NRSC AM bandwidth measurements with the loop antenna
Copyright 1994 Chris Scott
This note is a reprint of the article "Looking at Bandwidth with the Loop Device", published
September 7, 1994 Radio World. The original photographs have been replaced.
With the first year of AM bandwidth measurements behind us, my associate Roger Hall and I have learned important lessons about the best way to approach these new tests.
We tested a number of stations and gained some practical knowledge in this area and identified some potential pitfalls. About thirty percent of the stations we tested initially failed, with an even split between bandwidth and harmonic related problems. In all cases these problems were curable-the testing led us to transmitter defects, usually modulator problems or harmonic trap misadjustment. Aside from simply meeting the spectral purity rules, these tests had beneficial side-effects; true defects were identified and repaired.
According to Bernie Stuecker, Chief, Equipment and standards branch at the commission, low
signal-to-noise, interference from near-frequency stations, and the sometimes misunderstood
effects of antenna factor are probably the most common culprits affecting accuracy of these
Stuecker sets testing standards and procedures to be used by field operations personnel when determining station compliance. While at his office discussing AM bandwidth testing methodology, I asked him about the once common practice of using a communications receiver to check harmonics. Specifically I asked him if it was acceptable practice. "We would see [harmonic measurement] reports on file at the station indicating that the second harmonic was barely audible or so many S units below carrier. If it is anything other than inaudible, how do you know if it meets the attenuation specifications?" Stuecker asked. He indicated that they use a field strength meter to check harmonic levels.
We found that the receive antenna used with the spectrum analyzer is critical;
our experience comparing various antenna types used with the Tektronix 2712
showed that a broadband shielded loop is arguably the best choice.
Tek published a related technical brief on the subject available in pdf format here.
Signal to noise important
June was our busiest month conducting these measurements, and nearby thunderstorms contaminated the noise floor on more than one occasion. Locations away from power distribution systems and industrial areas were naturally the best. Achieving proper signal-to-noise and confirming that the client station was actually generating the recorded energy were probably the most important tasks.
In difficult cases we found it necessary to shut off the station to verify the emission source. Normal considerations for taking AM field strength measurements apply, and we found the best results locating at the specified one-kilometer or closer; when measuring the bandwidth of omnidirectional stations we found that proximity as close as one or two wavelengths showed the same results but with fewer noise problems.
The rules specify "approximately one kilometer" for operating stations and in my opinion this is a repeatable compromise between high signal-to-noise ratio and being truly representative of the signal that is radiated into the far field. It is easy to mistake ambient or electrical noise for the station's emission; if the measurement is erratic, it pays to corroborate it at a lower noise location or with a transmitter tap sample.
The level that a sample port provides should be checked before connecting a spectrum analyzer direct because many instruments will be permanently damaged by more than a hundred milliwatts.
When chasing harmonic problems with a transmitter tap the engineer should note what typed of
pickup is used; a reflectometer-type normally exhibits a six decibel per octave increase in
sensitivity, while resistive dividers should be flat. Our equipment power came from a fairly large
UPS that produced a sinewave output and is well shielded. Some units radiate and should be
tested prior to beginning a measurement series. We have since constructed a customized well
filtered and shielded inverter, supplied by vehicle dc.
If the major noise source is another local station, the twenty-five db null obtained with the shielded loop will not be enough to reduce it to near the level of the ambient noise. Obviously, arranging for the interfering station to be off air during the measurements will eliminate this problem, but if it happens to be the competition, it may be difficult to convince them to do this in the middle of the day, particularly if measurements need to be repeated for any reason. In this case, with the client station off-air it's best to record a spectral plot of the ambient RF environment and inclued it in the final data, demonstrating what cannot be blamed on the staiton. We usually recorded a plot for report inclusion showing the station nulled at least twenty decibels to identify which signals followed the null. One way to couple energy into a spectrum analyzer is to use a simple whip antenna connected directly to the fifty ohm input of the instrument. While this may be useful for quick-and-dirty checks, attempting to get meaningful harmonic data this way will almost certainly be misleading, often mistaken by ten or even twenty decibels. At least three things must be known about the test antenna; the antenna factor, or relationship between its output and the field that its placed into, the impedance match which affects performance with various lengths of coax, and what mismatch loss needs to be considered. This performance can be measured over the frequency range of interest (in our case 500 kilohertz to five megahertz) and combined into a calibration factor. Commercially available antennas for this purpose are rare, so we developed and calibrated our own.
Although the FCC rules granted a grace period for measuring close-in bandwidth,
yearly harmonic measurements were still required. Now that the grace period
has passed, both close-in bandwidth measurements and harmonic measurements
are required. Our experience showed that if the stations are in compliance
at the second and third harmonics, higher order products were the same level
The most favorable locations for close-in bandwidth tests are less than ideal for harmonic measurements because groundwave attenuation increases with frequency and is particularly noticeable at the third harmonic. In one case while corroborating data recorded at one kilometer, measurement at three showed several decibels improvement. Whether or not separate measurement locations are acceptable to the commission remains a question; this practice should probably be the exception, used only when compelling reasons exist.
Transmitter power levels of five kilowatts and above must meet the full eighty decibel specification. These often have harmonic traps and, once properly adjusted, usually have little trouble with compliance. Transmitters without traps may be more of a challenge. If all else fails, adding a trap to the atu should cure the problem. Although we used a Potomac FIM-41 field intensity meter for harmonic data, accurate measurements are available from a spectrum analyzer with a proper antenna. One caveat here: We saw some artificially high harmonic indications which were created inside the front end of the spectrum analyzer. As good as modern spectrum analyzers are, accurate eighty decibel on-screen dynamic range may be a stretch under certain conditions. We initially increased the input attenuation, which changed the ratio more than the input level; this is the tipoff. Selecting a lower first mixer level helped some, but to get consistently accurate harmonic data we needed an external tunable notch filter.
Filter insertion losses should be measured in the lab and tabulated for field reference. One final point is the effect of various program material. Many stations are trending toward talk. This restricted bandwidth audio often paints a rosier picture than music.
More repeatable results can be obtained using the USASI noise as recommended in NRSC-2. However, the rules again are mute as to whether this program source is acceptable.
We now examine design and construction of two loop antennas: a shielded, untuned receiving loop to be used in the field with the spectrum analyzer, and an unshielded transmitting loop used to generate a frequency-independent standard field, to calibrate the former. The technique described here is very similar to the way NIST calibrates AM field intensity meters.
Shielded loop antennas respond primarily to the magnetic component of the RF field, and provide good directionality in the form of a figure-eight with nulls at right angles to the plane of the loop. Compared to an amplified (E-field) whip. Empirical tests prove the shielded loop to reject substantially more electrical noise. The key benefits of the loop stand up well for AM emission testing. Minimal electrical noise pickup, with proper site selection an interfering signal can be nulled at least twenty decibels, and antenna factor is easily generated using the method described below.
The loop is also an excellent general purpose pickup; positioned near an AM tower, it provides a nice clean sample to a scope, spectrum analyzer, or frequency counter- all of which are useful to have available when "proofing." Normally the sensitivity of the loop increases directly with the speed that the lines of flux cut across the windings; this translates into a six decibel per octave increase, which is useful for harmonic measurements.
More sensitivity makes the lower level harmonic energy easier to resolve. Actual measurements proved this to hold well until parasitic reactance becomes a significant portion of the loop impedance. Our initial shielded loop was constructed of 1.25 inch square aluminum tubing with a gap at the top to prevent the shorted turn effect. Two turns of number eighteen insulated wire were wound inside, using wooden collars to maintain wall spacing. More turns can be used, but the increased inductance and stray capacitance will reduce the frequency at which antenna factor anomalies begin. Often, for communications purposes, the loop winding is resonated with a parallel capacitor, greatly increasing the Q and the output. Although this can be useful for improving the antenna sensitivity and selectivity for harmonic relationships, it can't be used for bandwidth testing because it's easy to bandpass out the sideband energy you're searching for.
Although theoretically, balance will be adequate by using shielding, we grounded the winding midpoint and used a broadband 1:1 transmission-line transformer for enhancement. A good measure of balance is null depth; this loop averaged more than twenty decibels. The standard field antenna consisted of a single turn supported by steatite bushings installed at the top of a shielded meter and matching box. The loop is fed through resistors and a balun. The RF current must be carefully metered as this must be held constant.
A reliable antenna factor can be accurately generated using the near-zone magnetic field of a small constant current loop. NIST calibrates field-intensity meters this way. Because the loop size is very small compared with the wavelength, the antenna current remains essentially constant throughout the conductor, resulting in the radiated near-zone magnetic field being constant over the decade .5 to five megahertz. This field does not follow the inverse distance rule, falling off rapidly beyond a couple of meters. We spaced the loops two meters apart center-to-center, which proved to be a good compromise between field strength and sensitivity to placement repeatability. The reference radiator was mounted one meter above ground on a wooden platform, and the shielded loop antenna on a tripod at the same height. Although this mini-antenna range setup is considerably more tolerant to the presence of adjacent structures that VHF antenna testing, reflective objects should be kept to a distance of at least three times the spacing between antennas.
We excited the reference radiator with one-hundred milliwatts of harmonic-free RF. At this distance our FIM-41 meter indicated six millivolts per meter with co-planar (normal) alignment, and twenty-one when aligned co-axially, the way NIST does it. According to the FIM, this field varied less than three percent over the decade in question. With this standard field established, calibration of small aperture magnetic field antennas properly positioned becomes pretty straightforward. If constructed exactly like our shielded loop, the six decibel per octave rule will hold well, until slightly above two Megahertz, where the slope starts to taper. When combined with both a modern spectrum analyzer and a notch filter, an accurately calibrated antenna completes the package necessary to begin "proofing".
A tripod supported the loop antenna in the field, and the entire apparatus is guaranteed to generate interest from passers-by who must surely believe that you're DF'ing for Nazi spies.