Author: goldspireblog

One question that I am frequently asked by my clients early on in the product
development process is this — “What do you think is the greatest risk on this
project?” My answer is always the same — “The greatest risk to the overall
success of your product development effort is EMC compliance.” Depending
upon the sophistication of my client, they may or may not have an understanding
of EMC compliance. What oftentimes catches many of my clients by
surprise is the realization that they will be unable to sell their product without
passing a series of EMC compliance certification tests. Breaking this news to
clients usually sets them back on their heels, and leads to an introductory
discussion of FCC Part 15, CISPR 11 and CISPR 22, radiated and conducted
emissions, and other things that cause one’s eyes to glaze over just before
they roll back into one’s head. “All of these things are entirely manageable,” I
tell my clients, “However it is important that we deal with these issues at the beginning, rather than at the end, of the design process. Doing this will add
several thousand dollars to the total development cost as well as adding some
development time, however investing that money now will save you many
thousands of dollars later. Trust me.”

I have spent many, many hours inside EMC anechoic test chambers with products for which I have developed the electronics, as well as others where I did not develop the electronics (I don’t need to tell you which ones always performed better at the outset……), and in most cases the radiated emissions limits applicable to the product are 47 CFR Part 15.109, 47 CFRPart 15.209 (for an intentional radiator) and/or CISPR 11/CISPR 22 which are generally accepted radiated emissions standards in the EU (where they are more commonly known as EN55011 and EN55022, respectively). As most consumer, industrial, and medical devices that are produced will ultimately sell into European markets, radiated emissions testing nearly always includes evaluating the device to CISPR 11 standards. In this installment of DesignGold I’d like to talk about my experiences with the CISPR 11 radiated emissions standard and why it is so difficult to pass. I will conclude with some design guidelines that will (hopefully!) be useful to engineers and PCB designers out there regarding how to get your devices through the CISPR 11 Class B limit the first time you try.

“CISPR” is an acronym for “International Special Committee on Radio Interference”, an organization that was established in 1934 for adopting standards for the performance of radio systems. Looking specifically at the CISPR 11 standard, it is a document that is concerned with the electromagnetic emissions from Industrial, Scientific, and Medical equipment, more commonly referred to as “ISM” equipment. Within the CISPR 11 standard there are two Groups broken out as follows:

1. Group 1 — Unintentional radiators: devices that do not intentionally radiate electromagnetic energy as part of their operation
2. Group 2 — Intentional radiators: devices that intentionally radiate electromagnetic energy as part of their operation

Within these two Groups there are defined two Classes:

1. Class A — Equipment suitable for use in all locations except residential settings and establishments directly connected to a public low-voltage power supply network for domestic use
2. Class B — Equipment suitable for use in residential settings and establishments directly connected to a public low-voltage power supply network for domestic use

In this article I am going to focus on Group 1 devices — unintentional radiators, as they are far more common than Group 2 devices which must meet a host of additional requirements regarding their emissions.

Looking at the definition of the two Classes, Class A defines equipment that is designed primarily for use in commercial and industrial environments (think factory floors, hospitals, commercial buildings, etc.), while Class B defines equipment that is primarily used in the home. The majority of consumer products that exist today and with which everyone is familiar generally fall into this category — personal computers, printers, TVs and radio receivers, bread machines, etc. It is equipment that comes under the purview of Class B — common, everyday “appliance type stuff” that all too often gets bogged down in the radiated emissions testing process, simply because the CISPR 11 Class B limit is a bitch to meet.

Per the CISPR 11 standard, radiated emissions measurements are performed across a frequency range of 30MHz to 1GHz utilizing a receiving antenna positioned 10 meters away from the device. At that distance, the standard dictates that the measured electric field cannot exceed a specified value. This maximum electric field is illustrated by the graph below:

The horizontal axis of the graph is frequency while the vertical axis is electric field strength expressed dBuV/m (dB relative to 1 microvolt per meter). The Class B limit is denoted by the lower red line in the graph labeled “B”. Notice the discontinuity that exists at 230MHz — the radiated emissions limit increases by 7dB above this frequency. You’ll often hear EMC engineers refer to the CISPR 11 Class B limit as the “thirty thirty-seven rule” since those are the two radiation limits across the measurement band.

It has been my experience that the 30dBuV/m limit causes real difficulties for many products at frequencies below 230MHz, necessitating the last-minute inclusion of ferrite sleeves around power cables, or worse, the need to completely redesign printed circuit boards with ferrite beads, common-mode chokes, or even redesigning the board layer stackup. When stuff like this happens just before a long-awaited product release the ripple effect through the organization can get bloody very quickly. Let’s take a look at the radiation mechanisms from printed circuit boards, and then I’ll generate some numbers that illustrate why the Class B limit is difficult to meet.

Any time-varying electric current that flows has associated with it electric and magnetic fields that exist in the space surrounding the current flow. The structure and behavior of these fields are governed by Maxwell’s Equations, a set of four equations that describe how electric charges and currents interact with each other, resulting in the radiation of electromagnetic energy. Using Maxwell’s Equations it is easy (I use that term somewhat loosely) to show that a time-varying current radiates electromagnetic energy into space. We can apply this principle to a trace on a printed circuit board — we can imagine breaking the trace up into very tiny segments and then considering each of those tiny segments as a “current element” of some very short length which we can denote “dL”. Consider a single current element like in the drawing below:

This tiny segment of a PCB trace is sitting along the z-axis in the drawing, and it carries a current which varies sinusoidally with time. Maxwell’s Equations tell us that such a current element will give rise to an electromagnetic field which is relatively complex in structure, the majority of which we can ignore for the purposes of our discussion. The only part of the electromagnetic field that we want to consider is that portion of the electric field known as the radiation field, that is, the portion of the electromagnetic field that actually escapes into space and propagates away from the current element, never to return. The expression for this part of the electric field is:

\begin{aligned} E_{\theta} &= j\frac{30kI\,dL}{R}\, sin\,\theta \, e^{j(\omega t - kR)} \\ \text{where} \quad k &= \frac{2\pi}{\lambda} \\ \end{aligned}

You can see that the radiated electric field has a pattern associated with it — the sin \,\theta means that the electric field is zero along the axis of the current element, making the pattern look like a doughnut. The field strength is a maximum in the direction perpendicular to the axis.

If we now consider a case where we have two PCB traces very close together, and they each are carrying the same time-varying current, but in opposite directions:

Each current element will generate an electric field of exactly the form we have shown in the above equation. However, since the currents in the two current elements are flowing in opposite directions, the fields produced by the current elements will be 180 degrees out of phase with each other with respect to time. If these two current elements are spaced very close together compared to the wavelength of the electromagnetic field being generated by the currents, the two electric fields will very nearly cancel each other out — the end result being no radiated field!

This principle is employed everywhere in electronics — it is the main reason that ethernet cables utilize twisted-pair conductors to carry signals, and differential pair PCB traces are always spaced close together and are routed parallel to each other. The currents in these conductors are of equal magnitude but traveling in opposite directions, and by forcing the currents to flow in extremely close proximity to each other their electric fields cancel, eliminating any radiated energy.

It is this exact reason that the layer stackup in a printed circuit board is so critical. As we have stated, in order to keep time-varying electric currents from radiating energy, we need to make sure that wherever any time-varying current is flowing there is a current equal in magnitude, but opposite direction, flowing in very close proximity — we accomplish this in PCB design by using ground planes, sometimes called image planes, interleaved into the board stackup so that every layer in the PCB that has current-carrying traces also has a ground plane immediate adjacent to it. Why does this work? Look at the diagram below:

U1 and U2 are two devices that are mounted on the PCB with a trace that connects them. Current flows along the trace from U1 to U2, which then returns to U1 through vias that connect the top layer of the PCB to a ground plane underneath. The current flow in the ground plane from U2 to U1 is very closely coupled to the current flowing in the top trace from U1 to U2. This close coupling is what is needed to minimize the possibility of the current radiating electromagnetic energy off into space. For those of you paying very close attention, you might ask, “Okay, I see that, but what about the current flow in the two vias? Those currents aren’t closely coupled!” Yes, you are correct — however the vias themselves are extremely short, and the frequencies at which radiation from vias generally becomes a problem is considerable higher than 1GHz.

Closely coupled currents that are equal in magnitude but flow in opposite directions is sometimes called “differential mode current.” A purely differential mode current does not radiate electromagnetic energy for all the reasons we have described. Let’s take a look at an example using a PCB where the differential mode characteristic of closely coupled currents fails. Consider the drawing below, showing a top view of a PCB:

As in the previous drawing, there are two devices U1 and U2, with current flow between them carried by a trace on the top layer. In this case however, there is a cut in the ground plane on the bottom side of the board in order to accommodate a separate ground “island”. The gap around the “island” prevents the return current flowing from U2 back to U1 from flowing directly underneath the trace on the top side of the board — the return current is split into two separate paths that must flow around the edges of the gap. The outgoing and the return currents are no longer closely coupled — it almost looks like there are three completely separate currents flowing on the PCB. With no closely coupled currents flowing in the opposite direction, these three currents have effectively “gone rogue”, and are free to radiate electromagnetic energy to their heart’s content!

Breaking the return path like this for currents on a PCB is asking for trouble once it gets into the EMC chamber. I’d like to go back to the equations I developed above and use them to illustrate how even just a tiny bit of “rogue current” (these “rogue” currents are usually known as “common-mode” current) will result in exceeding the CISPR 11 Class B limit. Recalling the electric field of a short current element of length “dL” carrying a time varying current of magnitude “I”:

\begin{aligned} E_{\theta} &= j\frac{30kI\,dL}{R}\, sin\,\theta \, e^{j(\omega t - kR)} \\ \text{where} \quad k &= \frac{2\pi}{\lambda} \\ \end{aligned}

There is a lot more information in this equation than we need to consider for our analysis. Since we are looking at worst-case conditions here, we utilize the fact that the electric field is at a maximum at right angles to the axis of the current element, so we can just set the value of sin\, \theta to unity. We also need only concern ourselves with the magnitude of the field and completely ignore the phase, which allows us to just replace the complex exponential factor with unity. This leaves us with an expression for the maximum electric field radiated from a current element as a function of distance:

|E| = \frac{30kI\,dL}{R}

Replacing the wavenumber k with its equivalent expression in terms of frequency, we can write the equation like this:

|E| = \frac{60\pi f}{c_0 R} (I\,dL)

where c_0 is the speed of light. Notice that I have separated the current and the length of the current element and placed them inside parentheses. This quantity — the product of a current and a length is known as a dipole moment, so we can associate the radiated electric field from this short current element with the size of its dipole moment. As one might expect, the dimensions of dipole moment in this case is “ampere-meters”, abbreviated as A-m. The question we seek to answer is this: “What dipole moment will give rise to a radiated electric field at a distance of ten meters that is just equal to the CISPR 11 Class B limit?“

There is a bunch of messy algebra to rearrange the last equation we wrote in terms of the answer to the question, so I’ll spare you from having to wade through it and just give you the result (I encourage anyone who wants to go through the math to do so — if you find I’ve made an error, please let me know!). The expression below is the “money shot”:

(I_{\mu A}L_{mm}) = 10 ^ {\left\{\frac{E_{dB\mu V/m}}{20} - log\,f + 10.202\right\}}

This expression gives us the answer in all the right units to make things easy. The units of dipole moment are given as (\mu A-mm) and the electric field is specified in dB_{\mu V/m} . The graph below illustrates the behavior of the dipole moment as a function of frequency:

The blue trace in the graph is the dipole moment expressed in (\mu A-mm) while the orange trace is the dipole moment expressed in (\mu A-in) . Let’s use the orange trace to generate some quick observations:

1. At 30MHz, a one-inch trace carrying a current of 660.4 microamps will induce an electric field at 10 meters that reaches the CISPR 11 Class B limit
2. At 220MHz (just below the 230MHz jump to 37dB) that same one inch trace needs to carry only 90 microamps to reach the CISPR 11 Class B limit
3. At 1GHz that one inch trace will reach the CISPR 11 Class B limit at a current of 44.4 microamps

Remember — its the dipole moment that matters! If the trace is longer than one inch, the maximum current it can carry before the CISPR 11 Class B limit is reached decreases — if you double the trace length, the maximum current is halved.

At this point, I have (hopefully…) shown you why the CISPR 11 Class B Radiated Emissions limit is a real bitch — the amount of “rogue” current running around on a PCB or a power cable in the 30MHz to 1GHz frequency band is measured in MICROAMPS, in other words, it doesn’t take much current that isn’t closely coupled to its return current to bust the Class B limit!

So — what to do? In the design of any product, it is imperative that an approach of “design for EMC” be adopted at the outset when it comes to PCB design, power supply strategy, cable design, and especially enclosure design. All too often I have seen mechanical engineering teams plow along inside their own silo, designing an enclosure for a product with an attitude of “We’ll just stuff the PCB in there and bolt it down with four screws” while paying absolutely no attention to the reality that the enclosure forms a fundamental component in the EMC performance of the finished product. Make no mistake — “design for EMC” involves BOTH electrical and mechanical engineering.

As far as the electrical design goes, the following list of items has been extremely useful in my work, and I have saved clients a lot of money and time by adopting them in my design strategies:

1. Take PCB stackup planning seriously. The cost differential in adding an additional layer or two to a PCB is negligible in most cases. I’ve seen poorly-performing printed circuit boards that, after redesigning them with additional ground layers exhibit a reduction of over 30 dB in their radiated emissions
2. Switching power supplies are noisy — they can generate hundreds of harmonics. Take a close look at how the high frequency switching currents flow to ground and back to the power supply IC, and keep the size of that closed loop as small as you possibly can! Remember, you don’t want those currents to “go rogue”…..
3. Use internal PCB layers for distributing power — don’t route power signals on the top or bottom sides of the board unless you simply cannot avoid it
4. Every PCB trace should be located on a layer that has a ground plane immediately adjacent to it
5. If you must route a long trace on the top or bottom layer of a PCB, avoid placing it close to the board edge. A good rule of thumb is to not place any PCB trace closer to the board edge than three times the board thickness
6. If the PCB is to be powered with an external power supply (a “wall-wart” or some other type of supply) utilize a common-mode filter right at the point where the power comes onto the PCB
7. Do not route PCB traces across a break in an adjacent ground plane. If this is unavoidable, find a way to make it avoidable.

If anyone out there has additional items they would like to share on this topic, please let me know!

Until next time,

David

There are multiple sources of power loss in antenna systems — ohmic losses in the antenna conductors themselves and dielectric losses in coaxial cables and PCB materials are probably the two most common sources of loss.  Its unlikely that many antenna engineers these days have ever had to contend with an oak tree or juniper bush sucking power from their transmitting antenna, however such problems really do exist!  At low frequencies where transmitting antennas can get quite large, the presence of vegetation around the antenna structure can result in power loss and a subsequent degradation in antenna performance.

The first part of my career as an engineer was spent in Civil Service — I was a VLF (very low frequency) antenna engineer in the Submarine Broadcast Systems Division at a large Navy electronics laboratory in California.  Very low frequency radio signals are used by numerous Navies around the world to communicate with their fleet of submerged submarines.  As it turns out, radio signals in the frequency range of 20 kHz to 60 kHz can penetrate seawater to a depth of several tens of meters.  A submarine typically drags a long wire antenna behind it which streams along just below the surface, allowing the radio signals to be received.  The trouble with using VLF radio signals for communication is their long wavelength — at 20 kHz an electromagnetic wave has a wavelength of 15 kilometers, and radio transmitting antennas generally need to be at least one-quarter wavelength long in order to function efficiently.  As such, a quarter-wavelength antenna at 20 kHz is impossible to build.  Most VLF transmitting sites in use today employ transmitters in the 20 kHz to 40 kHz range that generate powers as high as 1 megawatt, and use antenna systems comprised of multiple towers over 1000 feet in height.  Capacitive top-loading for these antennas frequently incorporate grids of wire thousands of feet long that are suspended in the air by additional towers and spanning tens of acres of real estate.

Despite their tremendous physical size, VLF transmitting antennas are electrically very small, that is, their overall physical dimensions are significantly less than one wavelength at their operating frequency.  This makes a VLF antenna very similar to one of those small surface mount chip antennas you often find on printed circuit boards that are used for Bluetooth or WiFi.  Anyone who has designed with these types of surface mount PCB antennas knows that they are very inefficient — the placement of these antennas on a PCB must be done with care, as any objects in close proximity to the antenna (other components, ground planes, connectors, etc.) can significantly degrade their performance, reducing their efficiency even further.

The same problem exists with VLF antennas — it is not uncommon to find VLF transmitting antennas in operation today that have efficiencies below 30%.  As many of these transmitting systems are required to radiate many hundreds of kilowatts, any small degradation in antenna performance can result in very large increases to the electric bill each month!  As such, anything that can be done to mitigate power losses that occur in VLF antenna systems has the potential to reduce the overall operating cost of the facility.

The photo below is an aerial shot of the U.S. Navy’s VLF transmitting facility at Jim Creek, Washington. This radio facility has been in operation since 1952 and operates on a frequency of 24.8 kHz:

This antenna is known as a “valley span” antenna — multiple wires are strung across the valley and supported by towers along the ridges on either side.  These wire catenaries are each over one mile in length and form the capacitive top-loading for the antenna system.  At the center of each catenary a vertical downlead extends to nearly the valley floor, and then travels along a series of short supporting towers to the main transmitter building located at the center of the valley.  The entire antenna system has the appearance of a gigantic hair net suspended from towers on each ridge.  Notice that the foliage across most of the valley has been removed — this clear-cutting of foliage is intentional and is meant to reduce the power losses in the antenna system.  When the Jim Creek transmitter first went into operation, it was observed that the radiated power of the antenna system was somewhat higher in winter than in summer.  It was found that these differences were a result of power loss occurring in the foliage growing in the valley underneath the antenna system (Watt, A.D., VLF Radio Engineering, Pergamon Press, New York, 1967, pp.149-153).  As I have already stated, anything which degrades the efficiency of an antenna like this contributes to a significant increase in the cost of electricity to operate the station.

How can one actually measure the power loss in the foliage across an entire valley like this?  Well — back in 1991 I had the opportunity to do just that!  My team of engineers was assigned the job of estimating the power loss in the entire antenna system due to the foliage in the Jim Creek valley and providing a measurement report to the U.S. Navy.  To accomplish this task, we constructed a  “Frankenstein-looking”  apparatus out of a bunch of chicken wire and 4×4 lumber:

That’s me — standing in the middle of what is actually a parallel-plate capacitor.  The top and bottom plates are fabricated from chicken wire and supported by the 4×4 lumber.  Immediately to the left of me in the photo is a large toroid — Litz wire wound around a large truck tire inner tube (for those of you that have never heard of Litz wire, take a look here).  The toroid and chicken wire capacitor are connected in series, forming a tuned circuit.  A capacitance decade box, also visible directly above the toroid, is connected in parallel with the chicken wire capacitor so its overall value can be adjusted so the entire tuned circuit is resonant at 24.8 kHz (the operating frequency of the Jim Creek transmitter).  Our source of RF energy was a Hewlett-Packard signal generator driving a 100 watt commercial amplifier.

So — how does this work?  Our team spent some time scouting around the valley beneath the antenna to see the kinds of vegetation present, and we identified several “garden variety” shrubs.  After choosing a specific shrub for testing, we assembled the chicken wire capacitor so that the shrub was situated at the center of the bottom capacitor plate and protruding up into the space between the plates.  In order to make this assembly process easier, the bottom plate of the capacitor was made from two sections of chicken wire each with a semicircular notch cut into one edge.  When placed together, the two notches formed a circular hole which surrounded the base of the shrub.  The two halves of the chicken wire were then connected together with a bunch of alligator clips to ensure electrical continuity.  This tuned circuit, which now included the effects of the shrub, was energized with 24.8 kHz RF energy.   The capacitance decade box settings were then varied until the entire tuned circuit resonated at 24.8 kHz.  Our method of detecting resonance was simple, but extremely cool — using a two-channel oscilloscope, we connected one channel across the amplifier output in order to monitor the tuned circuit voltage.  The other scope channel was connected to a current probe that monitored the current flowing in the tuned circuit.  By placing the oscilloscope display into X-Y mode, the voltage and current waveforms combine and appear on the scope as an ellipse (the voltage and current are out of phase with each other) — for those of you old enough to remember, this is known as a Lissajous Figure.  Adjusting the capacitance decade box so that the ellipse collapses to a straight line indicates that the voltage and current are in phase — resonance has been achieved!

Now that the tuned circuit is resonant, the next step is to measure the circuit’s bandwidth.  Noting the value of the current at resonance, the frequency of the signal generator is varied both above and below 24.8 kHz to observe the frequencies at which the current drops to 0.7071 times its value at the resonant point (this is the 3dB point on the resonance curve).  Taking the difference between the high frequency and the low frequency yields the bandwidth of the tuned circuit.  With the bandwidth determined, the next step in the measurement is accomplished with a bucksaw and a set of pruning shears — the entire shrub is cut up and completely removed from the chicken wire capacitor, and the bandwidth measurement is repeated.  With the shrub no longer present, the total loss in the tuned circuit decreases, subsequently decreasing the bandwidth of the circuit.  By looking at the change in bandwidth between the two configurations, the “equivalent series resistance” of the shrub is determined.  Since the current flow in the tuned circuit is known, the power loss in the shrub is determined.

The diagram above is the schematic diagram for the test setup that we used.  The shrub appears in the circuit as a resistance along with the other ohmic resistance that is present in the circuit (mostly the resistance of the Litz wire in the toroid, but there is also some resistance in the wiring and miscellaneous connectors used in the setup).  What follows is the math used to extract the equivalent series resistance of the shrub.

When the shrub is part of the circuit, the “Q” of the tuned circuit at resonance is given by:

Q_{shrub} = \frac{2\pi f_0 L}{(R_{loss} + R_{shrub})}

Once the shrub is removed, its equivalent series resistance is no longer present, so the expression for the “Q” at resonance becomes:

Q_{noshrub} = \frac{2\pi f_0 L}{R_{loss}}

The “Q” of a tuned circuit is inversely proportional to its bandwidth — another way to express the “Q” of a tuned circuit is like this:

Q = \frac{f_0}{\Delta f} = \frac{f_0}{BW}

Using this definition, we get a pair of equations:

\begin{aligned} BW_{shrub}& = \frac{R_{loss} + R_{shrub}}{2 \pi L}\\ BW_{noshrub}& = \frac{R_{loss}}{2 \pi L} \end{aligned}

Solving these for the equivalent series resistance of the shrub:

R_{shrub} = 2 \pi L (BW_{shrub} - BW_{noshrub})

Notice that the inductance of the inner tube toroid needs to be known — we measured the inductance of the toroid in the lab, so this was a known quantity.

So far, so good — but how does this relate to the power dissipated by a shrub when it is sitting on the valley floor under the antenna? We’ve computed an equivalent series resistance for the shrub, and we can compute the power it dissipates by utilizing the current flow in the tuned circuit at resonance:

P_{shrub} = I^{2} R_{shrub}

How do we use this resistance when we consider the entirety of the foliage that sits beneath the antenna? The answer may surprise you — we actually don’t use this resistance at all! We introduced the idea of an equivalent series resistance of the shrub solely to permit us to compute the power dissipated by a single shrub. With the actual dissipated power known to us, the next step in unraveling this measurement for the entire valley beneath the antenna is realizing that the power dissipation that occurs in the shrub is a direct consequence of it being immersed in the electric field of the antenna system! If we can measure the electric field that exists inside the chicken wire capacitor around the shrub, we can establish a relationship between the power dissipation in the shrub as a function of the surrounding electric field.

Fortunately, in order to get at the value of the electric field inside the chicken wire capacitor we don’t need to measure it directly. At VLF, the electric field inside the chicken wire capacitor is so slowly varying in time that it can be treated as a static DC electric field. Even more important, so long as we only consider the behavior of the field near the vertical axis of the capacitor (we stay away from the edges of the chicken wire), the electric field is nearly uniform (that is, its value does not change with position). Using this very-good assumption (once again, at VLF this works!), the electric field inside the chicken wire capacitor where the shrub is located is just equal to the voltage across the capacitor divided by the spacing between the top and bottom plate:

E = \frac{V_{capacitor}}{d_{plates}}

During our measurements, we did not directly measure the voltage across the capacitor — we don’t need to make this measurement, as we have already measured the bandwidth of the circuit when the shrub was present inside the chicken wire capacitor. How are these two quantities related? It turns out that, for a series resonant circuit, the voltages across the individual reactances at resonance are equal to each other (although they are 180 degrees out of phase) and have a magnitude equal to:

V_{peak} = Q V_{in}

where V_in is the input voltage across the entire tuned circuit. Making a few substitutions, the expression for the voltage across the capacitor is:

V_{peak} = \frac{f_0}{BW_{shrub}}V_{in}

It is a good thing we don’t need to measure this voltage directly — it is in the ballpark of several thousand volts! However, with this information at hand, the magnitude of the electric field inside the chicken wire capacitor is known to a good degree of approximation, giving us a relation between the power lost in the shrub when it is immersed in an electric field equal to that inside the capacitor. Estimating the power loss for any other electric field strength is made by noting that the power absorption by a conductor immersed in an electric field is proportional to the square of the electric field strength. When the Jim Creek radio facility is transmitting, the electric field on the valley floor is easily measured with standard field strength equipment, and the resulting field map is overlaid on a topographic chart.

The last step in the entire process of measuring power loss from shrubbery was perhaps the most difficult — we needed to count the shrubs! Here again, the use of a good approximation tool came in handy — we picked out a number of areas around the valley approximately 50 feet by 50 feet in size, and walked these areas while counting the shrubs. The areas we chose were based on how uniform they looked to the human eye as well as where in the valley they were located. Extrapolating this information out to the entire valley using photographic data allowed us, at long last, to arrive at a power loss estimate due to shrubbery residing in the valley.

How accurate were our measurements? When dealing with an antenna as large as the Jim Creek antenna, there are so many variables that contribute to the total power loss it is nearly impossible to get an answer with an accuracy better than 25% to 30%. As part of our measurement work we had the advantage of being able to review power loss data taken throughout the history of the station’s operation. The primary objective of our measurement effort was not so much to determine the actual power loss at a high level of absolute accuracy as it was to determine that no significant changes to the antenna system power loss had taken place. This knowledge is valuable from the standpoint of establishing an overall trend in antenna performance — this information is invaluable for predictive maintenance purposes.

A final comment about that picture of me inside the chicken wire capacitor — one of the perks of this job was my entire team got the opportunity to have our loss resistance at 24.8 kHz directly measured. The photo was my turn in the capacitor — I don’t remember what my loss resistance was, but I was sure to keep my head down, otherwise I would have burned the hair off the top of my head if it touched the top plate.

Until next time,

David

Thank you for stopping by to take a look at what is going on here at GoldSpire Design!  This blog is dedicated to all things associated with electronics, electrical engineering, and product development.  I’ll be posting things that everyone will hopefully find educational, thought provoking, entertaining, and useful to you in your business or product development efforts.  If any of my readers have any ideas for new and interesting topics to discuss, please bring them up so we can all brainstorm about them.

Thanks again for coming, and hope to talk to you soon!

David