromagnetic susceptibility testing,” Phillips Lab., Kirtland AFL3, NM, Rep. no. PL-TR-91-1036, Nov. 1991.

[4] R. E. Richardson, “Mode stirred chamber calibration factor and relax- ation time,” IEEE Trans. Instrum. Meas., vol. IM-34, no. 4, pp. 190-193, Dec. 1985.

Effects of AC Interference on Photovoltages of Junction Diodes

Cheng-Kuang Liu, Han-Chang Tsai, and Jine-Tine Liou

Abstract-The effects of ac interference on the photovoltage of diodes are analyzed experimentally and theoretically. The dependence of the interference susceptibility on the power and wavelength of incident light is shown. The frequency dependence of the interference effect is also illustrated. A model of the interference path is proposed and verified. Under the assumption of a capacitive coupling, an analytic expression is obtained for the prediction of the ac-interference photovoltage of a p-n diode. Computed results and experimental data are presented. Methods for avoiding this interference are discussed.

I. INTRODUCTION

Electromagnetic interference (EMI) has become a major problem for circuit designers. As circuitry becomes more complicated and faster, more emphasis is placed on the electromagnetic compatibility (EMC) problem. Among EMC problems, much attention has been focused upon the high-frequency interference [l], such as radio- frequency interference. A coupled interference may sometimes be neglected or smoothed out by an averaging process in practical applications. However, it may become noisome in the presence of a nonlinear device such as a diode or a transistor.

Microwave rectification effects on field-effect and bipolar transis- tors have been investigated [2]-[4]. The rectified (through device nonlinearity) high-frequency interference noise shifts the level of a low-frequency signal or the quiescent operating point of the victimized transistor. It is then expected that a low-frequency in- terference may result in a degradation or malfunction of an ultra-low frequency victim. One can apply high-frequency interference models to predict the interference strength. However, a low-frequency model is preferred in some cases owing to its simplicity.

It is evident that an optical signal in an optoelectronic system is free of EMI. However, E - 0 and 0-E transducers in the optoelectronic system may suffer from EMI. The interference is revealed by the current voltage nonlinearity in the transducer. Since the response of transducers is usually not large, a study of the strength of an electric interference noise is essential. An investigation on effects of low-frequency interference on an optically illuminated diode is thus motivated.

A noise problem is usually resolved by the method of experimental trial and error. Alternately, one may apply prediction or evaluation methods to save the expense and time in debugging. These methods are based upon the understanding of interference mechanisms [l] and theoretical models. A small signal model or a transmission line model has been developed [2]-[4] for nonlinear devices. In these models, the information of device nonlinearity, such as the transconductance of a transistor or the slope in the current voltage curve of a diode, is required to analyze the rectification effect in the presence of interference. An approach based upon the mechanism of carrier flow inside the diode is presented here, to analyze an optically illuminated diode under a low-frequency voltage interference. With a coupling

Manuscript received March 1, 1993; revised March 27, 1995. C.-K. Liu and H.-C. Tsai are with the Department of Electronic Engineering,

J.-T. Liou is with Winbond Electronics Corp., No. 2, R & D Road, VI,

IEEE Log Number 9413139.

National Taiwan Institute of Technology, Taipei, Taiwan 10772, R. 0. C.

Science-Based Industrial Park, Hsin-Chu, Taiwan, R. 0. C.

0018-9375/95$04.00 0 1995 IEEE