slightly reduced deviation (modulation index M2), is applied to the other terminal of the mixer at point B. The
mixer output will be the sum and difference frequencies of the two waves. The difference frequency is selected
by an IF filter. This difference frequency at point C will have a modulation index M3 which will be the
difference of the modulation indexes of the two FM waves (M3 = M1 - M2). This resulting waveform with
reduced deviation may be passed through a filter whose bandwidth is approximately 3M/M1 times that required
of the large deviation wave. The signal is then frequency-detected in a circuit such as a discriminator. The
second FM wave (M2) can actually be derived by feeding the output signal of the frequency discriminator
through a low-pass filter to frequency-deviate a VCO. The larger the gain the feedback loop, the more the input
deviation is reduced in the IF.
4. To explain the threshold improvement gained by using FMFB, the threshold mechanism of conventional
FM will be detailed. The threshold occurs in a conventional FM receiver when the random-noise peaks exceed
the carrier amplitude prior to the frequency detector (discriminator) for a sufficient percentage of time. Each
time a noise peak exceeds the carrier amplitude, an impulse in amplitude (a spike) appears at the frequency
discriminator output. This noise appears as a random sequence of spikes which are heard as sharp pops in an
audio system or seen as spots on a TV screen. For voice, data, or TV channel, operation below threshold is
5. The FMFB, then, must reduce these noise spikes, and thereby reduce the threshold. This is accomplished by
feeding the detected signal and noise back to the VCO. The noise that is fed back will reduce the incoming
noise, thereby reducing the threshold of the system and, at the same time, the improvement factor for high-level
S/Ns will remain that of the transmitted wave. These are significant improvements in terms of equipment. To
demonstrate this action, let's take two examples.
a. First, a 100,000-to-1 (50 dB) S/N is desired at the output or baseband signal. By using conventional FM
(Figure 3-3), a minimum C/N of 560-to-1 (27.5 dB) is required; for FMFB, only 70-to-1 (18.5 dB) is needed.
This means that by using FMFB, we are able to reduce the carrier power by 9 dB, which is a factor of 8.
b. As a second example, take a more typical baseband S/N of 3,000-to-1 (35 dB). Again referring to
Figure 3-3, we see that the carrier power can be reduced by 6.6 dB, or a factor of 4.6. Realize that a decrease of
6 dB is equivalent to halving the receiving antenna diameter or doubling the slant range.
6. Shown in Figure 3-3 are the modulation indexes needed to design an optimum system. Since bandwidth is
directly proportional to the modulation index, the RF bandwidth can be computed. For the two examples just
presented, the modulation indexes are between two and three times larger for FMFB; hence, we can expect the
RF bandwidth for optimum FMFB to be two to three times larger than conventional FM.