Charles W. Rhodes /
06.27.2007 12:00 AM
FCC Report Reveals More Potential Interference
There appears to be a great euphoria among broadcasters over telecasting to pedestrians and moving vehicles with the new technologies demonstrated at NAB2007. I wouldn’t want to upset this mood, but that development may have been overtaken by events I should brief my readers about. This month, I will report on test results by the FCC Laboratory on interference rejection of modern consumer DTV receivers. Unanticipated interference to DTV reception was discovered and reported in FCC Office of Engineering and Technology document 07-TR-1003 dated March 30.
Interference due to tuner overload was worse at a moderate desired (D) signal power (–53 dBm) than under strong D signal power of –28 dBm. It was even worse at lower D levels –68 dBm.
This report suggests that the RF automatic gain control circuits in modern receivers may have had enough pre-detection bandwidth that strong undesired (U) signals caused the RF amplifier gain to be reduced so the mixer was less subject to overloading.
Interference due to tuner overload was a problem in the past only where a receiver was near the U signal transmitter. That particular problem has been resolved, but the area of such interference is now farther out—at and beyond the D = –53 dBm contour, all the way to the noise-limited contour of –84 dBm. Fixing this problem would be much more difficult than improving RF automatic gain control.
Interference to a D signal on channel n was worse when the U signal was on n+/–2 than if on n+/–1. This was unanticipated, and the report attributed it to the wideband RF automatic gain control, which these modern receivers appear to have.
Some receivers were found susceptible to interference due to a signal on channel n+7. This too was unanticipated. I believe this interference mechanism is novel to DTV receivers. A signal on channel n+7 can affect the carrier recovery circuits of DTV receivers because it looks like single sideband noise, which increases the phase noise of the local oscillator. There are no restrictions on D/U ratios for n+7 or any other UHF taboo channel to protect from DTV-DTV interference.
The FCC Laboratory tested each receiver to find out whether the tuner was a single or double conversion. All were found to be single-conversion tuners, while the FCC based its protection factors on the measured interference rejection of the double- conversion tuner in the prototype DTV receiver tested at the Advanced Television Technology Center in 1995.
This is a glaring disconnect between consumer receiver interference rejection and that of the prototype tested in 1995. Because FCC Part 73 rules are based on that double-conversion tuner prototype, there are no D/U ratios for DTV-DTV interference. I interpret this to mean that when unlicensed transmitters are permitted to share broadcast TV spectrum, they cannot cause harmful interference if they avoid using DTV channels (n) and also avoid using the first adjacent channels n+/–1.
Harmful interference as I understand the term is due to U signals stronger than D signals by a margin larger that the FCC D/U ratio for that set of channel relationships.
The FCC Labs also found that interference from two undesired signals on n+k and n+2k where k is an integer up to at least +/–8, can cause interfere with channel n. However, because these U signals must be on what in the analog world were called “taboo channels,” such interference cannot be considered harmful because it results from third-order intermodulation products (IM3) generated in the afflicted receiver.
In other words, the interference can be blamed on poor receiver design. Furthermore it is not harmful because no FCC rule has been violated.
This is another new and unanticipated form of interference confronting broadcasters. It will become chronic when unlicensed transmitters become popular. They will become available to the public within 21 months or even less.
The FCC report noted: “Pairs of undesired signals placed on channels n+k and n+2k, where k is a positive or negative integer create an opportunity for third-order intermodulation occurring in the DTV tuner to create spectral products that fall in the desired channel n. We had anticipated paired-signal effects would be significant only at high signal levels; however, detailed measurements on one DTV receiver (chapter 11) demonstrated that such effects can constitute a dominant interference susceptibility, even at desired signal levels very near the minimum signal threshold for the TV, when such signal pairs exist.”
Certain other channels may also be subject to such interference. In my Feb. 21 column, I showed that the IM3 spread out over a number of channels. IM3 for two U signals covers the D channel and six or more other channels.
How can this possibly be? The answer is that for a signal pair well removed from channel n, for example, n+3 and n+6, these signal pairs reach the mixer and overload it, creating IM3, some of which falls in the desired channel n. When the D signal is above about –50 dBm, and the U signal pair is outside the pre-detection bandwidth of the RF automatic gain control, the RF amplifier will be operating at maximum gain, thus amplifying the U signals on n+3, and n+6 or n-3 and n-6. What is the solution?
Either the mixer must have a higher third-order intercept power or the RF selectivity ahead of the mixer must be improved, or both. The classic way to improve mixer linearity is to increase local oscillator power injected into the mixer and to design the mixer to handle this power.
In my RF test bed, I use mixers rated for local oscillator power input of +17 dBm. The total signal power input to such mixers must be below +7 dBm. Each of these U signals would have to be kept below +4 dBm at the mixer input. Now if the RF amplifier has a gain of say 15 dB, the maximum U signal power, (per signal) is limited to +4 dBm – 15 dBm = –11 dBm. This small calculation tacitly assumes the RF amplifier at maximum gain is delivering +4 dBm per each U signal and output is perfectly linear.
A LOT TO BE DESIRED
Yes, there may be some attenuation of these U signals, especially the signal on n+2k in the tracking filter of the tuner, but this just gives us a margin, not a miracle. The RF selectivity of these modern DTV receivers leaves a lot to be desired; the FCC found that even signals on n+8 and n+16 caused interference.
As for RF selectivity, last time I looked, tuners had two sets of tuned circuits—a single-tuned circuit between the antenna port and the RF amplifier, and a double-tuned circuit between the RF amplifier and the mixer. But as this column has noted, the latest thing in tuners is an integrated circuit built on a silicon substrate. Just how such a structure could support a tracking filter (three tuned circuits whose resonant frequencies track the D signal) escapes me.
So how did Zenith, more than a decade ago, design a DTV tuner with sufficient RF selectivity to protect the mixer from interfering signals about 60 dB above the D signal? They did it with a double-conversion topology and a first intermediate frequency filter at 1,200 MHz and a fixed tuned bandpass filter just wide enough for the D signal. This filter was located after the first mixer. The first mixer must have been very good, not to have suffered overloading.
In short, tuners today are built with excellent first-adjacent channel interference rejection at moderate and strong D signal levels by means of a smart form of RF automatic gain control. That takes care of the expected adjacent channel interference, but not the unexpected interference in a large part of a station’s coverage area.
As there are no D/U numbers given in he FCC rules for DTV-DTV interference, it appears that some manufacturers of tuners believe there is no need to worry about taboo channel interference in tuners that will never see analog TV signals. Meanwhile, tiny tuners with mediocre RF selectivity are popular today. Tomorrow, DTV receivers for pedestrians, which are by their nature battery operated, will have even greater problems with interference because their tuners will have very little power, they will have minuscule antennas and be very close to the ground where signals are especially weak.
All the above might be considered good news, meaning that there is bad news coming.
Reports from the Canadian Research Center and FCC Laboratories studied the effect of interference to DTV reception from one and two U signals, but not three or more signals.
CHECK OUT ALLOTMENTS
Look at the DTV allotment table for your community. You might want to note how many permanent channel allotments there are in the UHF band, and look also at thes high VHF-band allotments. High VHF is probably a sleeper.
It may be subject to unanticipated interference from other DTV signals in the high-VHF band and from the unlicensed transmitters soon to appear on the band. Two undesired signals of equal power produce a certain amount of IM3 in a given tuner. If there are three U signals of equal power, their total power is about 5 dB greater than one alone. This means that the IM3 is 15 dB greater than with just one U signal. This IM3 spreads out across three channels for one U signal, or across seven to 20 channels for two U signals, or across up to at least 17 channels in the case of three U signals.
You may not want to think about the spectrum spreading with four or more U signals. Note that I didn’t say four or more DTV signals because signals from unlicensed transmitters will behave in the same way as DTV signals, which get to the mixer of DTV tuners. Spectrum plots of IM3 for one, two and three U signals were published in the Feb. 21 issue of this magazine and should still be available at the Web site. I should soon have some spectrum plots involving four DTV signals and the IM3 they generate.
I am amazed at how calmly the broadcast industry is taking this interference problem. Just think about how many viewers believe that with DTV signals, one will not need a rooftop antenna because, don’t you know, digital technology works miracles. Well, read the FCC Labs Report. Interference due to tuner overload will exist clear out to your noise limited contour! This is your problem.