PRACTICAL PAPERS, ARTICLES AND APPLICATION NOTES
PRACTICAL PAPERS, ARTICLES
Here are two articles which should be of interest to EMCS members. In the first, on open field test site validation, W. Scott Bennett exposes errors in the ANSI C63.4 definition of an ideal site for a vertically polarized source. (Scott is the author of Control and Measurement of Unintentional Electromagnetic Radiation, which was reviewed in the Summer ‘97 Newsletter.) In the second, Daryl Gerke and William Kimmel of Kimmel Gerke Associates tackle the issue of grounding, with a useful “sewer” analogy.
Our intention in this section of the Newsletter is to provide EMC practitioners with an outlet for sharing knowledge or perspective gained from experience. Tips on problem finding, solving, avoiding or understanding are especially appropriate. Opinions and ideas are as welcome as irrefutable facts. Unlike archival works, such as IEEE Transactions or Symposium papers, Newsletter submissions will not undergo the scrutiny of multiple peer reviews prior to publication. They may be edited for length or content, but only with the concurrence of the author. Reader feedback is also welcome, either as a Letter (or e-mail) to the Editor or directly to the author.
To submit an article for consideration, send it via fax, e-mail or real mail to this Associate Editor. See page 3 for addresses and phone numbers.
With vertically polarized antennas it is often difficult to measure acceptable values of normalized site attenuation (NSA) on otherwise acceptable radiated EMI test sites. That difficulty is due in part to errors in the standards that must be met. Simple formulas for NSA on the ideal open-field test site are developed here, and standards errors are quantified.
Mathematical expressions for site attenuation are obtained by inverting a well-known expression for power transmission from one antenna to another [1]. Then, to obtain expressions for normalized site attenuation (NSA), both antennas are assumed to have antenna factors of 1 meter-1. That implies that the transmitting antenna radiates equally in all directions and the receiving antenna receives equally from all directions. In other words, normalization eliminates any directional variations in site attenuation other than those caused by the site itself.
If the “site” is free space, then, for the same distance from source to
receiving antenna NSA will always have the same value, regardless of direction. In free
space, NSA will vary only with wavelength, or frequency, and the distance from one antenna
to the other. In mathematical terms, 
or, in decibels, 
In these expressions
Rc is the characteristic impedance of the measuring system (typically 50 ohms),
l is the radiation wavelength (in meters),
Z0 is the characteristic impedance of free space (120p ohms), and
d is the distance separating the antennas (in meters).
NSA has no units, because the units of the antenna factors (meters-1) cancel those of l and d (meters).
In free space, radiated electric and magnetic fields, or E-fields and H-fields, are
both inversely proportional to the distance from their source to their observation point.
In other words, if the source is uniformly directional and d is that distance,
then the E-field and H-field magnitudes are both proportional to 
So, if ïEdï and ïHdï denote those
magnitudes a distance d from the source, then it is easily seen that 
Also, as a sinusoidal E-field and H-field propagate away from their source, they remain
in phase and, at any distance d, ïEdï = Z0ïHdï. Therefore, if the units of ïEdïand ïHdï are volts/meter and amperes/meter, the radiated power density
at any distance d will be ïEdïïHdï = Z0-1ïEdï2 = Z0ïHdï2 in
watts/meter2. And, as the fields propagate in any direction from d = 1
to any greater distance d, the power density will be attenuated by an
amount equal to 
. Thus, in free space NSA is the
product of 
and any one of those four equal quantities,
regardless of direction.
Free-space on actual test sites is limited, however, and radiated E-fields and H-fields
will be reflected by objects that limit the free space. As a result, on actual test sites
radiated E-fields and H-fields are sums of two, or more, components. One component travels
directly to the observation point, and the others arrive there indirectly,
because of reflection. Therefore, if ïEobsïand ïHobsïare the magnitudes of the total electric and magnetic
fields arriving at the observation point on a test site, the power density attenuation at
that point will be 
. That says that on any test site NSA
is the product of 
and any one of those three equal ratios.
On an open-field test site with a highly conductive ground plane, the ideal to be simulated is an empty half-space over a perfectly-conducting horizontal ground plane of infinite extent. The only reflecting body on the ideal test site is the ground plane. Therefore, the observed fields Eobs(t) and Hobs(t) will each be the sum of two components — one arriving directly and one arriving indirectly because of reflection by the ground plane.
Now, at any observation point the directions of the E-field and H-field of any radiated wave will always be perpendicular to each other and perpendicular to the direction in which they propagate. Also, a radiated wave on an open-field test site is said to be horizontally polarized if its E-field is horizontal and parallel to the ground plane. And, that wave is said to be vertically polarized if its E-field lies in the vertical plane that contains both the source and the observation point. Therefore, horizontal polarization means the E-field is horizontal, and vertical polarization means the H-field is horizontal.
An H-field that is reflected by an infinite perfectly-conducting ground plane to which it is parallel has a reflection coefficient of rH = +1. In other words, just above the ground plane the incident and reflected H-fields of a vertically polarized wave will be equal in magnitude and have the same direction. On the other hand, an E-field that is reflected by an infinite perfectly-conducting ground plane to which it is parallel has a reflection coefficient of rE = -1 _E = -1. That means that just above the ground plane the incident and reflected E-fields of a horizontally polarized wave will be equal in magnitude with opposite directions.
Figure 1. Open-field test site measurement geometry
Therefore, with a vertically polarized source on the ideal open-field test site, the
radiated H-field arriving at any observation point will have two collinear horizontal
components. However, whether those components have the same or opposite directions will
depend on the difference in the distances they have traveled. As shown in Fig. 1, the
distance traveled by the direct wave is 
, and the distance traveled by the
reflected wave is 
. In these expressions h is the measurement
distance, s is the source height, and a is the height of the
receiving antenna (the observation point). And, since a is always greater than 0, r
is always greater than d.
Suppose the H-field arriving directly at the observation point a distance d
from the source is 
c. Then, the reflected H-field
arriving at that point after traveling the distance r will be 
cos(w(t - (r - d)/c)) because of the
difference in propagation times (r-d)/c. (The divisor c = 3 ´ 108 meters/second is the velocity of light and
also that of the H-field and E-field.) And, because Hd(t) and Hr(t)
have either the same or opposite directions, the total H-field at the observation
point, Hobs(t), will be their sum.
So, with vertical polarization on the ideal open-field test site, and noting that 
,
it is seen that
And, from basic trigonometry it follows that the magnitude of Hobs(t) is
Thus, since rH = +1, the power density attenuation from source to receiving antenna will be
and multiplying that by yields NSA.
Therefore, with vertical polarization on the open-field test site
Now suppose the polarization is horizontal. Then, in the preceding equations H is replaced with E and, since rE = -1, it follows that the power density attenuation from source to receiving antenna will be
Therefore, with horizontal polarization on the ideal open-field test site
In other words, expressions for NSA differ for vertical and horizontal polarization only in the sign of the cosine term. And, that difference results because rH = +1 and rE = -1.
In measuring NSA, the value recorded for each source height and frequency is the minimum value obtained over a specified range of receiving antenna heights. In calculating those values for NSAdB with the above equation for horizontal polarization, there is complete agreement with the values given in ANSI C63.4 [2], for all frequencies and geometries. However, the values obtained with the above equation for NSAdB with vertical polarization do not agree with the values given in that standard. For a measurement distance of 10 meters, source heights of 1.0, 1.5 and 2.75 meters, and vertical polarization, the values for NSAdB in ANSI C63.4 are too high by the amounts shown in Fig. 2. For a measurement distance of 30 meters and vertical polarization, the values of NSAdB given in ANSI C63.4 all appear to be only 0.1 dB higher than they should be.
Figure 2. Positive error in values of NSAdB given in ANSI C63.4 for the 10-meter measurement distance with vertical polarization.
Therefore, to accurately determine how well a given 10-meter open-field test site simulates the ideal with horizontal polarization, measured values of NSAdB can be compared to the values given in ANSI C63.4. However, with vertical polarization, to make that determination with the same accuracy, measured values of NSAdB should be compared to values given in ANSI C63.4 after they have been reduced by the amounts given in Fig. 2. For additional measurement geometries, the required value of NSAdB can be accurately determined for either polarization using the equations given above.
[1] Friis, H. T., “A note on a simple transmission formula”, Proceedings of the IRE, vol. 34, pp.254-256, May 1946.
[2] ANSI C63.4-1992, “Methods of Measurement of Radio-Noise Emissions from Low-Voltage Electrical and Electronic Equipment in the Range of 9 kHz to 40 GHz.”
Scott Bennett is an
EMC consultant,
and can be reached at
7093 East County Road 74,
Carr, CO 80612
Tel: 970-897-2764.
Grounding is one of the most important, yet least understood, aspects of electrical engineering. As EMC engineers, most of us have seen grounding as a major contributor to EMC problems at both the systems and equipment levels. Entire books have been written on the subject, so this article will be but a brief attempt to highlight some key concepts on grounding. We’ll do this by asking several simple questions.
First, what is a ground? Ask an electronic design engineer, and you will get a different answer from a power engineer. Ask someone dealing with ESD, and you may get another answer. That’s part of the problem — the term is often vague, ambiguous, and context dependent.
The answer we like is that a ground is simply a “return path for current.” That path may be intended, or it may be unintended. Electrons don’t care, as they can’t read schematics anyway. Just remember that electrons “always return to their source”.
One answer we don’t particularly like for EMC issues is the classical “equipotential reference” definition. Anytime we have a finite current flowing in a finite impedance, we must have a finite voltage difference. (Yes, Ohm’s Law really works.) If the voltage difference is small, perhaps we can ignore it, and use the “equipotential” approximation. But often times the non-equipotential reality is what causes EMC problems.
We have a favorite analogy for grounds — the “sewer analogy”. Consider a ground as a path for “used electrons” if you like, and remember, it’s a path, not a cesspool. To expand the analogy, what is a desirable attribute of a sewer? Low impedance to flow. Is it acceptable to have more than one sewer? Sure, such as separate sanitary sewers and storm sewers (like separate “analog” and “digital” grounds.) Separation prevents unwanted mixing of sewage (or currents) that might otherwise share a common path.
Why use a ground? Grounds are used for many reason, including power, safety, lightning, EMI, and ESD. Sometimes one ground may perform several functions — intended or unintended. This is why grounding can be such a sticky issue.
Grounding requirements may vary widely due to vastly different amplitudes and frequencies. The latter is a key issue, since ground impedance is often highly frequency dependent. For example, round wire inductance become an issue above about 10 kHz, which necessitates lower inductance ground planes, grids, or bond straps at higher frequencies. Transmission line effects also become an issue when ground dimensions become significant. A common EMC rule of thumb for these effects is physical distances greater than 1/20 wavelength — about 250 km for 60 Hz, but only about 15 cm at 100 MHz. As a result, specific grounding techniques may vary widely. A general purpose ground may not solve a specific problem at all. Let’s look at several of these areas.
Power Ground — In power systems, this is often referred to as the “neutral”. This conductor carries current back to the sources, such as a transformer, service panel, or battery. This type of ground may carry large currents (many amps) but needs only to work at relatively low frequencies (50/60 Hz or DC). As a result, the constraints are resistance and current carrying capability.
Safety Ground — In power systems, this is the conduit or the “green wire”. Unlike the neutral, this conductor is only supposed to carry current during a fault condition, such as a short circuit to a metal cabinet. Like the neutral, it must carry large currents at low frequencies, but only for a relatively short time.
Lightning Ground — This type of ground provides a controlled connection
to the earth for lightning currents to follow. A lightning ground must carry huge currents
(upwards of 200,000 amps) but only for a fraction of a second. Due to the transient nature
of lightning, the frequency content is in the 300 kHz to 1 MHz range. The constraints for
lightning grounds are inductance and peak currents. This requires solid connections with a
minimum of bends.
EMI Ground — For “noise” control, an EMI ground must often work over a very
wide frequency range. At the same time, currents are often small (microamps or milliamps).
Thus, low inductance paths with low inductance/resistance connections are often needed.
ESD Ground — This is a special case of an EMI ground. A human body discharge can result in tens of amps, but only for nanoseconds. Due to the very fast rise times, ESD frequency content is in the 300 MHz range. Thus, low inductance grounds are mandatory for ESD. Sometimes we use “resistive” or “soft” grounds to limit the ESD current as well.
Single point, or multi-point? In simple terms, it depends on the threat frequency. For low frequency threats, such as stray 60 Hz power currents, the single-point ground is preferred. This prevents “ground loops”, which can let unwanted noise currents mix with intended signal currents in a common path. The strategy is to steer currents with careful wiring practices.
For high frequency threats (RF, or even power line transients), the multi-point
ground is preferred. Above about 10 kHz, parasitic inductance and parasitic
capacitance allow noise currents to follow alternate paths, so the single point ground
becomes difficult, if not impossible, to implement. In addition, transmission line effects
become an issue, also mandating multiple connections. Furthermore, to lower the
inductance, planes and grids are needed.
The single-point/multi-point guidelines extend to cable shield grounding as well. For low
frequencies, the preferred approach is to ground cable shields at one end, while for high
frequencies, the preferred approach is to ground cable shields at both ends.
For systems facing a combination of high frequency and low frequency threats, hybrid grounds may be needed. Capacitors can be used to provide multiple “high frequency” connections, and inductors can be used to isolate “high frequency” paths. Once again, these techniques apply to cable shield grounding. For example, we’ve solved numerous transient (high frequency) problems in instrumentation systems that traditionally use single-point (low frequency) techniques, by simply adding a small high frequency capacitor to the open end of a cable shield.
What about “earth” grounds? An earth connection is really only needed for lightning, but by convention it is widely used for power line safety as well. But you don’t need an “earth” ground for many situations. For example, an airframe can be used as a ground (return path for current flow), but you don’t see many 747s flying across the country dragging an earth ground cable with them. A common misunderstanding we often see is the urge to “drive more ground rods” to lower ground impedance. While this may help if your primary concern is lightning, usually it indicates a desperate approach that fails to address the real ground problems in the system.
One final comment. Safety issues must always take precedence over any other grounding needs. Before making any grounding changes, always be sure the system remains safe.
Daryl Gerke, PE, and William Kimmel, PE, may be contacted at Kimmel Gerke Associates, Ltd. at 1-888-EMI-GURU (Toll Free), or at their web site at https://www.emiguru.com.