RF Column 17 - February 1993 Copyright (c) 1993,1995 H. Douglas Lung ALL RIGHTS RESERVED TOPICS: Frequency Measurements: Counter accuracy limitations Measuring frequencies using a spectrum analyzer Checking calibration using a standard frequency station Calculating RF power densities near parabolic antennas Example showing calculations for an SNG truck RF safety Satellite digital video compression update IMPORTANT NOTE: Because ASCII text does not have a standard character for superscripts, the characters ^2 in the formulas in this article mean take the square of the associated variable. -------------------------------------------------------------------- Last month I reviewed the FCC regulations on frequency measurements. This month I'll cover some of the methods for measuring frequency, with more to follow in future columns. Several months ago I promised more information on the NTIS paper and graph for calculating power density in the near field of microwave (parabolic) antennas. This month, its here. Finally, I had a chance to look at a video compression system in operation last month. By the time you read this, it is likely that the "standards" will have changed again, but read on for my observations. Counting frequencies... For most engineers, a frequency counter is the instrument of choice when measuring frequencies. While the sophisticated counters from Fluke and Hewlett Packard are great for complex frequency/time measurements, I've found that the inexpensive counters from OptoElectronics in Fort Lauderdale, FL (305-771-2050) work very well for field work. I use their Model 2810 counter. Other manufacturers of inexpensive counters include Ramsey Electronics (716-924-4560 in Victor, NY) and StarTek (305-561-2211) in Fort Lauderdale, FL. There are some caveats to using inexpensive counters for FCC frequency measurements. First is accuracy. Just because the counter reads out to 10 Hz. at 999 MHz. does not guarantee it will be accurate within 10 Hz. Counter accuracy is largely dependent on timebase accuracy, usually specified in parts per million (PPM). Cheap counters without a compensated crystal oscillator timebase have accuracy's between 1 and 2 PPM. To quickly see how this affects a frequency measurement, consider that a 2 PPM timebase may be off by as much as 2 Hertz at 1.0 MHz. This error increases with frequency. At 176 MHz., the maximum error due to timebase error will be 176 MHz. times 2/1,000,000, or 352 Hertz. Note this counter could be used for FCC frequency measurements for TV broadcast service at this frequency, if the error was subtracted from the FCC tolerances. The FCC permits up to 1,000 Hertz deviation from the specified visual carrier frequency. Since the counter accuracy could be off by as much as 352 Hz., plus or minus one count of the least significant digit (let's say 10 Hz. for this example), the measured frequency must be within 1,000 Hz. minus 352 Hz. minus 10 Hz., or 638 Hz. Since the least significant digit of the example counter was 10 Hz., round down to 630 Hz. The same counter would not be suitable for TV carrier measurement at UHF frequencies. At 700 MHz., the timebase error would be 700 MHz. times 2/1,000,000, or 1,400 Hz., outside the 1,000 Hz. tolerance the FCC specifies for TV broadcast or precision offset LPTV stations. Timebase accuracy can be improved in counters by using either a temperature compensated crystal oscillator (TCXO) or a crystal oscillator in a temperature controlled oven. OptoElectronics offers a TCXO option for their counters which improves timebase accuracy to 0.2 PPM. This makes it accurate enough for UHF TV carrier measurements if the error is allowed for (as described above). At 700 MHz., the timebase error on a TCXO counter with 0.2 PPM accuracy would be 700 MHz. 0.2/1,000,000, or 140 Hz. plus or minus one count (least significant digit). Ramsey offers an ovenized crystal oscillator timebase with 0.1 PPM accuracy as an option for their inexpensive counters. In more expensive lab type counters, accuracy's of 0.05 PPM or better are common. A timebase with an oven crystal oscillator needs time to warm up and stabilize. A TCXO compensates for temperature, but it too should be allowed to stabilize as well, particularly if it is moved from one temperature to another. As you've probably noticed, you can't simply hook a counter up to an RF sample port in the line to the antenna and get a reliable frequency reading on TV broadcast signals. The aural carrier, usually 7 to 10 dB below the visual carrier, messes up the readings. The simplest way to get around this problem is to measure the visual frequency at the exciter, before it is combined with the aural carrier. In most transmitters the aural carrier can also be measured separately. Audio modulation will make aural carrier measurements difficult and a large amount of chroma on the visual carrier may make visual carrier measurements less accurate. Some exciters, particularly those used in LPTV transmitters, only have one output with visual and aural carriers combined. Most have a switch to shut off the aural carrier for visual frequency measurements. Sometimes you have to go looking for the switch. The new economy LPTV exciter from Acrodyne has the switch inside the exciter, under its cover with twenty or more screws holding it down. In these exciters the aural carrier is generated at 4.5 MHz. and mixed with the visual carrier. This makes it next to impossible to directly measure the aural carrier frequency. If the visual carrier is shut off, the aural carrier disappears too. Fortunately, if you remember last months column, the FCC doesn't require direct measurement of aural carrier, it only requires that it be precisely 4.500 MHz. above the visual carrier, whatever that might be, within 1,000 Hz. Therefore, measurement of the frequency of the 4.500 MHz. aural subcarrier meets the FCC's requirements. Acrodyne conveniently provides a 4.5 MHz. sample BNC jack on the front panel of their exciter. Many of the LPTV transmitters in use today don't have convenient ways to measure their frequency. Here's some ideas I'll throw out to see what TV Technology readers can contribute. How about a design for simple filter (cavity or stub) with high enough Q to filter out the aural carrier? This would permit clean visual carrier frequency measurements. The filter could either peak the visual carrier or notch the aural. For aural intercarrier measurements, how about a simple diode detector design with a 4.5 MHz. filter following it? This would meet the FCC requirement for aural frequency measurements as the counter would directly count the 4.5 MHz. difference. Keeping things cheap, we'd need a way to tune the filter used for visual carrier measurements without a spectrum analyzer. I see two ways to do it. For a peaking type filter, insert the filter in the RF sample line to the output power metering and tune for maximum reading. For a notch filter, hook it in line with the on RF going to the 4.5 MHz. detector/filter arrangement and adjust for minimum 4.5 MHz. as read on a meter with an RF probe or simple scope (waveform or oscilloscope). I like to have a spectrum analyzer with me when checking transmitter sites (full power or LPTV) and prefer using it, rather than a separate frequency counter, for frequency measurements when possible. I like to see what carrier I'm measuring and you can't do that with a frequency counter. The frequency and amplitude of individual carriers can be measured and, in the case of the delta markers on the Tektronix, the frequency difference between the carriers can be measured as well. Over the last couple years, manufacturers have started including accurate frequency counter options within their spectrum analyzers. I've used the Tektronix 2710 and 2712 analyzers with counters and had good results with them when they had the high stability timebase option. Hewlett-Packard also offers frequency counting and high stability timebases in their 859X series of spectrum analyzers. Both Tektronix and Hewlett Packard sell analyzers with optional timebase accuracy's of up to 0.1 PPM. Tektronix had an interesting feature in the 2710 spectrum analyzer (which is probably available in current analyzers as well) that permits the internal frequency reference of the analyzer to be calibrated to a known external reference frequency. As stated last month, the FCC considers NBS stations WWV/WWVH and low frequency WWVL and WWVB the standard for frequency measurement. I found I could run a long wire antenna out of the transmitter room to the tower, hook it up to the 2710, tune in WWV at 10, 15 or 20 MHz. and calibrate my analyzer direct to the National Bureau of Standards (NBS)! I found the WWV signal had to be at least 10 dB above the noise during the deepest fades for this to work. Using a video bandwidth of 3 KHz., a frequency span of 1 KHz. per division (or less) and letting the 2710 chose the sweep speed automatically helps. Consult the manual for guidance through the menu steps. Any spectrum analyzer with a frequency counter option and sufficient input sensitivity should be able to use WWV for an accuracy check, even if it can't calibrate to it. Hook up a half wave dipole or long wire and try it out. Average the frequency over several counts, as fading and propagation delays will cause some variation in readings. Tektronix offers an AM demodulator in their 27XX line of spectrum analyzers and it's available as an option on some Hewlett Packard analyzers. With a speaker or earphone, you check the calibration of your clocks as well as the calibration of the spectrum analyzer! Satellite Uplink Power Density - Revisited... Last year in my series on RF radiation hazard measurements I mentioned an interesting National Technical Information Service (NTIS) publication called "An Efficient and Accurate Method for Calculating and Representing Power Density in the Near-Field Zone of Microwave Antennas" written by Richard L. Lewis and Allen C. Newell and sponsored by the U.S. Environmental Protection Agency. Dane Ericksen, consulting engineer at Hammett and Edison, uncovered this publication when preparing an RF Hazard study for our uplink in Hialeah, Florida. Because part of a road was within a dish diameter of the uplink antenna, the Office of Science and Technology Bulletin 65 formulas couldn't be used, since they didn't consider off-axis exposure closer than a dish diameter to the uplink main beam. You may recall from my previous column that calculation of power densities in the near-field zone was difficult because of the interaction of RF energy reflected from different parts of the dish with different phases. The complexity is obvious if you refer to Figure 2 from the NTIS paper. Note that all the distances are relative to wavelength. This nomograph is valid for dish diameters greater than 30 wavelengths. For C-band uplinks, that is a diameter of approximately 1.5 meters. For Ku band uplinks, the valid dish diameters are those greater than 65 centimeters. Here's an example using the Ku SNG truck we discussed three months ago. As you recall, it had a 2.4 meter dish and a maximum power to the antenna of 500 watts. Since 2.4 meters is over 100 times the wavelength of a Ku frequency of 14.25 GHz., the NTIS nomograph is suitable. The NTIS paper gives a formula for calculating the maximum power density for 1 watt of antenna input power. It is: Power Density = 38.6 - 20 log D D is the diameter of the dish in centimeters and the log is to base ten. The result is expressed in dBm/cm^2. Substituting 240 centimeters for D, the power density is -9 dBm/cm^2, for one watt input. Our 500 watts is 27 dB greater than one watt (using conventional dB conversion formulas), so the resulting maximum power density is -9 dBm/cm^2 + 27 dB or 18 dBm/cm^2 or, converting to it to the units used in the ANSI standard, 63 mW/cm^2. This is actually higher than the near field density calculated using OST-65 formulas, which we found to be 26.5 mW/cm^2. The ANSI limit for power density at this frequency is 5 mW/cm^2 or 7.0 dBm/cm^2. Our power density, using the NTIS formula, is 11 dB above this. Notice Figure 2 has lines with various dB values written on them. This is the relative power distribution with respect to the maximum we calculated. In our case, the closest line to -11 dB (which we need to be within ANSI limits) is the -12.5 dB line between the -10 and -15 dB curves. The Y axis distance refers to the distance from and perpendicular to the center of the main beam of the antenna. The Z axis distance is the distance out from the face of the reflector. Looking at the left side of the nomograph, locate where the -12.5 dB curve reaches a maximum. It is slightly beyond one half a dish diameter from the center of the dish, so it extends beyond the dish. It never goes above 0.6 dish diameters from the center of the dish. This converts to 0.6 times 2.4 meters or 1.44 meters, 0.24 meters beyond the edge of the dish, which is 1.20 meters from the center. This is approximately 9.5 inches. Therefore even though using the NTIS publication's formula results in a greater near-field power density than OST-65 formulas, the NTIS nomograph shows off axis power density in the region greater than 1.44 meters from the center of the main beam is within the ANSI limits. The distance from the reflector, shown on the bottom axis, is also a function of the dish diameter. The 0.75 mark corresponds to the diameter (2.4 meters) squared divided by the wavelength (2.11 cm at 14.25 GHz.) times 0.75, or 204 meters. These are rough calculations, but they are sufficient to show that the NTIS paper predicts off axis power densities within ANSI limits not only for the public on the ground, but operators inside the truck at all normal dish elevation angles. It is necessary to point out that on axis, OST-65 calculated the ANSI limit was exceeded up to 363 meters from the antenna, as shown in my previous column. Therefore, operators must still be careful where they point the antenna! By the way, the typesetter broke up some of the formulas in that earlier article, but they were still correct if you followed them. If you had difficulty with them, contact me and I'll generate a fresh formula list and fax it to you. If there is enough interest, I'll repeat them in a future column, hopefully in clearer fashion. Video Compression... I had an opportunity in early December to visit General Instruments (GI) outside San Diego California. I saw their Digicipher video compression system in operation and I can tell you, it works. I noticed no degradation on 4:1 compression other than some slight chroma bleeding (delay) which did not appear to be a compression artifact. It was there even during still scenes, which makes me suspect an NTSC encoder/decoder problem. I probably wouldn't have noticed it without a monitor showing the D-2 original material to compare it with. At 6:1 compression, some shimmering showed up on the parquet floor during fast moving basketball video, but I don't think most viewers would catch it. These were worst case tests with all channels loaded. Looking at an HBO test feed via satellite, GI was kind enough to let me do some tests to see how well it handled weak signals. As we moved the dish off the satellite, Digicipher continued to provide a perfect picture even after an analog signal on the same satellite had developed enough sparkles to make it unusable even for news footage. As the dish moved further off the satellite, the analog image got worse, though you could still make out what it was. The Digicipher signal started to break up in spots, then, as the signal weakened some more, disappeared. What I saw confirmed what GI's engineers said - that if you can receive a decent analog picture, you can receive a Digicipher picture. I've also talked about compression with Oak Communications (now TV-Com) who are working with Leitch and Phillips on video compression equipment and Scientific Atlanta (SA), who are working with a British company on a system. There is much discussion of compatible compression systems operating under MPEG II standards. I feel there will be a standard, even if a defacto one, for video compression. GI is hoping its early entry into the market and installed VideoCipher base will make it the defacto standard. SA and TV-Com are looking to MPEG and standard chip sets for a standard. GI is being pushed to conform with an MPEG standard. Several cable programmers and at least one broadcast network (PBS) have made a commitment to video compression. Most are targeting mid 1993 through 1994 for the conversion. I expect some sort of a standard will evolve by NAB time. I'll cover the opportunities video compression offers TV broadcasters in future columns, along with my observations on the competing systems as I have an opportunity to test them. That's it for this month. Next month look for more details on video compression techniques, a continuation of the discussion on frequency measurement with some more tips on how to check accuracy and some lessons learned from recent experiences with transmitter RF systems. I'm interested in your comments, experiences and tips - send them to me at 2265 Westwood Blvd., Suite 553, Los Angeles, CA 90064, or for fastest response, direct an EMAIL to me on CompuServe. My CIS number is 70255,460. I can also be reached via telephone intermittantly at 305-884-9664 in Miami or 818-502-5739 in Los Angeles. I've been busy, so feel free to bug me if you haven't gotten a response from earlier correspondence! ((8/95 > UPDATE! - Use dlung@gate.net for e-mail!)) Copyright (c) 1993,1995 H. Douglas Lung ALL RIGHTS RESERVED