Last month, we discussed the vector network analyzer’s functions and its on-site calibration. We covered the need for calibrated adapters and gave a brief explanation of how to use gating techniques. Now we can discuss measurements.
The analyzer supplies the incident (reference) signal to the device under test, which transmits part of the incident signal and returns the rest. The analyzer’s receiver compares the transmitted signal and the returned signal to the reference signal (see Figure 1).
Figure 1. A network analyzer uses an incident signal, along with the reflected and transmitted portions of that signal to measure various characteristics of RF transmission lines, antennas and other components. Click here to see an enlarged diagram.
Before you begin taking measurements, make sure that the voltages present on the transmission line are not too high. The analyzer puts out 20dBm when turned up all the way, and can cope with levels somewhat higher than that before the front end of its receiver burns out. The analyzer will indicate the presence of high signal levels on the transmission line by displaying error messages, usually indicating that the analyzer is losing its lock on the reference signal. The next indication of excessive signal levels will be the failure of the analyzer’s front end. To prevent this, devices are available that limit the signal power to the analyzer. If you have any doubt about the levels on the line, you should use such a device.
Data collection and storage
The vector analyzer takes measurement data, compares them with a matrix of calibration data and stores them, along with information about its settings. This means that, after taking measurements, you can disconnect the analyzer from the line and display the results in whatever format you want, including time- and frequency-domain presentations. (Of course, if you change the instrument settings, such as the frequencies to be measured, you have to take a new set of data.)
The first measurement you probably want is a frequency-domain presentation of VSWR. Some engineers prefer to use return loss. It really makes no difference, since both measurements are based on the same data.
A display with a bandwidth of 12MHz shows not only the channel of concern; it also offers a better picture of the system’s bandwidth. In the time domain, that bandwidth can give you a good idea of the match between the transmission line and the antenna. A narrower bandwidth will make it more difficult to determine the location of a mismatch near the antenna. A wider bandwidth will cause a reflection that masks the actual match to the antenna. As a rule of thumb, the time-domain VSWR at the antenna should be 1.04 or less.
For analog systems, the VSWR should be near or below 1.1:1. This measurement is significant at the three carriers: visual, aural and color. It is most significant at the visual carrier because that is where the greatest power density occurs. Even if you can’t bring the system below 1.1 at the aural carrier, it is still acceptable for normal use. The same applies to a peak or two that exceed 1.1 at frequencies other than the visual carrier. In complex systems with many elbows, it simply may not be possible to get everything down around a 1.05.
For DTV, a good target for tuning is a return loss of 30dB. This is often not possible, especially at the higher UHF channels. Some argue that it is more important to look at the average return loss or VSWR over the channel than at some single part of the channel, where the VSWR may be a little higher than desired. Engineers are investigating that very question. It is simply a bit too soon to know the exact impact of antenna VSWR on the digital signal, other than to say you would prefer a good match.
The Smith Chart
At first glance, a Smith Chart representation of the entire system appears to be just a bunch of circles, but these circles offer useful information. A group of small circles tightly wound but offset from the center of the chart indicates a mismatch close to the input. Conversely, large circles centered on the chart indicate a mismatch at a distance from the input. This just tells that a mismatch exists; it won’t tell you exactly where the mismatch is located, as the time-domain presentation will.
Finding the mismatch
When trying to determine if there is a mismatch in the antenna or if the problem is in the elbow complex right at the antenna, you must measure the system’s response at the end of the transmission line right at the antenna. Some network analyzers allow you to gate the transmission-line response out of the system. All allow you to calibrate the system at the antenna. The resulting measurements will be the response of the antenna only, without the rest of the system. The hardest problems to really pin down perfectly are those that occur right at the base of the antenna. There is usually a lot of hardware at that location, such as a group of elbows, impedance-matching sections, tuning sections, etc.
When tuning matching sections, remember that the analyzer is not a swept device giving an instantaneous response. You must wait until the equipment completes a set of measurements and updates the display. This doesn’t take long and it is obvious when it occurs, but it means that you must move the slugs very slowly. The best way to do this is to have the rigger make a small move and then wait for the equipment operator to advise him to move further.
Coax and waveguide measurements
The next big set of measurements, those for coaxial and waveguide systems, take place at a much wider bandwidth. For waveguide measurements, the band of frequencies cannot exceed the cable’s cutoff frequency. In other words, don’t try to look at a waveguide from 50- to 200MHz. The waveguide will have reverted back to the warm-air-duct state and not look too good in any presentation. There is normally a reasonable bandwidth, such as 50- to 75MHz, where discontinuities will show up in the time domain. A narrower bandwidth, such as 12MHz, is better for tuning matching sections for the channel of interest. You can use a wider value to look for a problem.
In coaxial systems, a bandwidth of 100MHz will reveal most problems such as dents, bad center-conductor connectors, bad gas barriers, etc. The antenna will look terrible with this bandwidth, unless it’s a panel antenna with wide bandwidth. Any tuned sections will also look bad because they will normally have been optimized for the desired channel.
A final set of measurements with a bandwidth of 350MHz will show even the smallest of discontinuities. Again, the problem with using such a large bandwidth is that you must be careful not to exceed the cutoff frequency of the cable. For television systems, that is really only a problem with cables larger than seven inches. With such a large bandwidth, the display will show down to the individual insulators on the center conductor of the line. You’ll be able to spot as small a problem as a single bent-over insulator. Obviously, such a problem will probably not cause difficulties over the years. But, you won’t know if it’s a bent pin or a dead mouse until you actually open the line up and pull it apart. Don’t laugh. Dead mice, birds, bird nests, broken center-connector parts and various tools have all been found in coaxial line and waveguide systems.
Print and save
Print the test results to a printer or, even better, save them to an electronic file so you can import them into MS Word and have a clean copy for reference when a future problem makes it necessary to repeat the test.
Don Markley is president of D.L. Markley and Associates, Peoria, IL.