From EMC
This article will review common techniques and their basis for measuring emitted fields below 30 MHz. This region is often thought of as starting at about 9 kHz (the borderline frequency for many computing device requirements) but occasionally is extended down much further, as in the measurement of power line harmonics or in certain military standards. The further below 30 MHz we get, the more important an understanding of how inductive and radiated fields are linked becomes. There are three main areas which differentiate LF (low-frequency) measurement from the more familiar techniques employed for HF (here, taken as above 30 MHz): 1. Antennas: Most HF measurements are specified in terms of electric field strength, and require that the measurement be conducted with antennas that are sensitive to electric fields. In contrast, LF measurements are taken with antennas that are sensitive to magnetic fields and with antennas that are sensitive to electric fields, depending on the standard involved.
2. Near Field vs. Far Field: LF measurements are often taken in the near-field. The lower the frequency, the more likely this is to be the case. In the near-field, the details of the source-whether it is loop-like or dipolar-and the type of antenna it is measured with, have a large impact on the measured result.
3. Scaling: Many standards specify limits at a particular distance. Extrapolating measurements taken at a different distance to the specified distance can be tricky in the near field.
Equipment for LF Measurements Describing the changes in equipment that occur as you lower the frequency of measurement below 30 MHz is the easy part. Two types of antenna are commonly employed: magnetic loop antennas, which are sensitive to magnetic H-fields but insensitive to electric fields, and electric field antennas, which are sensitive to electric E-Fields but insensitive to magnetic fields.
Both types are available in active and passive models. Active models have solid state buffering built into the antenna to reduce antenna factor variation and interaction with the measuring equipment.
Loop antennas are available in a variety of sizes. Since loops sense magnetic fields by induction, the loop antenna's unloaded voltage output is proportional to the number of turns in the loop and the rate of change of magnetic flux across the loop's plane, which is proportional to the loop's area and the frequency of operation.
LF loop antennas used for EMI measurement are electrostatically shielded. The sensing loop winding is placed inside a conductive tubular shield that is grounded, but has a break at one point. Note that if this break were not present, the loop would be shielded against magnetic fields as well, because the outer shield would act as a grounded, shorted turn.
As stated above, the sensitivity of the loop antenna is proportional to the number of turns and the loop area. The antenna factor, which is added to the spectrum analyzer/receiver reading, moves in the opposite direction - more sensitive antennas have lower antenna factors. So, it would appear at first glance that more area and more turns would always be better in a loop antenna. Unfortunately, this is not the case, because the inductance of the antenna goes up in proportion to the area and to the square of the number of turns. This inductance appears as the dominant term in the antenna's source impedance, which increases with frequency. If this becomes significant in comparison with the load presented by the spectrum analyzer and cabling (typically 50 ohms), the sensitivity will drop. If the inductance is very large, self-resonances can also result in the antenna. To limit this effect, several measures can be taken:
1. Sometimes the frequency range of a large loop antenna will be restricted, and measurements at higher portions of the LF range will be taken with a smaller antenna (quite common)
2. Tuned matching circuits may be employed
3. An active high-impedance buffering amplifier can be used
E-field antennas are wire-like. At high frequencies, they are familiar as dipoles, biconicals, and log-periodics. At low frequencies there are balanced (short dipole) and unbalanced forms, such as whips. For some important standards (e.g., NEBS GR-1089, MIL-STD-462), which require E-field measurements, a 1-meter active whip with a planar counterpoise is specified.
At low frequencies, a whip is non-resonant because it is much shorter than a quarter-wavelength. Such antennas "look like" a voltage source with a small series capacitance-their source impedance is large, negative, and reactive. In contrast to loop antennas, whose impedance starts low and increases with frequency, electric field antennas start with very high impedances that decrease with rising frequency, so the use of active electronic buffering (usually via a battery-powered FET circuit) is common.
Figure 1: A typical shielded loop antenna for low frequency measurement. Although it measures H-fields, these values are often converted to equivalent electric field strength values under the questionable assumption of far-field conditions (see text).
A.H. Systems model SAS-562 Loop Antenna A.H. Systems' Loop Antenna delivers high performance for a wide range of magnetic field testing. Whether used in a set to measure shielding effectiveness or individually to satisfy specific requirements, the Loop Antenna is an efficient, low cost solution.
Figure 2: An active whip antenna for electric field measurements. A 1 meter whip length is typical.
A.H. Systems model SAS-551 Passive Monopole Antenna A.H. Systems' Passive Monopole Antenna provides superior performance in electric field measurements. The Passive Monopole Antenna is used for transmitting to perform shielding effectiveness and immunity testing.
Test Sites and Equipment Usage In the HF range, accurate measurements require careful characterization of the test site via site attenuation measurements and height scanning during testing. The motivation is to control the effects of constructive and destructive interference from multiple signal paths.
At low frequencies, where measurement is usually in the near field, neither site attenuation nor full height scanning (such as the typical 1 - 4 meter FCC/CISPR scan) play as large a role (note, however, that in the emerging field of Broadband over Power Line, BPL, one of the alternate methods of measurement in the FCC's Rule and Order does include a low frequency height scan) as it does in the HF range. At low frequencies there is no formal LF equivalent of site attenuation, although gross anomalies such as screen room resonances or large nearby conductive objects should be avoided.
The basic usage of LF antennas is in many ways similar to that employed at higher frequencies. The measured field is the logarithmic sum of the indicated signal from the analyzer/receiver, adjusted for cable loss, preamplification (if any), and antenna factor. The antenna factor denotes the voltage to field strength conversion characteristics of the antenna as a function of frequency.
Usually, the standard to which a piece of equipment is being tested will determine the placement and orientation of the measuring antenna. For example, it is common for a magnetic loop antenna to be mounted on a tripod at 1 or 2 meter height, and rotated for maximum signal. E-field monopole whip antennas are oriented vertically, and placed in a fixed position. In military testing of modules the antenna will have its counterpoise bonded to the test table plane; in GR-1089 testing, the counterpoise is bonded to the facility groundplane.
The use of the spectrum analyzer/receiver is not changed much at lower frequencies, although the resolution and detector function are typically changed. For FCC/CISPR measurements, the most common settings will be 9 kHz resolution and the use of the mid-band ("B") quasi-peak settings. To speed swept measurements, peak detection can be used, with quasi-peak confirmation for those signals which approach or exceed the limit.
The biggest area in which LF measurements vary from HF measurements is in the way distance is used to extrapolate measurements taken at one distance to estimate what they should be at a different (typically larger) distance. At higher frequencies, we don't think about this much-we often assume that field strength falls off linearly with distance. This works pretty well if we are "far enough" away-i. e., in the far field. To understand what is going on, we need to understand the way fields are generated from different sources, and how they behave in the near field.
Sources and the Near/Far Field Transition Two idealized sources are important for understanding the situation at low frequencies and the difference between the near and far field. Although simple, these models capture the essential behavior of actual equipment. Consider the sources modeled in Figures 3 and 4. The first is a small time variant charge dipole ("Hertzian dipole)," while the second is a time variant current loop (Figure 3).
Figure 3: A Hertzian dipole is a small, idealized electric field source. In the far field, E and H fields are radiated. Their strength is inversely proportional to distance. In the near field, "inductive" components are dominant, but fall off rapidly. From C. R. Paul [2].
Figure 4: A current loop (sometimes called a magnetic dipole) radiator reverses the near field behavior of the inductive magnetic and electric fields. Here, the magnetic field is stronger in the near field. From C. R. Paul [2].
Let's look at the dipole first. If there were charges, but no variation, we know from elementary physics that the E-field strength falls off with the cube of distance. Also, for a static charge dipole, there is no magnetic field. However, when current flows back and forth along the dipole (varying the charge concentration at the ends), the situation gets more interesting. The inverse cube behavior is still present for the electric field, but new terms appear inverse (1/r) and inverse square (1/r2) behavior. In addition, magnetic field terms emerge-these have inverse and inverse square behavior-but there are no inverse cube magnetic terms.
An analogous situation exists for a current loop, with a reversal of roles (Figure 4). Here, a steady DC current would generate a magnetic field with a 1/r3 dependence. If the current varies sinusoidally instead of staying constant, an inductive magnetic component and an inductive electric component which vary as 1/r2 are generated. Finally, just as in the case of the Hertzian dipole, there are electric and magnetic fields which fall off more slowly, as 1/r.
The 1/r electric and magnetic fields are coupled-they are the radiating electromagnetic field which propagates away from the source at the speed of light. The higher order terms are inductive-energy is stored in these fields, but is taken out again as the fields change. Now, if we look closely at the expression, each "r" term has the coefficient b = 2p / l. We can rewrite the terms as 1/br)n, where n is 1, 2, or 3 as follows:
So, the relative strength of the different terms depends on the ratio of the distance to the term (l/2p). When r = (l/2p), all components are the same strength. This is the dividing line between the near and the far field regions. At a distance greater than this, we are in the far-field, and the behavior is essentially inverse with distance. This is the radiated energy field, where the electric and magnetic fields are coupled. At distances closer than this, the higher order inductive terms predominate, and fields fall off more rapidly.
In the near field, how rapidly the measured E and H-Fields appear to fall off will depend both on the source type AND the antenna used to take the measurement. For example, if an electric-field sensitive antenna, such as a whip, is used in the near field to measure a source that is more or less dipolar, the measured level will decrease with the cube of the distance. However, if this same source is measured with a loop antenna, only the H-fields will be measured, and they will appear to fall off as the square of the distance.
As another example, take the measured behavior of a coil operating at approximately 100 kHz. Such a coil might be part of a security system, and RFID reader, or even part of a power supply. Measurements taken at multiple distances with a loop antenna, such as at 1, 2, and 4 meters would show that an inverse cube dependence for the magnetic field. However, if these measurements were taken with an electric field antenna (assuming the field were above the noise floor) an inverse square dependence would be noted.
The FCC and some EU standards (e.g., CISPR 11) specify limits below 30 MHz in terms of electric field strength, but mandate measurement with a loop antenna, which measures magnetic fields but is insensitive to electric fields. The antenna factors supplied for this purpose convert from magnetic to electric field strength by assuming a conversion ration of 51.5 dB, which corresponds to assuming an E/H relation of 377 ohms. While this E/H ratio obtains in the far field, it is not true in the near field. This difference is caused by the fact that the near field ratios increase by different powers of distance in the near field. This idea is captured by the ratio of E/H, which is termed the "wave impedance." (Figure 5). A dipolar source is "high-impedance" in the sense that in the near field, the inductive electric field is relatively stronger than the induced magnetic field. Conversely, a current loop source is "low impedance" in that the reverse is true.
Figure 5: The wave impedance of idealized electric and magnetic field sources as a function of distance from the source. The wave impedance is the ratio of electric to magnetic fields. In the far field, it becomes 377 ohms regardless of the details of the source structure. From J. P. Mills, [5].
Table 1 shows the relationship between the type of source, the type of antenna used for measurement, and the near-field dependence of the measurement. Now, a few words about the regulatory importance of scaling.
Table 1: The antenna "picks out" the type of field it is sensitive to. This determines the "measured" dependence on distance and the measured scaling factor.
The Importance of Scaling Some standards specify field strengths at distances that are too great for practical measurement. Sometimes the distance is such that it exceeds available test site space; on other occasions, the signal being measured is below the equipment noise floor at the specified distance; and sometimes both factors come into play. Consider the FCC's default requirements for field strength below 30 MHz (47 CFR § 15.209) shown in Table 2.
Table 2
If a measurement is made at less than the specified distance, it must be scaled, or extrapolated, for comparison with the limit. Fields fall off as a function of distance. Depending on the type of source (loop-like or dipolar), the type of antenna (magnetic or electric), and the electrical distance between the EUT and the antenna, the right extrapolation number might be 20, 40, or 60 dB per decade of distance. 20 dB/decade corresponds to the (1/r) fall off of field strength in the far field. 40 and 60 dB/decade correspond to 1/r2 and 1/r3 behavior.
Since the source type isn't always known or ideal, the FCC has adopted the following approach:
This approach is sensible, and provides a predictable rule for comparing measured data to limits. It isn't always perfect, as a couple of numerical examples may make clear. That is, there can be multiple interpretations that give substantially different results.
Example 1 (Fun with scaling factors): Imagine that a transmitter is operated at 29.9 MHz, and is measured to have a field strength of 3000 uV/m at a distance of 3 meters. Since it is operating below 30 MHz, the application of the default 40 dB/decade scaling factor would extrapolate this to a 30-meter field strength of 30 uV/m, which is just at the limit. Note, however, that if the same transmitter had operated at 30 MHz, the required scaling factor would have been 20 dB/decade, and the 30-meter extrapolated measurement would be 300 uV/m, ten times higher!
Which is correct? Physically, the second one is, as at 30 MHz (and of, course 29.99 MHz) the far field starts at about 1.6 meters. If the second option of experimentally determining the rate of signal attenuation with distance had been chosen, it would have been found to be 20 dB/decade. But, the rule as written seems to allow both interpretations.
Example 2 (Crossing the near/far field boundary): Consider a compact, loop-like transmitter operating at 13.56 MHz, such as a handheld RFID tag reader. At this frequency, the near/far field boundary is at approximately 3.5 meters. If measurements are taken in the near field, for example at 1 and 2 meters from the device, the roll-off will appear to be inverse cube (60 dB/decade.) If measurements are taken at 5 and 10 meters, the roll off will appear to be inverse, or 20 dB/decade. And, if the default rule is chosen, the distance correction factor will be 40 dB/decade.
Some Concluding Thoughts Near field measurements are straightforward. The additional equipment required is neither complex to use nor particularly expensive to obtain. More interesting is the physics behind the interpretation of the measurement. An understanding of the source type and how it interacts with the antenna clears the confusion that sometimes arises in interpreting the measurements made and comparing them to limits specified for different field types and at different distances. ¡ö
References 1. Electromagnetics, 4th ed., J. D. Kraus, McGraw-Hill, 1992
2. Introduction to Electromagnetic Compatibility, C. R. Paul, Wiley-Interscience, 1992
3. CISPR 16-1-4:2003, " Specification for radio disturbance and immunity measuring apparatus and methods - Part 1-4: Radio disturbance and immunity measuring apparatus - Ancillary equipment - Radiated disturbances," Sections 4.2 and 4.3
4. "Electromagnetic interference reductions in electronic systems," Jeffrey P. Mills, 1993 PTR Prentice-Hall.
Thank you to A.H. Systems for providing examples of antennas. For more information about their products, visit www.ahsystems.com. © Copyright 2007 Conformity |
