The Challenge of Channel Election

My research into DTV-DTV interference resulted in some interesting spectrum plots concerning third-order intermodulation (IM3) products as a source of such interference. To review, undesired signals on certain pairs of channels can generate IM3 that falls into a third or victim channel (n). The (n) IM3 is noise, and the signal-to-noise ratio it reduces may be yours. To recap, the worst channel pairs are n-1, n-2 and n+1, n+2 (shown in Fig. 1). This is because:

These channels are so close in frequency to your channel (n) that the meager RF selectivity between the tuner's antenna port and the RF input to the mixer hardly attenuates the adjacent channel's undesired signal, and only attenuates the n+/-2 channel by a few dB.

(click thumbnail)Fig. 1: Spectrum plot of pairs of undesired DTV signals on n-1, n-2 and n+1, n+2 that may generate IM3 in channels n-3 and (n). Black trace from both n-2, and n-1 is simultaneously present; green trace was made with only one DTV signal present. Note that when both are present, the IM3 is much higher than that due to only one signal. IM3 varies with the cube of the input power, in this case, a 9 dB increase in IM3 from a 3 dB increase in signal power.

(click thumbnail)Fig. 2: Spectrum plot of pairs of undesired DTV signals on n-2, n-4 and n+2, n+4. Black trace was made with both signals present; green trace was made with only one signal present. Note that the maximum IM3 are on channels n-4 and (n). An IM3 from all six channel pairs falls into the "victim of victims" channel (n).

(click thumbnail)Fig. 3: Spectrum plot of pairs of undesired DTV signals on n-3, n-6 and n+3, n+6. Note that IM3 generated in the receiver by one undesired signal is increased by the presence of any additional signals of comparable strength at the mixer. Significant IM also exists in other channels, all due to overloading in the receiver.
There is another problem with n+/-1. The radiated sideband splatter from that transmitter also contributes noise in your channel (n). That noise is U -46.5 dB. As U could be up to 0 dBm, this noise can be up to -46.5 dBm. If your signal power is below -31.3 dBm, adjacent-channel interference alone blocks reception.

But that is the good news. Let's say that receivers commonly generate as much IM3 as the received splatter. The noise is now -43.5 dBm, so your signal must be stronger than U -28 dBm.

What if you should be so lucky that you have two adjacent channels? With such luck, your DTV signal could only be received where it is above -25 dBm at the mixer RF port, assuming the RF amplifier gain is one or less. Your field strength would then have to exceed 106 dB uV/m.

The FCC allows maximization to the extent that your competitors can hand you a D/U of -26 dB for n+1 DTV-DTV interference and -28 dB for n-1 DTV-DTV interference. That interference starts with the their radiated sideband splatter. There is not much headroom left when you have neighbors on adjacent channels. It can quickly vanish when folks move into n+/-2.

So, for those of you choosing your digital future channel, think about the digital neighborhood before filing.

Fig. 2 shows the n-4, n-2 and n+2, n+4 pairs. These neighbors aren't quite as noisy as the pairs shown in Fig. 1. Receiver RF selectivity may not be quite good enough to reject the n-4 signal, so the n-2, n-4 pair may adversely affect your service area.

The n+2, n+4 pair in your neighborhood can ruin things unless the FCC mandates adequate RF selectivity for all receivers. I would not build my business plan on that wish.

A signal on n+4 has two ways to attack you. The second harmonic of the LO frequency is 88 MHz above the desired channel center frequency. This beats with the second harmonic (generated in receiver tuners when overloaded) of any signal on undesired n+4, and the beat falls in the 44 MHz IF, where it passes on to the detector and ruins reception. In a test we did at the ATTC, we found that NTSC reception was 10 dB more sensitive to an undesired NTSC signal on n+4 than one on n-4. I expect the same for DTV-DTV interference, and soon will make that test.

Signals on n+2 and n+4 produce IM3 in channel (n) as shown in Fig. 2.

In my view, DTV on channel n-2 is incompatible with DTV on n-1 or n-4 in the same community. This also holds true for n+2, where n+1 or n+4 is allotted as just explained.

This brings us to the n-3, n-6 and the n+3, n+6 channel pairs as shown in Fig. 3. I believe these are the least dangerous channel pairs because even poor RF selectivity in the tuner should attenuate the n+/-6 undesired signal so only the n+/-3 signal is strong enough to challenge your broadcast.

IM3 is generated by the second harmonic of one undesired frequency (Fa) beating (mixing) with the fundamental of another undesired frequency (Fb). Fb>Fa. Let Fa = Channel 31, whose center frequency is 575 MHz, and let Fb = Channel 34 with a center frequency of 593 MHz.

There are four IM3s, but only two count as interferers. Fig. 3 shows the n-3, n-6 and n+3, n+6 channel pairs.

2Fa - Fb = 1150 MHz - 593 MHz = 557 MHz, center of Channel 28

2Fb - Fa = 1186 MHz - 575 MHz = 611 MHz, center of Channel 37

Note that two desired channels exist for each of the channel pairs of undesired channels. The IM3 resulting from the sum terms does not fall into the broadcast spectrum:

2Fa + Fb = 1743 MHz, well above the UHF broadcast band.

2Fb + Fa = 1761 MHz, also well above the UHF broadcast band.

This brings us to another great unknown: What will the RF selectivity be of future receivers built under FCC's DTV tuner mandate? In some cases, RF selectivity may be nothing more than switchable bandpass filters, Channels 2-6, or 7-13 or 470-806 MHz.

A better solution is a tracking, tunable RF filter, but the problems of such filters include the physical size of the inductors, the costs of factory alignment, the Q factor of the resonant circuits and the small fractional bandwidth desired at UHF (1 percent).

Zenith understood these problems well enough to use a double-conversion tuner, but it too has problems with noise and cost.

I understand that most DTV receivers have single conversion tuners. Will DTV tuners have tracking or bandpass filters next year?

If manufacturers of consumer receivers do not understand and appreciate the need for voluntarily providing good RF selectivity, then all channel pairs, including n-4 & n-8 and n+4 & n+8--will prove equally disastrous to terrestrial broadcasters as the "killers" n+2, n+4 & n-1 and n-2 & n+1, n+2.

DTV tuners are going into wireless phones in a year or so. Where will they put the tracking filters?By the time you read this, I will have returned from CES and the International Conference on Consumer Electronics (ICCE), both in Las Vegas. I will present a technical paper I co-authored with Gary Sgrignoli at ICCE. The purpose of our presentation is to attempt to alert consumer electronics manufacturers about our concerns over inadequate interference rejection capabilities of DTV receivers.

I will also be looking around at the CES for products with the CEA Smart Antenna Interface, No. 909, and for smart antennas that can be controlled via that standard. As always, stay tuned.