While the subject of antenna measurements was discussed in a past column, questions still remain concerning what should be done, what it means and how they should be performed.
Measuring the UHF TV panel array’s performance upon installation is usually done by choosing a remote site and measuring the signal level received from the transmitter. Periodic measurements at that same location will reveal the amount of any degradation so corrective action may be taken. Photo courtesy Radio Frequency Systems.
First, there is nothing magic about the new equipment being used for antenna measurements. The instrument of choice is currently the vector network analyzer from any of several different manufacturers, the most popular being Agilent. In the past, antenna measurements were performed with slotted lines, the well-known admittance meter from General Radio, or the impedance bridge from Hewlett Packard. Various schemes of test equipment were used to design home brew Smith Chart plotters and to incorporate older network analyzers into complex systems that would show plots across the entire channel at once. Those usually included a computer, various analog-to-digital converters and their inverse with a frequency synthesizer. The problem with such composite test systems was that the overall noise floor had a tendency to be greater than the reflected signals one was measuring.
However, the measurements taken with the slotted lines or admittance meter were normally quite accurate. The main problem was not in accuracy, but in speed. To understand what is happening in a wideband antenna, such as those used in broadcast systems, one needs to be able to see the results of many measurements across the band of interest. Just checking one or two points doesn't give an accurate representation of the system response. That meant taking a ton of individual measurements, measuring each frequency with a counter, and plotting them on graph paper. A sweep over 6MHz with the subsequent analysis took a few hours. There was a need for an instrument, such as the network analyzer, in one box that would perform all of those functions.
Before any measurements are even attempted, the network analyzer should be calibrated on the frequencies of interest. It measures on a number of discrete frequencies in the assigned band. For most HP instruments, the default is 201 points. However, that usually can be changed to 1601 points, which gives a smoother and more detailed result. The calibration process must include any impedance bridges, couplers and cables that are included in the final test setup. The calibration is performed using a calibrated load, open and short. Just leaving the end of the test cable open isn't enough to do this properly. The point where the open and short occur should be at the exact same location as the point where the 50Ω termination appears. It really doesn't make an enormous difference if one is measuring VSWR values around 1.5:1. But, if you are measuring values like that in a modern antenna, you have bigger problems than errors in calibration.
The calibration process measures the magnitude and phase of the reflected signal from the three test transitions. Then, knowing what the load, open and short should look like, a matrix of correction coefficients is generated. The instrument is now ready to perform the desired measurements. It should be remembered that the accuracy of all the measurements depends on the accuracy of that 50Ω load. The calibration process has taken away the errors in cables, directional couplers, etc. assuming that the load is really 50 plus j 0. If the engineer/technician performing the measurements at your station does not do this calibration prior to connecting to the antenna system, escort him gently to the edge of the site and wish him a fond adieu. Without field calibration, the measurements are junk. The 50Ω load is critical. It cannot be just any 50Ω termination. Off-the-shelf terminations such as one would normally use on circulators or directional couplers aren't accurate enough. The load should be calibrated by the manufacturer to have a return loss of more than 40dB. Further, the load should never be used for anything but the calibration of the test equipment. Most engineers maintain a new load at the office, which is periodically used to check the load taken into the field. When they differ, the field load buys the farm and a new load is purchased, usually for the price of 50Ω.
Remember the network analyzer only measures the magnitude and phase of the reflected signal on each of the 1601 or less frequencies in the assigned frequency band. Those are the only measurements — everything else is manipulation of those measurements by the test equipment itself. First, the measured signals are compared to the forward-going signals to determine the actual measured return loss in both magnitude and phase. That data then is corrected by use of the calibration matrix of coefficients. The corrected data then is displayed in one of several formats as chosen by the operator. These normally include the return loss, the log of the magnitude of the return loss, Smith Chart, polar plot and VSWR.
Equally important, the network analyzer should be equipped with the necessary option to give a time domain presentation. This is done by using a version of Fourier Transform to convert the measurements at individual frequencies to a presentation of what is happening as time passes — the time domain; in other words, to act like a TDR over a limited range of frequencies. This allows a quick analysis of how well the match is between the transmission line and the antenna. Further, the performance of elbows and connectors between sections of transmission line can be checked, waveguide can be tuned along its length, and the transmission line itself can be checked even down to individual insulators along a section of line. Until the system has been checked in this fashion, you don't know what is going on.
The problem is that the presence or absence of VSWR at the input of the line is not sufficient to determine how the system will operate. It is critical to know where that VSWR is coming from. If the problem is occurring at the gas barrier right at the output of the transmitting equipment, it probably will pose far less of a problem than if it is at the input to the antenna. Certainly, it won't cause ghosts or be as significant a problem concerning the bit error rate of the digital signal. The more significant problems result from the signals traveling up the line, being reflected, coming back down, going back up, etc.
Another enormous point in performing antenna system measurements is the necessity of a tuned adapter between the test equipment and the transmission line. The test equipment is normally 50Ω with a type “N” connector at the end of the test cable. The adapter must permit the connections to the transmission line, possibly including an impedance change to 75Ω, without introducing VSWR into the system. A review of the literature concerning network analyzers shows that the problem of the input connection is critical to the accuracy of the test results. Calibration requires having a calibrated load with its own adapter separated from the adapter being tuned by a reasonable length of known transmission line. Putting the adapters back to back is not satisfactory as they are then being tuned to an unknown impedance — the one where they come together — as opposed to 50Ω. If there is a section of transmission line in the middle, the adapters must both reach 50Ω. A lot of stations have started buying a factory tuned set of adapters — two is a good idea — and keeping them on hand for future measurements.
The problem goes back to the original point that the only thing the instrument measures is the magnitude and phase of the reflected signal. Everything else is manipulation of that data. If the instrument sees a large mismatch at the input, low-level reflections returning from smaller mismatches are lost, especially those occurring near the input where they are overwhelmed by the larger reflection. Some network analyzers permit bad reflections to be eliminated by “gating” functions that mathematically eliminate the “bad” reflections. This is fine if small mismatches are being ruled out, but not satisfactory for large errors. Remember what the equipment is doing — and remember that every time the data is manipulated, fondled, processed or changed, the opportunity exists for error to come creeping in — even in the digital world where the ubiquitous rounding error rears its ugly head.
The point now has been reached where the equipment is ready for the measurement and the initial measurements have been taken. The VSWR is not bad but not as good as we would like — what do we do now? Tune in next month for that — and you thought “To Be Continued” only occurred on TV.
Don Markley is president of D.L. Markley and Associates, Peoria, IL.
Send questions and comments to:firstname.lastname@example.org