The STL debate goes on

Actual elevations can vary considerably from those spewed out by the computer, especially in very irregular terrain.

In STL systems, the issue of which modulation scheme to use seems to be more of an argument for the salesmen to pursue than for the engineer to use in system design.

In what should be a quiet corner of the industry, a discussion seems to continue about systems for digital television. The argument seems to be that STLs won't work as well for digital television as they have for analog. However, the argument seems to be mainly built on fear and superstition rather than on sound engineering principles.

Some facts are evident. The systems for DTV do have a slightly lower value for power output and/or receiver threshold. This is primarily due to the bandwidth involved in such systems. Some of the manufacturers are using the same power amplifiers for DTV systems as were used in the existing analog radios. The increased bandwidth, especially in the dual systems where two signals occupy the same microwave channel, will normally lower the available power. This directly relates to the gain-bandwidth product that applies to any amplifier.

It is noted that the receiver threshold is a few dB poorer for DTV systems. Again, the bandwidth is increased to handle the requirements of the large data capacity required. An increase in bandwidth is always accompanied by an increase in noise that in turn raises the level of signal required at the receiver if the signal-to-noise ratio is to be unchanged. However, that doesn't mean that the system will be less usable or dependable. It simply means that a little more gain must be build into the system to maintain the same fade margin and reliability.

It is also interesting to note that the same difference of opinion seems to exist in STL systems as in the main broadcast signal. Manufacturers of STL systems are making use of different modulation schemes. Some are using an 8VSB system while others are using COFDM. They have their own reasons for the methods selected and are ready to defend their positions vigorously. However, both seem to work satisfactorily in STL service, which differs significantly from broadcast signals. In STL systems, the path is carefully designed to be free from obstacles, and the signal strengths are carefully calculated at the receiving point. There are no indoor antennas, nor do antennas move. Antenna orientation is controlled and fixed. In such an environment, the differences in modulation schemes seem to make less difference in the final system performance. This seems to be more of an argument for the salesmen to pursue than for the engineer to use in system design.

The primary factors in the system design continue to be the transmitting system output power, the path, the antenna systems and the receiver sensitivity. If those criteria are properly used to design a conservative link, the system should perform as well as the older analog systems.

In looking at some problems that have caused STL difficulties, one problem still exists. Numerous software programs are available that will perform a path plot. These programs all make use of either a 30-inch or a three-inch terrain database. To explain the terminology, a 30-inch database contains the elevation of terrain for each 30 inches of latitude and longitude. All other terrain points are determined by interpolation using the elevations at the corners of a square containing the point in interest. The cheaper programs usually come with a 30-inch database, as it requires much less memory. On the other hand, the better (maybe) programs use a three-inch terrain database that is much larger and more expensive. While the databases are obviously different in the interval between points, they also differ in the maps from which the data is obtained.

Database information is also available with three-meter intervals between points. That data is obviously the most accurate available and was taken from 7.5-foot topographic maps. Unfortunately, that data requires an enormous amount of space and is very expensive. As a result of all of these differences, many paths are run using a 30-inch database and are assumed to be accurate. After all, the computer says it is so and who are we to question the computer. Right? The problem is that actual elevations can vary considerably from those spewed out by the computer, especially in very irregular terrain. The author has seen errors of over 100 feet in elevations in very rough terrain or in the vicinity of large terrain irregularities such as ridges or bluffs.

The solution is fairly simple. First, use the computer program and database of choice to run the desired path. Normally, there will be one or two points that control the receiver and transmitter elevations needed to provide a path that gives 0.6 Fresnel zone clearance over the ground (don't forget to include trees when applicable). The better programs will allow the user to identify the coordinates of such points. Then refer back to the latest 7.5-foot topographic maps. Using the coordinates from the computer study, identify the point that is affecting the path and confirm the exact elevation. That elevation can then be added to the computer-generated plot to check for adequate path clearance. The same applies to the exact elevation of the ground at both the transmit and receive locations.

If there is still any worry about the elevation accuracy, go to the point that is in question with a calibrated altimeter and check the elevation yourself. The altimeter should be taken to the benchmark nearest to the point in question and set to the elevation of that benchmark. Then go to the point and record the elevation and then transfer that to the computer plot. If all of that doesn't work, you have really angered the terrain gods and there is no help for you.

Finally, your author has goofed (again). In a past article, the equation for determining path reliability was given as determined by Barnett and Vigants of Bell Telephone Laboratories. That equation may have been incorrectly printed. It should be as follows: T = a × b × 2.5 × 10^(-6) × f × D^3 × 10^(-F/10), where T = time out of service as a fraction; a = 4 for very smooth terrain including over water, 1 for average terrain with some roughness, and 1/4 for mountainous, very rough or very dry terrain; b = 1/2 for Gulf Coast or similar hot, humid area, 1/4 for normal interior temperate or northern areas and 1/8 for mountainous or very dry areas; f = frequency in GHz; F = fade margin in dB; and D = path length in miles.

Now, for the next goof — probably the dumbest of them all. The author recommended “Engineering Considerations for Microwave Systems,” published by GTE Network Systems, as a reference for microwave path planners. That book is no longer in print. If you can find one on someone else's bookshelf, take it to the nearest copy machine. Otherwise, you are out of luck. However, other good reference books on microwave path planning are available. One is “Radiowave Propagation” by Lucien Boithias, which can be obtained from McGraw-Hill. Unfortunately, it is like most textbooks in that it tells you more than you really want to know. Yet, all of the good stuff is in there — you just have to hunt a bit.

Finally, the questions seem to keep on coming regarding STL systems. To help resolve those questions, the author offers this column as a medium to treat those issues. Let us hear from you regarding your experiences or questions concerning digital STL systems. We will attempt to either get answers for your problems or, at a minimum, offer them up for others to provide solutions based on their experiences.

Don Markley is president of D.L. Markley and Associates, Peoria, IL.

Send questions and comments to: