Anticipating Signal Behavior

In 1998, the FCC revised--for planning purposes--its desired-to-undesired ratios for DTV-DTV interference between adjacent channels (ACI). Originally, the D/U ratio was -43 dB, but now it is -26 dB in one case and -28 dB in the other case.

In 1998, the FCC revised--for planning purposes--its desired-to-undesired ratios for DTV-DTV interference between adjacent channels (ACI). Originally, the D/U ratio was -43 dB, but now it is -26 dB in one case and -28 dB in the other case.

This change was made because the Advanced Television Technical Center brought to the FCC's attention that for ACI between DTV signals, the threshold U level for DTV reception varies with the D level. This is because the interference mechanism is nonlinear, so the behavior is nonlinear too. The DTV RF mask was also revised in 1998 because of this problem. The new mask reduced ACI by an additional 5 dB. This shouldn't be news to my readers, but it bears repeating.

This column has already reported that the nonlinear mechanism involved is the generation of third-order Intermodulation (IM3) products in all DTV transmitters. Some of these IM3--those closest to the undesired channel inside the adjacent channels--pass through the FCC RF mask filter and are radiated, and thus received by receivers tuned to either adjacent channel. These IM3 are within the desired channel so they raise the noise floor under the desired signal because DTV signals are noise-like. When the total noise in the desired channel is less than 15.2 dB below the desired signal, reception fails.

Now let's do the numbers. The maximum power in the radiated sideband splatter, aka IM3, is 44.5 dB below the adjacent channel (U) signals, but it is in the desired channel. The selectivity of DTV receivers discriminates against this co-channel noise by about 2 dB, because the -3 dB bandwidth of the DTV receiver is 5.38 MHz and signals outside this range are attenuated by 3 to 60 dB.

The co-channel noise is 46.5 dB down. So if the received U signal power is 31.3 dB (46.5 dB - 15.2 dB) above the received D signal, the desired signal cannot possibly be received. Co-channel interference, is therefore:

(click thumbnail)Fig. 1I = U - 46.5 dB.

This -31.3 dB ratio only applies where the receiver-generated noise is negligible, as shown in Table 1.

Fig. 1 plots the limit D/U over the range of D signals levels from the noise-limited minimum usable signal power of -80 dBm to -30 dBm. The curve remains flat at -31.3 dB down to D = -70.9 dBm, and then the receiver-generated noise begins to have effect. At D = -80.9 dBm, the D/U ratio is -20.9 dB, well inside the FCC limit of -26 and -28 dBm.

Thus, the service area is smaller than the noise-limited coverage expected. This is not due to receiver limitations. Remember that when the FCC published its estimates of DTV coverage in 1997, the D/U ratio was -43 dB. Therefore, things have changed!

Receiver operation below -71 dBm received power level shown in Fig. 1 will be unreliable if there is a powerful DTV signal on an adjacent channel, because the fading of the two signals will be independent unless the transmitters are co-sited. That has been established in FCC field tests.

(click thumbnail)This also means that by basing D levels on the F(50,90) propagation and U signal levels on F(50,10) propagation, some margin was built into the present protection ratios for co-sited allotments.

The steep slope of the plot in Fig. 1 below D = -71 dBm means that reception may be erratic where there is a strong DTV signal on either adjacent channel. The use of a low-noise amplifier (at the rooftop antenna) can mitigate this problem, as shown in Fig. 1. My calculations show that the minimum usable received DTV signal power would be reduced from -81 dBm to -86 dBm in a noise-limited situation, and that at the -80.9 dBm D level, the D/U could be -29.6 dB, a big improvement over -20.9 dB.


This column has reported this before, but just maybe it wasn't taken too seriously as the event was far into the future (after the NTSC sunset). So how can viewers survive this sunset of NTSC? Simply by using a low-noise preamplifier designed for DTV, not one designed for NTSC, despite advertising claims. (NTSC needs 64 dB uV/m signals, DTV needs 41 dB uV/M field strength signals--according to the FCC--so what is low noise for NTSC is high noise for DTV.)

Many amplifiers purported to be low noise and suitable for DTV have been shown to be otherwise. Data supporting this conclusion was recently reported by Evans Wetmore in IEEE Transactions on Broadcasting, June 2004, Vol. 50, No. 2, p. 200.

Last month, this column noted that extremely high D/U ratios could result well inside the station's coverage area if the signal of an adjacent channel is radiated with large beam-tilt. Those numbers are staggering. At this time, DTV is being radiated at low power by hundreds of stations, but before the NTSC sunset, I'll bet many of these will seek to maximize their DTV facility and to do that, will use beam-tilt.

Fig. 1 covers the case of one adjacent-channel signal. Many times, there are two. In that case, things get worse by 3 dB if both undesired signals are received at roughly the same power level.


Let's consider how receiver performance may make things worse. Over the range of D levels where the D/U ratio of -31.3 dB applies, the U level can be 26 or 28 dB higher. At such received U power levels, some receivers will be overloaded and the IM3 generated in the receiver will further increase the noise level under the desired signal.

What determines the level of IM3 generated by the receiver is its third-order intercept power (IP3). This term defines the signal power at which the signal and locally generated IM3 are equal. I don't expect consumer receiver manufacturers to say what their product's IP3 is, but let's see why it is an important receiver performance parameter.

(click thumbnail)
Table II takes us beyond Table I by assuming there is some receiver overloading. It assumes that the receiver-generated IM3 is equal to the total noise from Table 1. This results in a 3 dB noise penalty and a 3 dB increase in the minimum usable desired signal received power to offset the noise-floor increase. Table II provides the IP3 required of the receiver to let it operate at the stated minimum D level of Table II.

In this table, we see that the minimum receiver IP3 required varies from -11.75 dBm to +18.75 dBm. For example, a receiver with an IP3 of +3 dBm will work with one undesired adjacent channel DTV signal up to -25 dBm, but not above that level.

We also see that a receiver whose IP3 is above +20 dBm will work at U levels above -5 dBm. For those who are intrigued with this performance parameter, see Dallas Maxim Semiconductor application note "Improving Receiver Intercept Point Using Selectivity" App. Note No. 749, May 17, 2001, available from the company's Web site


A word or two of caution--the IP3 rating of a TV receiver is dynamic. That is, the IP3 may vary with the input signal levels. Conventional receivers have the RF gain automatically controlled by the signal level at the second detector (where only the D signal is present).

Thus, the undesired signal level at the mixer (where the IM3 is actually generated) is determined only by the desired signal level. If the D signal is weak, the RF amplifier will operate at its maximum gain; therefore, the undesired signal(s) may exceed the linear signal handling range of the mixer so that mixer overloading and additional IM3 result.

A wideband RF AGC circuit is much more appropriate to DTV receivers. In this case, strong undesired signals would force a reduction in RF gain thus automatically protecting the mixer from overload.

We need to know the IP3 for a range of D levels and not just let manufacturers pick a sweetspot. What needs to be defined is the safe operating range of a receiver. To determine this, we must test for IP3 over a range of D levels. In reality, we would be finding the maximum U level at which the receiver would operate for each D level tested. From such data one could calculate IP3, but much more important, the receiver's safe operating range.

IP3 = 0.5 (3*U - IM3).

Note that a 5 dB increase in receiver IP3 reduces receiver-generated IM3 by 10 dB, an excellent tradeoff for receiver designers once they understand the importance of DTV-DTV interference. This 10 dB decrease in receiver-generated IM3 would reduce that noise source to insignificance compared to the other noise sources, making this 5 dB increase in receiver IP3 with respect to the values in Table II extremely valuable to broadcasters.

Is this the end of the story? No. We must proceed to consider how DTV signals on UHF taboo channels may also generate additional noise within the desired channel.

(click thumbnail)Fig. 2
In the Aug. 18 issue of TV Technology, Fig. 2 of this column showed actual spectra of IM3 in Channels 28 and 37 due to a pair of DTV signals on Channels 31 and 34. If the viewer wanted to watch Channel 37, then these undesired signals are the n-6, n-3 channel pair. If the viewer wanted to watch Channel 28, these undesired signals are on the n+3, n+6 channel pair.

The result is always the same, the IM3 generated by strong undesired pairs of signals contributes to the noise in either desired channel. This figure is being repeated here also as Fig. 2 because this is so important.

The noise in Channels 28 and 37 is much higher than the sideband splatter around Channels 31 and 37. I hope to measure the power in each of these channels late this year, and you will read about it right here. Stay tuned.

The channel pairs discussed above are perhaps the least of our concerns because a well-designed tuner will attenuate n+/-6 by quite a few dB. The channel pairs n+/-2, 4 are more of a problem, especially the n+4, n+2 pair. My biggest concern is the n+/-1, 2 channel pairs where tuner RF selectivity will be minimal.