FREQUENCY MODULATION

How it works.

FM radio was invented in 1916 by Edwin Armstrong. A few stations went on the air in the 1930s but until the mid 1960s FM was the stepchild of broadcasting. In 1958 WTSP in St. Petersburg Florida returned its FM license to the FCC and sold the transmitter and antenna to a station in Mexico. Ten years later the same station under the new call sign WLCY had to reapply for an FM license. I'm sure they weren't the only station to misread the tea leaves. For a time in the 50s there were more FM stations going off the air than were coming on. FM stereo and a growing interest in high fidelity audio was the savior of FM radio.

What is FM?

The amplitude of the output from an FM transmitter is constant. The audio signal changes the frequency. The transmitter has its steady state or unmodulated frequency. Negative excursions of the audio move it down in frequency while positive excursions of the audio move it up in frequency.

On paper this seems simple enough. All you need is some sort of voltage controlled capacitor to place in the LC circuit of an oscillator and you have an FM modulator. No big transformers, high voltages, or high powers, are required. The voltage controlled capacitor is a reactance tube or a variable capacitance diode, varycap. In practice it gets a bit more complicated.

For a transmitter's output frequency to be stable enough for broadcasting it must be crystal controlled. A crystal oscillator can be pulled by changing the capacitance across the crystal but the change is very small only about 0.001 %. Frequency deviation as it is known in FM is plus and minus 75 kc in FM broadcasting. The frequency band is from 88 to 108 Mc with a nominal center at 100 Mc. So the required frequency change is 100 % x 150 kc / 100 Mc = 0.15 %. There have been many schemes from tubes through transistors to integrated circuits to accomplish this.

Monophonic FM.

The simplest method to understand is phase modulation. Phase modulation is accomplished using the same voltage controlled capacitor which is placed in one of the tuned circuits between the crystal oscillator and the final amplifier of the transmitter. It is placed in a low power stage and makes the phase of the signal vary in accordance with the audio signal.

If a phase modulated transmitter is received on an FM receiver the overall audio frequency response rises at a rate of 6 dB per octave, 20 dB per decade as frequency increases. This produces a straight upward sloping line when plotted on a log-log graph. If the audio is put through an RC low-pass filter the response will come out flat. The filter could be placed either in the transmitter or the receiver. This is very similar to flattening out the output from a magnetic cartridge.

In practice some of the filtering is done in the transmitter and the rest in the receiver. A filter is placed in the transmitter which starts rolling off at a very low frequency and becomes flat at 2122 cycles. So the transmitter's output frequency response is flat from low audio up to 2122 cycles. At that frequency it begins to rise at the rate of 6 dB per octave. There is another low pass filter in the receiver that begins to roll off at 2122 cycles up to the maximum audio frequency. This is known as the de-emphasis network and has a time constant of 75 microseconds.

FM Detection.

There are three main types of FM detectors. The Foster-Seeley Discriminator, the Ratio detector, and the Quadruture detector. Of these the first is almost never used because it has poor AM rejection and must be preceded by very hard limiters.

The ratio detector has very good AM rejection and low distortion. It was used in most high fidelity tuners and receivers in the tube era.

The quadruture detector was developed by RCA in an attempt to get around the patents of Armstrong. The detector was used in almost all TV sets both color and monochrome. Because of the very poor sound reproduction of tube based TV sets I had never thought of this detector as high fidelity. When I had the pleasure of working on a Citation tuner I found to my surprise that it used a quadruture detector. If the Citation used it, it must have been a pretty good detector.

The Ratio Detector.

Below is the circuit of the Ratio detector used in the Stromberg-Carlson model SR-402 AM FM tuner.

Figure 7.9 Ratio Detector.

For a verbal description click here.

There are three coils inside a discriminator transformer. (Note; even though the circuit is a ratio detector the transformer goes by the name of the other detector circuit which is a discriminator. The discriminator circuit is almost never used because of its poor AM rejection. But the transformer which is identical for both circuits still carries the name.) Now, where was I? Oh yes, 3 coils. The primary coil which is in the collector of the last IF transistor or plate of the last IF tube is tuned to the center of the FM IF, 10.7 MHz. The secondary which is center-tapped is also tuned to the center frequency. The third coil known as a tertiary winding is magnetically coupled to the primary but is untuned.

Imagine for a moment that the tertiary winding has no voltage across it. The voltages at the top and bottom of the secondary are 180 degrees out of phase with each other. There will be a positive voltage at the cathode of the top diode and a negative voltage at the anode of the bottom diode. The polarity of the DC voltages are determined by the direction of the diodes not by the phase of the RF signals. These voltages will be filtered by the 330 pf capacitors. These voltages will cause current to flow through the two 10 k ohm resistors. (Let's simplify the explanation by pretending that the 2.2 k ohm resistor doesn't exist. The imbalance it introduces can be compensated for by slightly off-tuning the secondary of the transformer.) As the junction of the two 10 k ohms is grounded the voltage at the center-tap will be zero. Without the tertiary winding it would stay like that regardless of the input frequency.

The current which is induced in the tertiary winding is in phase with the current which is induced in the secondary winding. That is because both currents are being induced from the primary winding.

The tertiary winding is untuned and it has a self resonant frequency which is much higher than 10.7 MHz. Thus it appears inductive and the voltage is leading the current induced in it by 90 degrees. The bottom of this winding is essentially at IF ground potential so the voltage applied to the center-tap of the secondary leads the current by 90 degrees.

Because the secondary is tuned to resonance at 10.7 MHz the voltage from the center-tap to the top is in phase with the current and the voltage from the center-tap to the bottom is 180 degrees out of phase with the current. Let us say for the sake of argument that the RF voltages across the top half of the secondary, the bottom half of the secondary, and the tertiary winding are all equal in amplitude and equal 10 volts. This condition is shown in the phasor diagram in Figure 7.10(a).

Figure 7.10 Phasor Diagrams of Discriminator Transformer Voltages.

For a verbal description click here.

The voltages across the tertiary winding and the top half of the secondary add as to the voltages across the tertiary and the bottom half of the secondary. This addition is shown in Figure 7.10(a). The voltage at the top of the secondary is 14.14 volts at an angle of 45 degrees while the voltage at the bottom of the secondary is 14.14 volts at an angle of 135 degrees. Note that both voltages still have equal amplitude although the relative phase angle has been altered. The diodes recover only amplitude information not phase information. Therefore the DC voltages at the cathode and anode of the top and bottom diodes respectively will be equal in value and opposite in sign. (+14.14 volts at the top and -14.14 volts at the bottom). Note that the DC and audio output from the detector is taken from the center-tap of the secondary. The tertiary winding has a minute amount of reactance at audio frequencies.

Now suppose that the frequency shifts upward. The phase of the voltage at the top end of the secondary will shift negatively because the resonant secondary now appears capacitive. Voltage lags current in a capacitive circuit so the phase will shift and let's say it shifted by -45 degrees as shown in figure 7.10(b). The voltage at the top of the secondary is now 7.653 volts at an angle of 22.5 degrees. The voltage at the bottom of the secondary is 18.478 volts at an angle of 112.5 degrees.

Remember that the two 10 k ohm resistors are holding the voltages across them equal and balanced to ground. Phasors represent the peak value so the voltage between the anode of the lower diode and the center-tap is 18.478 volts with the center-tap positive with respect to the diode anode. The voltage between the upper diode cathode and the center-tap is 7.653 volts with the cathode positive with respect to the center-tap. The total voltage is 26.131 volts. Each 10 k ohm resistor will have 13.0655 volts across it. The voltage at the center-tap will be 13.0655 v - 7.653 v = 5.4125 volts. Note that for an increase in frequency the output of the detector goes positive which is desirable for tuning meters and automatic frequency control (AFC).

When the frequency goes below center by the same amount it went above center the numbers are the same but all signs are reversed.

I wonder if anyone noticed the discrepancy with the numbers. When the incoming frequency is at the exact center the total voltage across the two 10 k ohm resistors was 28.28 volts and at a frequency that will give a 45 degree phase shift the voltage was 26.131 volts. If the voltage of the tertiary winding is decreased to 1 volt the total DC is 20.1 volts at center frequency and 20.05 volts at a frequency that gives a 45 degree phase shift. Decreasing the phase shift to 30 degrees gives 20.1 v and 20.08 v respectively. Reducing the phase shift and tertiary voltage reduces the amount of recovered audio. Clearly the designers of ratio detectors have to make some compromises to obtain a practical audio level and acceptable distortion.

The Stromberg-Carlson was put on the market before any multiplex stereo stations came on the air however stereo was on the horizon. There was an extra jack on the back which was labeled detector that mirrored the main audio output. Modifying it to bring pre de-emphasis signal to this jack was very simple.

There are no large currents or voltages in this circuit which reduces the probability of component failure. Resistors and capacitors can simply fail from old age. The most likely one to fail from this cause is the 5 μf capacitor which is a low voltage electrolytic. This capacitor in conjunction with the 2.2 k ohm resistor set the AM rejection. If the electrolytic should dry out and its value be reduced the AM rejection of the receiver will be considerably poorer than when the set was new. Do not assume that if 5 μf is good that 10 μf is better. The impedance and time constant of the circuit consisting of the 2.2 k ohm resistor and 5 μf capacitor in conjunction with the Q of the transformer secondary set the best value of AM rejection. Changing the value of the capacitor will reduce the AM rejection of the detector. In this particular case the substitution of a 4.7 μf capacitor would probably be acceptable especially since 5 μf capacitors are no longer made.

Note: In diagrams of the Ratio Detector you are likely to find in other textbooks the 2.2 k ohm resistor will be split between the top and bottom of the circuit. S-B designers are taking the AGC voltage from the bottom end of the 5 μf cap. If there were a resistor between the diode and this cap audio signals would appear on the AGC line. Also there would be an additional RC time constant in the AGC which might render it unstable.

The Quadruture Detector.

The circuit for this detector is somewhat simpler because a special discriminator transformer is not required.

Figure 8.6 A Quadruture Detector.

For a verbal description click here.

As mentioned earlier this circuit is found as the detector in the Citation tuner. This detector has no output for AFC or AGC. I had no diagram of the Citation. All I did was to restring the dial cord and touch up the detector coil a bit. I don't recall if it had an AFC switch. Perhaps its designers had temperature compensated the oscillator well enough that none was required.

The internal structure of the tube used in the detector such as a 6BN6 is such that grid 3 has a negative resistance characteristic. This makes the combination of grid 3 and the tuned circuit oscillate at the resonate frequency of the LC tank circuit. When a signal is present at grid 1 of the tube the oscillator is locked to the incoming frequency. If the tank circuit is tuned to the incoming frequency the relative phase of the signal and the oscillator will be 90 degrees, hence the name of the detector.

As the frequency of the incoming signal changes the relative phase of the signal and the oscillator varies. Alterations of the phase will cause the average plate current of the tube to vary. C1 filters out the RF variations at the plate and in conjunction with the parallel combination of R2 and the plate resistance of the tube adds de-emphasis.

Both C2 and C3 must be large enough to present a low reactance to audio frequencies as well as RF.

FM Receiver Alignment.

Before attempting alignment of the IF strip be sure it needs it. If you remove the tubes for testing be sure to mark them so you can get them back in the same sockets. Small differences in tube capacitances can significantly throw off the alignment of the IF section.

Repeat after me.

NEVER ATTEMPT TO ALIGN THE IF SECTION OF AN FM RECEIVER OR TUNER!

Although the drill sergeant would never put up with you asking why, I will. The why is that the IF tuned circuits are stagger tuned. That means that each one is tuned to a slightly different frequency. The purpose is to obtain a wide band response with a flat top. The Citation tuner went one step further and designed an IF amplifier that was phase linear. That is, the phase shift across the bandwidth of the IF was a straight line. They used a very high priced swept frequency instrument to align the IF tuned circuits at the factory. The high priced instrument still exists, but apparently there is no one left who remembers how to perform the alignment. My advice is "don't try it." That said, if you have a sweep frequency generator and a scope and know how to use them and the IF is so badly out of alignment that the receiver does not function then you have nothing to lose.

Other brands of FM tuners weren't so sophisticated so if you have experience in aligning stagger tuned IF strips and it is clear that the IF needs alignment then go ahead. But if you make it worse, don't blame me.

Detector Alignment.

Unless the tuner on your bench has been in the hands of someone who tightened up all the loose screws the IF and detector really won't need much if any alignment. In every old tuner I have seen the detector was slightly miss-aligned. You shouldn't need to turn the alignment slug more than 1/2 turn and more likely it will require less than a quarter turn.

The evidence for a misaligned detector is found by tuning across a strong local station which has only very weak signals nearby. When the detector is properly aligned you will observe three distinct listening peaks, not peaks on any meter. As you approach the signal you will begin to hear audio and it may not be distorted but it may be mixed with a little noise. The tuning meter will read quite low. As the tuning meter rises the sound will grow quite distorted. As you continue tuning the distortion will clear and the tuning meter will be at its peak. The stereo light will most likely come on. As you continue to tune the sound will once again grow distorted, the tuning meter will drop, and the stereo light will go out. The distortion will clear but the audio will be a bit noisy, and the tuning meter will read quite low.

A misaligned detector will give you two peaks and it may be ambiguous as to which one is the right one. Make small changes in the discriminator transformer or quadruture coil and tune across the station until you observe the signature of a properly aligned detector described above. If a symmetrical triple peak can't be obtained it may indicate trouble in the detector, or an IF that is seriously out of alignment.

Never never never, adjust the detector for maximum noise when tuned off a station. You will adjust the detector so it won't detect FM at all but might detect AM if the limiters aren't to hard.

FM Stereo.

The standards for FM stereo were set when compatibility still meant something to government regulators. Consequently the owner of a monophonic radio or tuner could still hear the program broadcast by a stereo station although it wouldn't be in stereo. This was accomplished by adding the two stereo channels, Left and right, together to form a monaural image. This is known as the sum channel or L + R. A stereo station generates another channel which is transmitted above the range of human hearing which is known as the difference channel or L-R. What good does that do? It's like this. If you add the sum and difference channels together you get (L + R) + (L - R) = L + R + L - R = 2L. The right channel is derived by subtracting the difference channel from the sum channel like this. (L + R) - (L - R) = 2R.

The 2 multiplying the recovered L and R channels was evident in early stereo tuners as the sound became louder when switching from mono to stereo provided there was a stereo program being broadcast. Second generation designs and later compensated for this loudness change.

How They Did It.

Early stations grafted stereo onto existing transmitters. No doubt they had to be modified to improve the bandwidth of the modulator. Stations using SCA (subsidiary communications authority) already had the bandwidth.

An earlier SCA standard used a subcarrier frequency of 43 kc which covered a total bandwidth of 28 to 63 kc. The FCC must have anticipated the stereo standard which overlaps this part of the spectrum because newer SCA stations used a stereo compatible frequency of 67 kc. Before the days of satellites stations with SCA were used to transmit Musac. The receivers were owned by the stations and rented to stores and offices that subscribed to the service. For a station to make a change would have been quite costly. The SCA signal is FM. So to recover an SCA signal you need the standard FM detector designed to have a wide audio bandwidth (no de-emphasis) followed by a filter and another FM detector. The name for the overall system is FM/FM.

The difference signal is transmitted by an overall FM/AM system. That an AM signal at a frequency of 38 kc is transmitted along with the audio signal. They didn't use FM because it takes up more spectrum space than an AM signal of comparable bandwidth. Musac on SCA only had an audio bandwidth of 8 kc.

So the difference signal is sent on this AM carrier at 38 kc which for an audio bandwidth of 15 kc occupies the spectrum from 23 to 53 kc. But the problem with conventional AM is that big fat carrier that does nothing but sit there and burn up power. In FM, adding signals to the modulator cuts down on the amount of modulation that is available for the sum signal which is the main event when it comes to transmitting information. Trying to modulate a 38 kc carrier would waste a lot of bandwidth and accomplish nothing because if the carrier were removed from the transmitted signal it could be put back at the receiving end. Adding the carrier back means it must be added with exactly the proper frequency and the proper phase. If the phase is off even by a few degrees the stereo separation goes to pot.

That is the reason for the 19 kc pilot signal. It serves one practical purpose of telling the receiver that a stereo signal is being received but its main purpose is to help the receiver regenerate the 38 kc carrier. In the simplest form of stereo demodulator the 19 kc pilot is run through a full wave rectifier and then a filter to remove the harmonics. The difference signal is passed through a filter to remove the audio and pilot signals, combined with the 38 kc carrier and detected with a pair of diodes. One diode is reversed from the other which gives the opposite phase signals for addition and subtraction. These recovered signals are combined with the audio sum signal through a resistor network and there are the L and R signals. Capacitors are added to the resistor network to apply de emphasis. It was not applied to the sum signal before this point. Preemphasis is added at the transmitter to the two L and R signals before they enter the stereo multiplex modulator.

The first stereo demodulator I owned was an adapter made by Night to be added into their stereo ready tuner. I think it cost 15 or 20 dollars. It was my cheep entry to the world of FM stereo. I sucked a little power from my amplifier, modified my Stromberg-Carlson tuner to send a pre de-emphasis signal to it and I was ready when WTCX-FM signed on with stereo. The first thing they played in stereo was the first 30 seconds of Also Sprach Zarathustra. Approximately 10 years later Stanly Kubrick would discover this dramatic music and thereafter it would be known as the 2001 theme. WTCX was 10 years ahead of the rest of the world.

Block Diagram of Cheap Multiplex Adapter.

For a verbal description click here.

Home

This page last updated December 9, 2011.