The FCC Laboratory report on the interference rejection capabilities of modern DTV receivers dated March 31, brought to my attention the fact that many viewers live in relatively weak signal areas.
Eleven percent of the noise-limited coverage area has a 1 dB or less signal level margin. Twenty-nine percent of the coverage area has less than a 3 dB margin and fully 84 percent of the coverage area has a received DTV signal power of less than –68 dBm, which the ATSC calls a “weak DTV” signal.
This is by area, not population, but still, it focuses upon the fact that the FCC Lab found the interference susceptibility of receivers is worse where the desired signal is weak, not strong. This interference is primarily due to tuner overloading by the undesired (U) signal(s).
This column has long been concerned with interference due to tuner overloading, and so I was concerned with the problem of strong signals, which can only exist near the transmitter. I was therefore as surprised as anyone with the finding that tuners were being overloaded by what the ATSC describes as a moderate DTV signal level of –53 dBm, and I am alarmed to learn that the problem only gets worse below D = –53 dBm. This means that such interference will be much more widespread than was previously imagined.
(click thumbnail)Fig 1: Median U (dBm) D = -68 dBmAbove about D = –50 dBm, the tuner’s RF amplifier gain is reduced by the RF AGC circuit, so at D = –68 dB, any signal two or more channels offset from the desired channel and which is 30 dB stronger is being amplified before the mixer. This is what causes cross-modulation (X-M) in the tuner. X-M looks like noise in the desired channel.
If there is only one such signal present it can generate X-M. It can also generate third order intermodulation products (IM3) between its own sidebands, which is the mechanism by which adjacent channel interference (DTV-DTV) occurs. When there are two or more strong undesired signals, they can generate IM3, which spread out over many channels. Alas, in many communities, there are, or soon will be multiple DTV signals, some of which are on channels of the form n+k and n+2k (k is an integer, either positive or negative).
TABOO CHANNELS
As this column has reported, IM3 products generated by signals on such channel pairs fall in the desired channel (D) where it increases the in-channel noise. As the spectrum of IM3 products generated by one DTV signal is three channels wide, three channels are affected by just one undesired DTV signal. In fact with an undesired signal on say, Channel 34, Channels 33 and 35 will each have equal noise, which is only 6 dB less than in Channel 30.
(click thumbnail)Fig 2: Median U for Channel Pairs at D = -68 dBmAnother example is in n+3 and n+6, there will be noise in channels between n–3 and n+9, surrounded by noise 6 dB lower on channels n–4, n–2, and n+8 and n+10. Any or all of these channels may be jammed by IM3 generated by the presence of DTV signals on just n+3 and n+6. This column has shown actual spectra of such IM3 products many times so I won’t repeat them here. Any of these channel pairs consists of frequencies, which were considered taboo channels for NTSC signals.
Fig. 1 was prepared from the data in the FCC Lab report. The desired signal power is –68 dBm. The plot shows the undesired signal powers at which reception fails as a function of the channel offset between the U and D signals for the median receiver tested.
What this means is that where your DTV signal is received at –68 dBm, 50 percent of the receivers will be unable to receive your signal if there is one undesired signal above the indicated power for any of these channel offsets. The only difference between channel offsets is the RF selectivity of the receiver. In Fig. 1, a signal more than nine channels from the desired signal would have to be extremely strong to cause interference. That is fine for DTV signals, but when the folks in the same apartment building activate their unlicensed device (UD) it may cause interference to many DTV signals in many homes. When anyone in the household uses a UD, TV reception will be jammed in that home. The total noise (X-M and IM3) and tuner noise in your channel will be 15.2 dB below your signal power at threshold of failure—noise is noise.
Fig. 2 shows the FCC measured median U power per U signal for channel pairs of the form n+k, and n+2k at which DTV reception fails for D = –68 dBm. From Fig. 2, you see the combined effect of X-M and IM3 for channel pairs n+k and n+2k for k values from –5 to +5. Fig. 1 and 2 are to the same scale.
In comparing both figures, you will note that the power per channel of each of these U channel pairs at which reception fails is much less than the U power for either member of the channel pair by itself. Note also that the U channel power at which reception fails is well within the range of received DTV signal power, that is, interference results when the U signal power per signal is –26 dBm to –32 dBm. The channel pairs all involve what were UHF taboo channels in NTSC. These are unregulated regarding power, as the FCC believed, back in 1997, that DTV reception was inherently immune to interference from other DTV signals on taboo channels. That might have been true for the double conversion tuner Zenith provided in 1995 for testing of its 8-VSB modulation signal, but the FCC has recently learned that there are no double conversion tuners in the DTV marketplace today. The RF selectivity of DTV tuners shown in Figs. 1 and 2 is probably much better than what will be provided by a tuner-on-a-chip, which I fear, will be widely used in by 2009.
GOOD NEWS, BAD NEWS
If the FCC were to establish protection ratios for DTV-DTV interference, a good place to start would be Fig. 2. However, this is only half the story. Below D = –68 dBm, in the vast outlying area of your station’s noise-limited coverage, different U limits must be established. The FCC found that the threshold U power is even lower below D = –68 dBm. For example, U for n+6 = –4.9 dBm @ D = –68 dBm, while at Dmin + 3 dB, U is limited to –8.9 dBm. The FCC Labs did not report any tests at Dmin + 3 dB with pairs of undesired signals. Meeting these requirements is clearly more restrictive than would be suggested by data in Fig. 2, where D = –68 dBm; but it would extend DTV coverage almost to the noise-limited contour, which would not be the case were new protection ratios based on D = –68 dBm data.
Without knowing the U limits for the channel pairs below D = –68 dB, it would only be guess work. More testing with these same receivers at the FCC Lab would be needed before this matter could be properly resolved.
Somewhere above D = –53 dBm, the RF AGC in receivers starts to reduce the gain of the RF amplifier, which in turn, reduces the U signal power at the mixer. This was observed in all DTV receivers tested by the FCC. At D = –28 dBm, the maximum expected DTV signal power into receivers within a few miles of the transmitter (–8 dBm according to the ATSC), does not cause interference. That is the good news. The bad news is that nearby unlicensed transmitters next door, or even on the same block or apartment house may overload nearby DTV receivers anywhere within the station’s noise-limited contour.
I believe the interference problems of spectrum sharing between UD and DTV cannot be solved along the lines the FCC has proposed. I also do not believe that signal sensing to determine which channels are vacant is practical. Think of all the IM3 from pairs of DTV signals overloading the frequency agile front-ends (tuners) of these signal-sensing devices. These IM3, as shown in Fig. 2 will be a much bigger problem for signal sensing, than tuner noise around –99 dBm. I hope the FCC Lab will test the prototype signal-sensing hardware with a pair of strong DTV signals on channels of the form n+k, n+2k over a range of k from at least –8 to +8, while n is the channel being examined by the prototype sensing system. Remember that each UD will be required to include such sensing technology. These will be consumer appliances to be mass marketed to the general public at a price point at which consumers in large numbers will buy them. So, I expect to find consumer DTV tuners in these signal sensing devices.
A TECHNICAL SOLUTION?
Rather than sandwich UD and DTV signals into the remaining broadcast spectrum (Channels 2–51 after Feb. 17, 2009) I favor some partitioning of spectrum, and to do that would mean repacking DTV channel allotments. This drastic step may be the only way to save terrestrial broadcasting from hopeless interference from unlicensed devices spread across the broadcast spectrum and across the coverage area of DTV stations. This is not a matter of if, but when repacking must be done, I believe.
