This review of measurement methodologies for the shielding or screening effectiveness of microwave coaxial cables shows the strengths and weaknesses of various approaches.
Shielding effectiveness (SE) is a key parameter when considering microwave coaxial cable assemblies and connectors for applications with strict electromagnetic-compatibility (EMC) requirements. Measuring the shielding or screening (as the IEC prefers1) effectiveness of microwave coaxial cables is not trivial, however, and requires a full understanding of the terminology and techniques practiced essentially by a distinct subset of the microwave industry concerned with EMC.
EMC parameters include emissions, immunity (susceptibility), and crosstalk. Emission defines the impact of the cable on the environment. Immunity shows the ability of the cable to meet specified performance in a given EMI environment. Finally, crosstalk is EMC interaction between the cables themselves. Although coaxial cable is a passive linear device and according to the reciprocity theorem, emission and immunity tests should lead to the same results, this doesn't always occur. A typical coaxial cable is not always a linear and, therefore, reciprocal component.
EMC parameters can be measured by many methods designed to detect low-level signals, although test results are often nonrepeatable. EMC test responses can depend on the type of test system chosen.
The most common shielding parameter is surface transfer impedance, ZT, which was first introduced by Schelkunoff from Bell Labs in 1934. It can be defined as a ratio of induced series electromotive force inside the shield in the secondary (disturbing) circuit (V2) to the disturbing circuit current (I1) flowing in the shield in the primary (outside disturbing) circuit of an electrically short piece of cable.
Unlike most cable parameters that determine the signal propagation along the cable, ZT characterizes the energy propagation across the cable through the shields. The importance of using an electrically short piece of cable is critical. For cable, the surface transfer impedance has a unit of Ω/unit length (typically mΩ/unit length):
Equation 1 is defined for the cable only. For connectors and electrically short cable assemblies, the surface transfer impedance is defined for the whole length of the assembly.
The surface transfer impedance applies to the SE against current. According to Ampere's Law, current is related to magnetic field. Therefore, the surface transfer impedance relates to magnetic of galvanic coupling. But capacitive coupling is also possible (for example, through the holes in cable braids). To include this coupling, there is also a definition of equivalent transfer impedance, which includes effects of galvanic, magnetic, and capacitive coupling. Thus, in this case the effective transfer impedance (ZTE) can be defined as:
ZF = the capacitive transfer impedance.
It is important to note that ZT depends only on the screening properties of the shield and doesn't depend on the nature of the outer circuit (in this case, the testing cavity surrounding the leakage source). Because of this, the surface transfer impedance is established as a primary screening parameter in the literature. The capacitive transfer impedance depends on the outer circuit geometry and permittivity (as any capacitance does). In practice, capacitive coupling should not be a concern for high-performance microwave cables, only for single-braided cables similar to RG316. According to ref. 3, capacitive coupling can be normalized in a way that ZF will be invariant to the outer circuit under typical conditions. Figure 1 shows typical transfer impedance data for microwave cables. Because surface transfer impedance measurements require an electrically short device under test (DUT), they are not suitable for microwave transmission lines, but can be used for microwave connector measurements.
Screening/shielding attenuation is defined as the ratio of the maximum power in the secondary (outer) circuit to the power propagating to the primary (inner) circuit. Since shielding attenuation measurements don't have a requirement for electrically short DUTs, they can be used for electrically long objects such as microwave cables. Screening attenuation is actually defined for electrically long objects only and doesn't depend on the mechanical length. Shielding attenuation measurements depend not only on the screening properties of the DUT but also on the measurement system, such as the velocity and impedance differences between the test cavity and the DUT. According to ref. 7, shielding attenuation is a secondary screening parameter, with a great deal of uncertainty inherent in the interpretation of the test results.
There are two important additional functions: the transfer function and a summing function. The general coupling transfer function is defined as the square root of power measured related to the square root of power sent to the system:
The coupling transfer function for cable measurements is different for the near and far ends (Tn and Tf). Additionally, it is a complex number. Phase effects are expressed by the summing function S which has the form of (sin x)/x and it has some difference for the near and far end Sn and Sf (Fig. 2). For low frequencies, the summing function becomes equal to unity while for high frequencies, the envelope can be calculated as:
where the near and far end cutoff points are
The point of intersection between the asymptotic values is the so-called "cutoff frequency." This frequency gives the condition for electrically long samples. Parameters εr1 and εr2 are the relative dielectric permittivity of the inner and outer systems and l is the cable length.
There is some confusion in the definition for cutoff frequency. Traditionally, in transmission-line theory, the cutoff frequency is a frequency from which the excitation of the next higher-order mode of propagation is possible. For example, in a coaxial line this frequency is when the TE11 mode can theoretically be excited along with a principal mode TEM. In the case of EMC measurements, the cutoff frequency is a frequency point where the summing function crosses the axis, and the DUT can be assumed to be electrically long (which has nothing to do with the traditional definition for cutoff frequency).
As a result, the cutoff frequency has a single definition for two different effects. This can be confusing since the high-frequency triaxial test setup has some issues with regard to both effects.
MIL-C-17 is the main military specification for the US coaxial-cable industry but it doesn't include RF screening requirements. MIL-T-81490 and MIL-C-87104 apply to special high-power coaxial assemblies for airborne applications, and both specifications have similar RF shielding test specifications and test methods. The test methods are based on a special triaxial cavity with the shorted both outer circuit ends (thus the outer coaxial system forms a coaxial resonator).
The mode-stirred method covered in MIL-STD-1344 (method 3008) is a connector test. The coaxial connector shielding covered by MIL-PRF-39012 includes a triaxial cavity method. There are some international specifications developed by the International Electrotechnical Commission, Technical Committee 46, Working Group 5. IEC TC46 WG5 has been working for over 30 years and is going to release some standards regarding to the coaxial-cable screening effectiveness. There is a Special International Committee on Radio Interference (CISPR) that is related to the IEC TC77. The European Standards (EN) were created by the European Committee for Electrotechnical Standardization (CENELEC). The Society of Cable Telecommunication Engineers (SCTE) also published its own documents. At present, most cable standards have been developed for low-frequency coaxial cables. Typical applications include a transmission line for a wireless base station or for a CATV system. Therefore, most of the test methods can be used for frequencies to 2 GHz. Until now, international shielding standards have had little influence on the US cable industry, but that may change in the future, most likely for companies that are involved in export or with manufacturing facilities overseas.
Unlike MIL-C-17, international shielding standards include EMC test methods. The most common SE test methods in the International Standards include:
Line injection and absorbing clamp are not recommended at frequencies higher than 1 GHz. Although ref. 5 describes the line injection method extension to 20 GHz, this approach is not widely used at higher frequencies.
The mode-stirred method is widely used for microwave components because it has no real limits on the upper operating frequency. The mode-stirred (reverberation) chamber is a large (compared to wavelength) cavity with highly conductive walls. Boundary conditions are continuously and randomly perturbed by a rotating conductive tuner or "stirrer." The excited field is not directed from one source and a field density can be uniform if the number of modes is large enough. The number of modes should be in the thousands in order to have correct measurements. The volume of the chamber should be as large as possible.
A typical stirred-mode chamber has two antennas: the reference and source antennas. MIL-STD-1344 describes long-wire antennas. In some references these antennas are replaced by horns because they produce lower VSWR. The antenna reflection is one of the reasons for the measurement uncertainty. According to ref. 5, there are two main measurement modes: the mode-tuned and mode-stirred modes. In the first approach, at frequencies below a few GHz, the antenna impedance is a function of tuner position, and the stirrer should be moved in steps. After every step, a correction can be made for the difference in the transmitting antenna's input impedance as a function of the tuner position. In the second approach, for frequencies above a few GHz, the stirrer can be moved continuously.
Figure 3 shows the setup for the continuous sample averaging system per MIL-STD-1344. The mode-stirred method, which is convenient for DUTs with complicated shapes, can be used up to high frequencies, as it actually only has limitations on the low-frequency end. The shortcoming of the mode-stirred test is that the measurement setup is expensive and requires complicated signal processing, and according to ref. 6, by 1991 there were few facilities in Western Europe capable of these measurements.
As noted earlier, method 3008 of MIL-STD 1344 covers connector SE but is applicable to multicontact connectors to 10 GHz rather than coaxial connectors. Coaxial connectors are covered by MIL-PRF-39012 (the triaxial cavity). Right now, the mode-stirred method is not included in the IEC standards for coaxial cable screening. The mode—stirred method has references in MIL-STD-461 as an alternative method for the electric field radiated susceptibility test RS103, but this standard was developed mostly for system and subsystem levels.
Of the SE measurement methods, the triaxial cavity measurement is the simplest and least expensive approach. Leakage measurements using this approach are based on collecting the leakage energy in a coaxial system surrounding the leakage source. The DUT is in a uniform transmission line terminated on a matched load. Of the many configurations for triaxial cavities, the most common is the construction published in 1961 (ref. 2) by John Zorzy. In this configuration, the second coaxial system has an adjustable short-circuiting plunger on one end and the other has a tapered transition to a standard connector.
The triaxial cavity test procedure is included as a main reference in the RF coaxial connector specification MIL-PRF-39012. Additionally, the IEC method includes a triaxial fixture of significant length but the short is not adjustable (Fig. 5). Zorzy actually did use an adjustable short for the connector surface transfer impedance measurement only, but not for a cable (see ref. 9). For cable-assembly testing, the short was in the fixed position. With this method, screening attenuation measurements are valid for electrically long DUTs, which means that the frequency range has to be above the cutoff frequency (the nontraditional cutoff point).
Because of these considerations, it would make sense to construct a long triaxial cavity (the IEC recommendation is a length of 2 m), although longer constructions are no longer cost effective. This is likely one the reasons that in MIL-T-81490 and MIL-PRF-39012 the triaxial cavity length is defined as 1 ft. When considering connector measurements only, the connector cannot be considered electrically long below 2 to 3 GHz. For lower-frequency connector evaluation, it would make more sense to use the surface transfer impedance approach, or the connector's electrical length must be artificially increased.
The triaxial measurement method is not a straightforward as it may seem. In a triaxial fixture, measurement results depend on the outer line impedance and the velocity difference and, according to the IEC, a correction should be added.
The outer line impedance is always higher than 50 Ω because of mechanical considerations. According to ref. 3, the IEC default value is 150 Ω and the velocity difference between the inner and outer systems is 10 percent. According to the latest IEC draft, the resulting expression should include correction 10log10(2Zs/R), where Z is the outer line characteristic impedance and R is the DUT characteristic impedance (50 Ω in most cases). If the outer line impedance is not at 150 Ω, it must be normalized to a 150-Ω value. In ref. 2, the outer circuit impedance was also normalized but for a 50-Ω value.
The IEC (ref. 3) notes the correction caused by the velocity difference from default value is:
εr1 = the relative permittivity of DUT,
εr2,t = the relative permittivity of outer circuit, and
εr2,n = the relative permittivity of outer circuit, normalized
A velocity correction must be applied if the difference between velocities is more than 10 percent.
Another difference between the two approaches is the frequency limitation. The IEC design is limited up to a frequency of 3 to 4 GHz. The limitation factor is the outer line cutoff frequency (traditional cutoff frequency in this case). According to ref. 2, the triaxial line can be used to 7.5 GHz. In ref. 9, Zorzy noted that the triaxial line approach could even be used to 18 GHz with a special mode suppressor.
If the TE11 mode is generated, mutual coupling will occur in the same line of a triaxial cavity. Figures 4 and 5 show two triaxial cavities, using tapered or stepped configurations, respectively. The transition is needed for connection to standard test connectors, even though it acts like a filter and can cause part of the TE11 mode energy to be transformed to TEM energy. The transition reflects the waveguide mode because of the reduced radial dimensions, and can result in measurement errors. One way to suppress the mode is by having a tight concentricity tolerance between the conductors.
Most of the references pay attention only to the first waveguide mode that is TE11. Using the High Frequency Structure Simulator (HFSS) EM software from Ansoft Corp. (Pittsburgh, PA), the author performed a computer simulation on a simple triaxial line loaded with cable similar to RG142. The outer line of the model had an inner diameter of 0.915 in. that resulted in an impedance of about 108 Ω. About three waveguide modes were active through about 18 GHz. In addition, some modes, including TE11, were excited in two cross polarizations. The interaction between the dominant TEM and the waveguide modes is not of concern (except for the resonance frequencies). Moding can influence SE measurements under some conditions.
During the HFSS simulations, a line of larger diameter was simulated. The line has 127-Ω impedance with more than 10 waveguide modes active. Most simulations didn't show resonances below 7.5 GHz(Fig. 6), leading to the conclusion that for operation to 18 GHz, the outer line diameter should be as small as possible. The test data can also be improved by employing mode filters. In most cases, the first three waveguide modes have TE behavior. All of these modes have an axial magnetic field component that results in a circumferential current component on the internal surface of outer circuit. The mode filter design can be based on this effect.
While low-frequency screening test methods are well defined, there is much to discover with microwave applications particularly when testing microwave coaxial cables as electrically long objects. Better correlation is needed among different shielding/screening methods, and corrective and normalization functions must be much better defined.
The author would like to thank Astrolab colleagues Stephen Toma and Weirback, as well as Richard Pavadore from Engineering Special Service for their help with this article.