Amazon.com Widgets

Guide to Linear Electronics: Radio receiver circuitry



Home | Glossary | Books | Links/Resources
EMC Testing | Environmental Testing | Vibration Testing



Introduction:

Although I have used the term Radio in the section heading, the basic techniques and systems which are used for radio reception are much the same as those employed in television, Radar, radio astronomy, and other systems used to receive high frequency electro magnetic radiations, apart from any modifications to the receiver which may be needed to meet the specific requirements of the operating frequency, or the signal bandwidth.

For me, and perhaps for many others of my generation, radio circuitry has always enjoyed a special degree of affection because it was my original introduction to the whole field of electronics, and I still think that there are two uniquely enthralling pleasures of revelation which await the practically minded- that of watching a black and white print image, of ones own, appear, for the first time, in the developing dish in a photographic darkroom, and that of hearing ones first short- wave radio signals, on headphones, on a home-built radio receiver. Of these two pleasures, that of radio has the advantage that, with care, all of the components are re-usable. One does not need to throw away ones failures, one can dismantle them, instead.

Radio receiver circuitry

Tuned radio frequency (TRF) systems

Demodulation methods

In most radio signals, the amplitude of the high frequency, (HF or RF), transmitted waveform, usually called the carrier, is modulated, (caused to vary), as a means of carrying the lower frequency information or program content of the transmission. With a high frequency signal, the fluctuations in the amplitude of the incoming signal would be far outside the audible range, and undetectable. In order to allow the receiver to convert such an amplitude modulated (AM) RF voltage into an audible or measurable signal, the process known as demodulation or detection is used, most commonly by simply rectifying the incoming RF waveform, shown in FIG1a, to give a waveform of the kind shown in FIG1b. If this voltage waveform could be averaged, as shown in FIG1c, the result would be a true replica of the original waveform used to modulate the carrier. Unfortunately, the easiest technique for doing this, just to use a resistor/capacitor/diode circuit, of the kind shown in FIG2, leads to a substantial distortion in the recovered modulation waveform, as shown in FIG Id, unless the CR discharge time constant is kept short in relation to the highest modulation frequency employed, and this lowers the demodulation efficiency.

FIG1 Action of diode demodulator

FIG2 Simple diode demodulator circuit

A rather better way of extracting the modulation component from such a rectified carrier signal is to interpose a low-pass filter between the rectifier and the signal output point, as shown in FIG3, but even this cannot entirely solve the problem of AM demodulation distortion, particularly at lower carrier, and higher modulation frequencies, so demodulation distortion remains a characteristic of all AM receivers.

The second problem with a demodulator system based on a simple diode rectifier, of the kind shown in FIG2, is that low applied voltages may either not cause the diode to conduct at all, or, if a forward bias is deliberately applied, to ensure that they do, or if a semiconductor junction is employed which does have some conduction at zero applied voltage, as shown in FIG4, there will be very little difference between the diode forward and reverse direction conduction characteristics for very small input signal voltages.

This leads to a very low demodulation efficiency, a characteristic which it shares with the other 'slope demodulator' circuits shown in FIGs 5 and 6, which rely for their rectification ability on the curvature of the Vy/C, or Vg//d, characteristics of the semi conductor device.

FIG3 Improved diode detector circuit

FIG4 Conduction characteristics of forward biased diode junction

FIG5

FIG6

The so-called 'grid leak' detector circuit, shown for a triode valve in FIG7, exploits the fact that as the positive-going peaks of the RF waveform applied to the valve grid approach zero potential, the flow of grid current into C1 will negatively bias the grid of the valve, which will push the mean DC level of the input signal down the/a/Vg slope, as shown in FIG8.

This makes the modulation envelope lop-sided so that the average anode current of the valve fluctuates in sympathy with the carrier modulation level. This scheme could also be used with a junction FET, but, in this case, the FET gate would need to be forward biased to the point at which diode type gate conduction was about to occur. Traditionally, the grid leak resistor would be connected across Cl5 to minimize the resistive damping of the input tuned circuit, (L2ICV{), but since Rx would usually be a megohm or greater in value, the damping due to this cause would be negligible, and Rx could equally well be taken to the OV line. This type of demodulator shares with the diode detector the problem of modulation distortion due to the finite R\ICX discharge time constant, as well as the problem that, for small signal levels, the demodulation efficiency is exceedingly poor.

FIG7 Valve grid leak demodulator circuit

Clearly, what is needed is some way of increasing the size of the incoming RF signal to a level at which the demodulation efficiency from any of these systems reaches a useful level.

Effect of the Q of the input tuned circuit Some assistance in increasing the magnitude of the applied signal will be given by the action of the Q, (circuit magnification factor) of the input tuned circuit, interposed between the aerial and the demodulator, in FIG9, but, to prevent the low input impedance of a demodulator diode, or that of the base/emitter junction of the transistor, from lowering the Q too much, it will be necessary to take the output from the tuned circuit from a low tapping point on the inductor, (L2), which will also lower the available output volt age.

FIG8

FIG9 Simple diode detector radio receiver circuit.

Figure 14·10 Radio receiver circuit based on FET slope detector

The FET slope demodulator of FIG6 has a very high input impedance, and can therefore be connected across the whole of L2, as shown in FIG10, with a significant gain in demodulator output.

The normal engineering solution to this basic problem -- that all AM demodulator circuits require an adequate RF input for satisfactory performance -- is to provide some pre-demodulation RF signal amplification, using the type of circuitry examined in Section 12. Simple receiver layouts of this kind, using one or more tuned RF amplifier stages preceding the demodulator, such as that shown in a contemporary form in the circuit layout of FIG11, would be described as a TRF (tuned radio frequency) receiver, and systems of this kind formed the bulk of early radio designs.

FIG11 Radio receiver circuit using cascode connected FET RF amplifier stage

Problems with selectivity:

There are a number of fundamental difficulties with this approach, of which the major one is that of pro viding adequate selectivity in respect of adjacent frequency signals. As seen in Section 11, a single tuned circuit gives a response curve of the kind shown in FIG12, in which the -6dB voltage gain points occur at frequencies separated from the resonant frequency by +l-fJQ. The relative voltage output from a parallel resonant tuned circuit at frequencies on either side of resonance can be calculated, approximately, for values of Q in excess of 10, from the formula given below (Dobbie and Builder, Radio Designers Hand book, fourth edition, pp. 415-416):

... where EJE is the ratio of output voltage at resonance, f_0, to that at a frequency, f, on either side of resonance.

This shows that the -12dB gain points occur at approximately +/-2/0/ß, the -18dB points at+/- 4fJQ9 and the -24dB points at +/-S/JQ. (Note. This type of calculation is only approximate, and shows a symmetrical gain/frequency curve, whereas, in reality, the cut-off characteristic must be somewhat steeper on the lower frequency side of resonance than the higher frequency side.)

Relative transmission (dB)

FIG12 Response curve of single loosely coupled LC resonant circuit

Using the above data, based on an operating frequency of 200kHz, and a Q value of 100 (a good average value for a well made tuned circuit), the-6dB points would occur at 198kHz and 202kHz -- a 4kHz passband: the -12dB points at 196kHz and 204kHz --an 8kHz passband: and the -18dB points at 192kHz and 208kHz -- a 16kHz passband. If we accept that an 8:1 (18dB), rejection ratio for an adjacent signal is acceptable -- it’s clearly not very good -- the ability of a receiver with a single input tuned circuit, having a Q value of 100, to provide adequate selectivity at a 200kHz operating frequency would require that adjacent stations were not closer than 16kHz in their carrier frequencies. For an operating frequency of 1MHz, these pass-bandwidth figures will be five times as great, and clearly not acceptable for the degree of crowding which exists on any broadcast band.

To improve the selectivity, one must either have more tuned circuits before the demodulator stage, or one must increase the Q of the circuits. Ganged (mechanically coupled) tuning capacitors are available to allow simultaneous tuning of more than one tuned circuit, which will help to solve the problem of selectivity, but while 2-gang tuning capacitors are inexpensive and easy to buy, 3-gang ones are costly and relatively scarce, and 4-gang types are very seldom found, and would be very dear even if they were available, so the number of variable capacitor tuned RF stages which the designer could employ in a TRF type receiver would be mainly limited by the availability of hardware.

Regeneration or reaction:

An ingenious solution to the problem of low Q values, and poor selectivity, which is particularly acute on the 'short wave' bands (approximately 2.5MHz-30MHz), is to use 'regeneration' or 'reaction'. In this technique, some energy is fed back from the output of the amplifying stage into the input circuit, in an identical manner to that employed in an LC type HF oscillator.

However, if the amount of energy fed back is carefully adjusted, so that the circuit does not quite break into oscillation, it’s possible to use the feedback signal to very nearly completely cancel the energy losses in the tuned circuit. With care, and delicate adjustment, operating Q values for the input tuned circuit as high as 50,000 can be obtained by this means. For a receiver operating at 20MHz, this would give a -18dB pass band of 3.2kHz, which would be adequate to separate most wanted signals. This arrangement would also have the great advantage that the incoming aerial signal would also be magnified by the Q factor, so that a 50µF aerial signal would become a 2.5V signal at the demodulator input -- a level at which efficient demodulation would occur.

Because of the simplicity and efficiency of such circuits, designs for one valve short-wave receivers using regeneration, such as that shown in FIG13, were very common in the amateur magazines during the 1930s and 1940s. A comparable contemporary design, using two transistors to achieve a similar audio output power, is shown in FIG14.

Super-regenerative receiver systems

FIG13 One-valve receiver regeneration

FIG14 Simple AM receiver using re-regeneration -- usable over the frequency range 150kHz-30MHz.

While, with careful adjustment, regeneration provides a very simple and effective method of increasing receiver sensitivity and selectivity, it suffers from the problem that no single setting of the regeneration control will be satisfactory over the whole of the desired tuning range. This is because any alteration in the resonant frequency of the tuned circuit will require an alteration in the L:C ratio in that circuit. But the Q of the circuit is defined by the relationship …so altering the tuning will inevitably alter the Q value of the circuit, and, in consequence, the amount of positive feedback needed to cancel the circuit losses.

This means that the user of the receiver needs to fiddle continuously with his regeneration control knob as he tunes his receiver, to keep on the correct side of the knife-edge between stable operation, and an audible 'howl', which is irritating both to the user and also to others within range of the RF signal inadvertently radiated by his oscillating receiver.

An ingenious solution to this problem, again introduced by Edwin Armstrong, in 1921, is the 'super regenerative' system. This takes advantage of the fact that even when a sufficient amount of positive feed back is applied, around an amplifier stage containing a tuned circuit, to cause it to break into oscillation, there is a finite time lapse between the moment of applying the feedback, and the onset of oscillation.

Armstrong's approach was therefore to apply an external 'quench' signal periodically to the regenerative circuit, to pull its operating point back from the point at which continuous oscillation will occur. This quench signal can be any convenient method of momentarily reducing either the stage gain or the amount of feedback -- even a grid-leak/capacitor combination, whose values have chosen to make the oscillator 'squeg', will work.

However, for simplicity and reliability of operation in a super-regen receiver, the best approach is usually just to superimpose a suitable amplitude sine-wave or square-wave on the HT supply line to the regenerative stage. In order to prevent the whistle due to the quench waveform from blotting out the wanted signal, the quench signal will usually be chosen to be at some ultrasonic frequency, which will facilitate its separation, by simple filtering, from the wanted signal.

Experience has shown that the best receiver sensitivity is obtained with quench frequencies just above the audible range. A simple modification to the re generative receiver circuit shown in FIG14, using an externally applied square-wave, of 1-2V amplitude, at 25kHz, to convert it into a super-regen. receiver, is shown in FIG15. Any suitable square-wave generator, of the kinds shown in Section 13, will work.

Although super-regen receivers are simple and effective, especially at higher signal frequencies where other receiver systems are less efficient -- and are widely used in 'Citizens Band' (CB) receivers for this reason -- they suffer from several snags which have prevented their more widespread adoption. These are the annoyingly loud inter-station noise (due, it’s said, to randomly occurring thermal noise impulses in the input circuit being amplified by the circuit up to full output voltage level), wide-band noise being radiated from the receiver in the neighborhood of its tuned frequency, and the relatively poor input signal selectivity.

Quench signal:

FIG15 Modification to convert circuit of FIG14 to super-regenerative operation.

Effect of selectivity on the received signal

So far, all our calculations on the ability of a receiver to reject an unwanted adjacent signal have failed to consider the effect of this selectivity on the sidebands of an AM broadcast signal, which are themselves composed of the sum and difference outputs of the modulation and carrier frequencies. Too narrow a reception bandwidth in the receiver will attenuate the higher modulation frequencies, and this is an inevitable effect with any system based on either a single LC parallel resonant tuned circuit, or a series of these connected in cascade to improve the selectivity, simply because the output voltage from such a tuned circuit will fall immediately the signal frequency moves away from the frequency of resonance.

Tuning successive stages, in a cascaded series of tuned circuits, to slightly different frequencies will increase the receiver bandwidth, but this is a difficult practice to carry out well. A much better approach, which is very widely employed, is to use pairs of tuned circuits which are electrically coupled to one another, as shown, for example, in the inductively coupled layout of FIG16.

FIG16 Bandpass-coupled pair of tuned circuits

Bandpass coupled tuned circuits:

Such bandpass-coupled pairs of resonant LC circuits, of which a number of possible layouts were shown in Section 11, have the great advantage that, for appropriate values of Q and coupling factor, the pass-band has a portion, adjacent to the resonant frequency, at which the signal output will be substantially constant, as shown for a critically coupled bandpass system in FIG17. Circuits of this kind can be cascaded, to improve the steepness of the skirt frequency response, without impairing the flatness of the portion of the response curve on either side of resonance, and this level response portion can be tailored, by choice of Q and coupling factor, to provide almost any desired pass-band.

FIG17 Frequency response curve of critically coupled bandpass pair of tuned circuits.

However, these factors will depend on the operating frequency chosen, and this factor, quite apart from the impossibility of obtaining 4-, 6-, or 8-gang tuning capacitors, precludes the use of bandpass-coupled RF stages in simple TRF receivers, and has led to the almost universal adoption of the 'superhet' system.

The supersonic heterodyne, or superhet system As with so many other useful ideas in the field of radio, this was introduced by Armstrong, in 1921, and pro vides an effective solution to the problem of selectivity, while avoiding the difficulty of trying simultaneously to tune a number of critically coupled bandpass-coupled circuits. The basic idea of the superhet receiver is illustrated in FIG18. In this arrangement, the incoming signal is converted, in a frequency changer or mixer stage, into a signal at some other fixed, intermediate, frequency at which the bulk of the pre-demodulator, RF, amplification is per formed. This frequency conversion is carried out by combining the incoming signal with the output of a local oscillator in a stage having a sufficient degree of nonlinearity as an amplifier to cause the generation of sum and difference products as a result of the interaction between the input signals. By the choice of a suitable local oscillator frequency, a composite output signal, carrying all the modulation information present on the incoming aerial signal, can be generated, and can be placed at any desired frequency. This allows the intermediate frequency to be chosen to lie at any convenient part of the RF spectrum.

FIG18 Basic layout of superhet receiver ---Mixer or frequency changer; RF IF amplifier demodulator AF output; Local; oscillator.

For example, for an incoming signal frequency of 1550kHz, and a local oscillator frequency of 1000kHz, sum and difference frequencies of 550kHz and 2550kHz will be generated, and will be present in the mixer stage output, along with the original 1550 and 1000kHz input frequencies.

For low to medium frequency receivers, such as are widely used for the medium- and long-wave broadcast bands, the IF frequency normally chosen is 455 or 465kHz, and this frequency band is kept free of most commercial broadcast transmissions. For earlier de signs of short-wave receivers, 1.6MHz, a frequency right at the upper end of the medium-wave band, and not, at that time commercially exploited, was commonly chosen as an IF frequency, but in more modem designs, 45MHz or even higher frequencies may be used. Since it’s desirable to avoid direct aerial circuit break-through at the chosen intermediate frequency, the frequencies adopted for IFs are usually those free from existing broadcast signals. It’s also normal practice, because it facilitates mixer stage design, to adopt the Oscillator high' style of operation, in which a receiver with a 455kHz IF amplifier stage, covering the frequency range 650kHz-1600kHz, would employ a local oscillator tunable over the range 1105kHz 2055kHz (i.e. above, rather than below, the received signal frequency). The use of a fixed frequency IF amplifier conveys an enormous advantage in receiver design, in that almost any desired amount of amplification can be provided, and the adjacent channel selectivity can be chosen to give any required cut-off characteristic and passband, whether by the use of a cascaded series of bandpass-coupled, or stagger tuned, circuits, or by the use of a quartz crystal, mechanical resonator, or ladder filter, or some other surface acoustic wave type of RF passband defining system.

There are, however, also some drawbacks with the superhet system, of which the chief ones are 'second channel' reception, mixer noise, drift of the tuned frequency, poor mixer conversion efficiency, whistles and cross-modulation effects.

Second-channel interference:

Second-channel or image frequency reception can al ways occur in a frequency changer, for the reason which can be illustrated using the case of the receiver described above. For example, if the local oscillator is tuned to 1105kHz to produce a 455kHz IF output from a wanted 650kHz incoming signal, it will also produce a 455kHz output with an unwanted 1560kHz incoming signal, if one should, by chance, be present. The only way by which this can be prevented is by making sure that the selectivity in the RF amplification, or other tuned frequency, stages between the aerial and the mixer input is adequate to reject the unwanted image frequency signal. For signal frequencies up to a few MHz, the task of providing adequate RF selectivity to reject an unwanted signal 910kHz away from the wanted signal frequency is quite easy to accomplish.

However, for a signal frequency of, say, 20MHz, an adequate degree of rejection of an unwanted second channel signal at 20.910MHz will be much more difficult to secure. This was the reason for the choice of a 1.6MHz IF in early short-wave superhets: that the image frequency would be 3.2MHz away from the frequency of the wanted signal, and therefore some what easier to reject.

Double superhets and direct conversion systems Since the efficiency of an IF gain stage will decrease as the operating frequency is increased, the 1.6MHz IF signal might then be down-converted to 455kHz for further amplification and pass-band filtering, as shown in the schematic layout of FIG19. This process is termed 'double conversion', and such a receiver is called a 'double superhet' to distinguish it from the 'single conversion' method used in the simpler design of superhet receiver.

FIG19 Layout of double superhet. 1st mixer; 2nd mixer; Demodulator; Earth.

The process may be extended yet further, for example to a 'triple conversion' receiver, with three IF gain blocks, and three successive frequency changer stages. In the opposite direction of development, there are what are termed 'direct conversion' receivers, shown in FIG20, in which the incoming signal is mixed with a locally generated one, usually derived from a crystal controlled oscillator, operating at, or near, the same frequency as the wanted signal.

FIG20 Direct conversion receiver.

In the case of a transmitter whose carrier is keyed to send a Morse coded message, the output from the mixer could then be an interrupted audible difference frequency tone. In the case of a single-sideband sup pressed-carrier transmission, provided that the oscillator frequency was chosen correctly, the output from a direct conversion receiver could be a normal audio band speech signal, without the need for any additional demodulator.

For a double-sideband transmission, with carrier, there would be an audibly objectionable 'howl' if the locally generated oscillator signal in a direct conversion receiver differed in frequency from the received carrier by an amount which was within the audible pass-band, and this limits the usefulness of such a system. Nevertheless, there are double-sideband receivers, in which synchronous frequency local oscillators are employed. These are called 'homodyne' or 'synchrodyne' systems, and are discussed later.

Whistles and mixer noise:

A further problem with all superhets is the presence of whistles, which may occur anywhere within the tuning band. These are usually due to the presence at the input to the mixer stage of harmonics of the intermediate frequency signal -- mainly generated at the demodulator stage -- which then interact with incoming RF signals. Alternatively, oscillator harmonics can pro duce whistle generating IF frequency signals by mixing with signal harmonics.

In principle, a superhet will always have a somewhat lower signal-to-noise ratio than a simple TRF receiver, because of the additional noise introduced by the mixer stage. This noise occurs for exactly the same reasons as the thermal noise which is found in any other high gain amplifier system, and is a function of the effective conversion bandwidth, the effective input impedance of the system, and the ambient tempera ture. The problem is worsened by the relatively low stage gain (conversion efficiency), of the mixer stage, which means that its output noise level -- bearing in mind that its output impedance may be fairly high --may be amplified nearly as much as the desired input signal.

Frequency drift:

Another difficulty peculiar to the superhet is that of oscillator frequency drift, which causes a corresponding drift in the tuned frequency of the receiver. This problem is mainly caused by changes in the ambient temperature of the oscillator capacitors and inductors, and methods for minimizing this are discussed later.

However, frequency drift can also be caused by changes in the input impedance of the oscillator or mixer devices, as well as by component ageing effects, or by the effects of mechanical vibration or external capacitative or inductive fields.

In VHF receivers, where high frequency IF stages, such as 45MHz, are used to reduce image frequency signal intrusion, the instability of tuning, with even a well designed variable frequency LC oscillator, would be quite unacceptable, and this problem is worsened because such receivers will inevitably require that the local oscillator operates at a frequency placed above the desired signal frequency. Practical receivers of this kind must therefore either use drift cancelling circuit techniques, or frequency synthesizer systems, based on a stable frequency quartz crystal reference oscillator. These techniques are explored later in this section.

A further associated problem is that of local oscillator frequency pulling, because of the effect of the aerial input signal on the impedance which the mixer stage presents to the local oscillator. This can largely be eliminated by good design procedures, such as the inclusion of a buffer amplifier between the oscillator output and the mixer input.

Cross-modulation:

Two other associated difficulties are those of inter modulation and cross-modulation within the mixer stage. The first of these effects is due to harmonics of input signals, produced by the essential nonlinearity of the mixer, creating spurious higher frequency signal images. The second, again due to mixer nonlinearity, is the more annoying problem in which a strong in coming signal, by causing the mixer input working point to move up and down its transfer characteristic, will add its own program modulation to that of the wanted signal.

Avoidance of these faults requires care in mixer stage design, and some limitation of the size of the signal level at the mixer input. For this reason, pre mixer RF gain is usually chosen to be high enough to ensure that the overall signal-to-noise ratio of the receiver is mainly due to the aerial input circuit, but not so high that inter- and cross-modulation effects become noticeable.

Practical mixer circuitry:

In thermionic valve operated equipment the most common frequency changer stages are the triode hexode, or heptode, valve types illustrated in FIG21.

These are, essentially, screened grid or RF pentode valves in which the screening grid has been separated into two sections, with an additional control grid inserted between them. This grid is internally connected to the control grid of a triode, which is contained within the same envelope and which shares the same cathode as the hexode/heptode. The internal triode section can then be employed as a separate RF oscillator whose output will modulate the electron stream flowing through the hexode/heptode section. This gives the required sum and difference signal generation, but screens the oscillator signal from the aerial input circuitry to avoid unwanted radiation of the local oscillator RF output. Unfortunately, although the triode- hexode/heptode mixer valve can give a good IM and cross- modulation performance -- if the circuit operating conditions are chosen correctly -- it has, by modern standards, a relatively poor noise figure.

In low-cost portable transistor radio receivers, where the principle requirement of the manufacturer is to keep the total component count as low as possible, earlier circuit designs usually employed a self-oscillating mixer, of the general type shown in FIG22, using a bipolar junction transistor, in spite of the fact that this type of circuit can only give a relatively poor performance in technical terms.

FIG21 Triode-hexode and triode-heptode frequency changer valves

FIG22 Circuit layout of single transistor frequency changer

In contemporary low-cost design practice the preferred approach is to use a single integrated circuit, such as the LM1868, in which all the functions of mixer, IF amplifier, demodulator, automatic gain control, and probably also those of a low power AF output stage, are fabricated on a single chip. Although the IC designers are forced to choose circuit configurations which minimize the use of hard-to-make capacitors and resistors, and this constraint can lead to some odd-looking circuit layouts, the basic electronic structures used in this type of IC will generally be similar to the circuit configurations used with discrete component designs. So such ICs, although effective in keeping the manufacturers component count down to a low level, will still usually only offer a comparably indifferent technical performance. Such single chip IC radio systems will usually also incorporate an FM receiver section: a method of signal transmission which is examined later in this section.

The best practicable performance for a multiplicative mixer (one in which the output signal is the product of the aerial and local oscillator signals) is given by a system in which there is a 'square-law' relationship between the input voltage and the output (anode, collector or drain), current of the mixer device, and this condition is met most nearly by a junction FET operated at or near zero gate bias.

An efficient, low-noise, mixer circuit is provided by the circuit shown in FIG23, where the aerial signal is applied to the gate and the local oscillator signal is injected into the source circuit of a junction FET, though this does not give a very good isolation of the oscillator signal from the aerial circuitry, unless an RF or other buffer stage is interposed between the aerial and the mixer. The relatively limited optimum working voltage range of such a circuit also limits the range of signal voltages which can be handled without input overload.

FIG23 FET mixer layout.

This circuit can be elaborated using the cascode layout of FIG24, which gives a somewhat better degree of input/output/oscillator circuit isolation.

These advantages are shared by the dual-gate MOS FET circuit shown in FIG25, a layout which is very commonly used in medium quality discrete component superhet systems. The 7d/7g characteristics of the dual-gate MOSFET are not quite as favorable from the point of view of avoidance of cross-modulation as those of the junction FET, though some MOS FETs are designed specifically for mixer applications, where the gate characteristics have been modified to suit this application. MOSFETs don’t generally have such a low noise figure as junction FETs. They may, however, have better input overload behavior.

Single- and double-balanced systems:

All of the mixer circuits so far described are commonly known as single-ended systems, in which there are single signal and oscillator ports. This type of layout can be elaborated into push-pull systems, de scribed as 'single balanced' and 'double balanced' layouts. These offer a better conversion efficiency -- and consequently an improved s/n ratio -- as well as improved port to port isolation: this factor is better with dual- than with single-balanced types.

FIG24 Cascode connected FET mixer system

FIG25 Dual gate MOSFET mixer

The simplest single-balanced system is that using a diode bridge layout, shown in FIG26, where the oscillator signal is injected into the centre tap of the secondary winding on a wide-band RF transformer --usually 'tri-filar' wound on a toroidal ferrite core. This layout gives good oscillator to input, but poor oscillator to output isolation. Diode mixers of this type can give excellent performance in respect to cross-modulation and input overload, but need a high oscillator output voltage at a low source impedance, and suffer from the drawbacks of both a conversion loss and a relatively poor noise figure.

A double balanced diode layout, often called a 'ring mixer', is shown in FIG27. This gives good port-to-port signal isolation, and excellent IM and cross-modulation characteristics. Using 'hot carrier' or 'Schottky-type' (metal/semiconductor junction) diodes, this layout is usable up to the UHF (gigahertz) range.

Single and double-balanced mixer layouts using MOSFETs and junction FETs are shown in FIGs 28-30.

FIG26 Diode bridge single-balanced mixer

FIG27 Double-balanced ring mixer

FIG28 Single-balanced mixer using

FIG29 Balanced mixer using junction FETs

FIG30 Low noise balanced mixer based on junction FETs

Oscillator stages:

A number of circuit designs which could be used as the local oscillator in a superhet have been described in Section 13. The important considerations in this particular application are that they should be stable in frequency, and have a low noise and spurious signal content in their output waveform.

Temperature compensation

The importance of frequency stability is dependent upon the degree of selectivity of the receiver, and the tolerability of tuned frequency drift with time. Most LC tuned circuits will suffer from drift as a result of changes in the ambient temperature of the oscillator circuit, though, with low power solid-state circuitry, 'warm-up' frequency drift following switch-on is no longer a particular problem.

Some compensation for thermal drift in LC oscillators can be achieved by the incorporation within the tuned circuit of additional capacitors having a negative or positive temperature coefficient. These may simply be connected in parallel with the tuning capacitor, and would typically be of metallized ceramic construction.

Such capacitors are usually designated N- or P-, with the coefficient specified in parts per million per degree Celsius. For example, an N-750 marking would denote a temperature coefficient of -750 parts/million/°C, and a P-100 one a capacitor with a +100 ppm/°C characteristic. Similarly a NPO marking would imply a near zero temperature coefficient for the component. Choosing the correct value and type of temperature compensation components is usually a laborious and tiresome exercise, especially if compensation over a wide temperature range is sought.

A high degree of oscillator frequency stability is particularly desirable in the case of receivers designed to receive CW signals (transmissions consisting of a simple sinewave carrier, interrupted by morse or other coded keying patterns), where the signal is made audible by a beat frequency oscillator (BFO) stage.

This operates by heterodyning the signal, in a further mixer stage, with an oscillator having an adjustable output frequency close to that of the IF. In this case a shift in incoming signal frequency due to oscillator drift will result in an audible change in the BFO pitch.

Oscillator frequency drift will also be embarrassing where the receiver is used to receive a 'suppressed carrier' signal, where if the reinserted carrier frequency moves away from the desired frequency, the received signal may become unintelligible.

The freedom of a local oscillator output from spurious signals is obviously necessary if whistles and the reception of signals at unwanted frequencies is to be avoided, and this demands care in the layout of the resonant LC circuit, to avoid inadvertent inductive loops, or inconvenient stray capacitances.

The requirement for a good signal-to-noise ratio in the local oscillator output arises because any modulation component of the local oscillator output, such as noise, will be added to the output sum and difference signals, in just the same way as the modulation present on the carrier of the wanted signal. So, if a 60dB signal-to-noise ratio is required for an input signal of ??µF amplitude then the local oscillator signal must have a signal-to-noise ratio of at least 80dB. Also, in some cases, mixer nonlinearities may exaggerate this problem by increasing the extent to which local oscillator noise is added to the composite signal.

IF amplifier stages:

The basic requirements for both RF and IF amplifier stages are the same -- that they should give a useful degree of gain, a high signal-to-noise ratio, an adequate input linearity (to avoid cross-modulation), and that they should be stable in operation. However, whatever the means by which adjacent signal selectivity is obtained, the fixed frequency IF stages have the advantage that they won’t usually require tuning except in setting up the receiver. Any of the HF filter systems examined in Section 11 can be employed, and two typical IF amplifier blocks in which the selectivity is defined, in the first case by a group of four band pass-coupled tuned circuits, and, in the second instance by a surface acoustic wave ladder filter, are shown in FIGs 31 and 14.32.

FIG 31 IF amplifier stage using bandpass coupled tuned circuits. AGC

FIG32 IF amplifier using ceramic SAW ladder filters

Automatic gain control (AGC)

An inevitable difficulty with the reception of amplitude modulated signals is that the larger the signal, the larger the demodulated audio frequency output. This means not only that strong signals will be louder than weak ones, which will necessitate continuous adjustment of the receiver output gain control as it’s tuned from one signal to another across the band, but also that fluctuating signal strength, due to changing reception conditions, will lead to 'fading'. Because of the very high IF gain which is usually available, it becomes possible with superhets to derive a further signal-dependent voltage, proportional to the average magnitude of the received carrier, and use this voltage as a gain control mechanism, to reduce the gain of the IF stages (and usually, also, the RF stages if used) to try to ensure that the magnitude of the IF signal presented to the demodulator is of a substantially constant size.

A simple AGC circuit is shown in FIG33, in which a negative gain control voltage, suitable for controlling the gain of a valve or MOSFET/FET operated IF amplifier, is derived from a diode demodulator.

Some HF filtering is necessary in the AGC loop to avoid HF instability due to signal feedback, but the time constants for the gain control circuitry must be chosen with care, since too long a response time in the control system will reduce its ability to correct rapid 'flutter type' fading, while too short a time constant will attenuate the lower modulation frequencies, which will be interpreted by the AGC system as un wanted fluctuations in signal strength.

It must be remembered also that the gain control system is a closed loop servomechanism, and if the accumulated phase shifts in the control loop approach 180°, the whole system may become unstable, leading to the problem known, from its sound, as 'motor boating'.

To avoid needlessly reducing the gain of very weak signals, it’s common to arrange the design so that there is a lower threshold level in the AGC control system, below which the AGC voltage won’t be applied. This is usually termed 'delayed AGC'. Similarly, to in crease the effectiveness of the AGC system, designers may sometimes employ a DC amplifier in the AGC loop. A low distortion demodulator system, using both amplified and delayed AGC, is shown in FIG34. (JLH, Wireless World, October 1986, p. 17).

Signal demodulation:

Most of the demodulator systems described above in relation to TRF type receivers will operate quite satisfactorily as the demodulator, or 'second detector' in a superhet. However, since there is almost always an adequately large IF output signal available, and the designers of AM receivers are seldom concerned about minimizing demodulator distortion, a simple diode demodulator, of the kind shown in FIG33, is nearly always used.

FIG33 Automatic gain control voltage generation circuit

FIG34 Low distortion demodulator incorporating amplified AGC

Stable frequency oscillator systems:

The use of high signal frequencies in receiver or oscillator systems exaggerates the problems of frequency drift, simply because the same proportional drift becomes a higher frequency error at higher frequencies. For example, a 0.01% drift at 100kHz will be 10Hz -- a negligible amount in most applications.

However, the same proportional drift at 100MHz will be 10kHz -- an error which would be unacceptable in almost every case. A relatively simple solution to this problem, provided that the coverage of the receiver is relatively limited, is to use a fixed frequency, quartz crystal controlled, local oscillator, coupled with a tunable IF. This would, however, suffer from the drawback that if the chosen, tunable, IF was low enough not to suffer from frequency drift, there would be a problem due to second-channel breakthrough, due to inadequate RF selectivity, while if the chosen IF was high enough to avoid this problem, the IF stage -- which would also be a superhet receiver -- would also suffer from frequency drift.

FIG35 The Barlow-Wadley loop drift cancelling oscillator system [Wideband IF, amplifier; 2nd mixer 2nd IF amp. 3rd mixer 3rd IF amp.]

A number of systems have been evolved to reduce the extent of this problem, of which the two most interesting, in that they represent entirely different, and original, design philosophies, are the Barlow Wadley drift cancelling loop, and the phase-locked loop frequency synthesizer system.

The Barlow-Wadley drift cancelling system

The way this system works is shown, in schematic form, in FIG35. In this the aerial signal, amplified by a conventional tuned RF stage, is fed to a first mixer, along with the output signal from a good quality LC tuned local oscillator, operating at a frequency which is above that of the incoming signal. The resulting first IF signal is then amplified at an IF frequency of , say, 45MHz and fed to a second mixer stage.

Because of the inevitable frequency drift in the first local oscillator stage, the 45MHz IF output will also suffer from drift. However, attached to the oscillator system is a stable frequency quartz crystal oscillator, operating at, say, 1MHz, and the output from this is fed to a harmonic generator circuit, to produce an array of output frequencies, at 1MHz intervals, extending up to, say, 100MHz or more. This harmonic series is then mixed with the output of the first variable frequency LC oscillator, and the composite output from this mixer is fed to a selective amplifier tuned to a frequency which is higher than the 45MHz first IF, by an amount equal to that of the second (say, 3MHz) IF amplifier. The output signal from this will suffer from an identical frequency drift, and in the same direction, as that of the first IF output signal, and the resultant eiTors will cancel, giving a drift-free input to the second IF amplifier/demodulator combination.

As an elaboration of this basic principle, the first IF can be chosen to have a fairly wide, flat-topped frequency pass-band, and the second IF stage can then be made tunable over, say, the 2-3MHz range, to provide a band-spreading facility for the receiver -- since frequency drift in a 2-3MHz receiver will seldom be a major problem. The first variable frequency oscillator can then be used to select the incoming signal frequency in 1MHz blocks.

Frequency synthesizer techniques:

These can be divided into 'partial' synthesis and 'full' -- i.e. 'digital' -- synthesis designs. The partial synthesis method combines a quartz crystal oscillator with a standard LC variable frequency oscillator (VFO), as, for example, in the simple circuit arrangement shown in FIG36. In this a receiver designed to cover the frequency range 28-30MHz, with a first IF frequency of 10.7MHz, employs a VFO tunable over the frequency range 1.7-3.7MHz. The output from this is heterodyned with that from a 37MHz quartz crystal, and passed through a 38.7-40.7MHz selective amplifier, before being mixed with the aerial signal in the receiver. This technique is adequate for coverage of a limited and specified waveband, but would require that a range of crystals were available to allow tuning over a wider input signal range. Full frequency synthesis techniques rely on the use of a phase-locked loop (PLL), for the control of the oscillator, so I propose to explain this system first.

FIG36 28-30MHz receiver system.

The phase-locked loop:

This circuit arrangement, shown in schematic form in FIG37, provides a powerful technique for forcing a variable frequency voltage controlled oscillator (VCO) to operate at a specified frequency, as well as a means for deriving a DC voltage output related to this frequency, if this should be needed.

PSD Low-pass filter

DC amplifier; RF output; Control voltage input

FIG37 The phase-locked loop

In the simple circuit arrangement of FIG37, the PLL consists of just four elements, an input mixer or 'phase sensitive detector' (PSD), a low-pass filter, a DC amplifier, and a voltage controlled oscillator, whose output frequency is determined by the control voltage applied to it. If the free-running frequency of the VCO is close to that of the input signal, and if it’s postulated that, momentarily, the input frequency (ft), is the same as the VCO output frequency (ft), the PSD output will be a DC voltage which is related to the phase difference between these two signals. If this is amplified, and applied to the input of the VCO, it will cause this oscillator to speed up, or slow down, over part of a cycle, in such a manner that f1 and f2 are brought into phase and frequency synchronism with one another. If the free-running frequency of the VCO is truly that which requires a zero DC control voltage, then ft and/2 will also be at phase quadrature, since that is the condition at which the PSD will have a zero voltage output, and if the loop amplifier has a high enough gain, a near-quadrature phase relationship be tween the signal and the VCO will also be the condition for non-zero VCO control voltages. Moreover, it’s also found that if f_t and f2 are not initially identical, then, provided that the difference frequency is within the pass-band of the low-pass loop filter, known as the capture range, then the VCO will be drawn into, and will remain, in synchronism with the input signal. This is described as the PLL being in lock'.

If a frequency divider is incorporated in the VCO/PSD loop, then the VCO can be forced to oscillate at a multiple of the input frequency. Similarly, if a frequency divider is introduced into the control frequency path (ft), then the VCO can be caused to oscillate at a sub-multiple of f_t.

The choice of filter bandwidth is dictated by the required capture range, and the required purity of the VCO output, since increasing the bandwidth of the filter increases the amount of wideband noise which will be added to the control voltage.

Digital frequency synthesis techniques:

The complete superhet receiver shown in outline in FIG38 uses an elaboration of the frequency multiplication, frequency division possibilities of the phase-locked loop circuit to generate a fully variable, but crystal-controlled, local oscillator frequency.

In this, a crystal controlled oscillator, operating at a frequency f0, is used as a reference frequency generator, whose output is frequency divided by a factor of ?, before introduction as the control signal, f, to the PSD. The output from a high frequency voltage con trolled oscillator is also divided before being applied to the PSD, and the PSD output is filtered and amplified before application as the control voltage to the VCO. Then, when the PLL is in lock, the output from the VCO, f3, will be determined by the crystal frequency according to the relationship …and will be very stable in frequency. By the use of a microprocessor to control the two division ratios, the required input signal frequency of the receiver shown in FIG38 can be controlled by receiver front panel push-button selection, and the selected signal can be shown, for convenience, on a digital frequency display.

A number of manufacturers now offer single-chip frequency synthesizer ICs, which greatly facilitate the use of this technique.

FIG38 Complete superhet receiver using frequency synthesized local oscillator system.

Synchrodyne and homodyne receiver systems:

The difficulty of achieving adequate selectivity with TRF type receivers, and the many technical problems associated with superhet systems, has prompted exploration of the possibilities of direct conversion receivers, of the kind discussed above. The advantage of a direct conversion receiver, if the local oscillator frequency can be controlled so that it’s truly identical to that of the carrier of the received signal, is that demodulation of the incoming signal will occur with out the need for a specific demodulator circuit: the input sum and difference frequencies produced by the amplitude modulation of the carrier will be directly transformed into an audible signal. This avoids the problems of poor demodulator sensitivity and linearity. Moreover, since the sidebands of adjacent interfering signals will also be transformed into audio signals, but of a much higher pitch, the receiver selectivity can be provided by post-mixer AF filtering, for which there are a number of useful low-pass filter circuits, as shown in Section 8. The only problem with this system is that if the pre-mixer selectivity is inadequate, cross-modulation can occur at the input to the mixer, to introduce an interfering signal which is proof against subsequent attempts at removal.

The difficulty, obviously, is to generate a local oscillator signal which is truly in synchronism with the required aerial signal. In the homodyne circuit, best implemented as the demodulator system for a broad selectivity superhet, a solution to this problem is at tempted by using the carrier of the desired signal for this purpose, as shown in FIG39. In this, a very narrow bandwidth IF stage, in parallel with the normal IF amplifier, is used to generate a signal, derived from the input signal carrier, but amplitude limited to remove its modulation component, which can be employed to synchronously demodulate the incoming signal. This system does work, but small changes in the signal tuning will alter the relative phases of the two signals applied to the mixer, and this will substantially alter the magnitude of the audio output from the mixer.

In the synchrodyne receiver, it’s also required that the local oscillator signal fed to the mixer is synchronous with the wanted signal. An attempt is usually made to satisfy this requirement, in simple systems, by introducing a small amount of the aerial signal into the local oscillator circuit. If the local oscillator frequency should drift away from the signal frequency, the loss of synchronism becomes audible as a loud whistle, whose pitch is determined by the difference frequency, and this fact, especially noticeable when tuning from one signal to another, coupled with the difficulty of maintaining synchronism of the local oscillator, has prevented the use of this technique on anything other than a laboratory scale. However, a synchrodyne receiver design in which phase-locked loop techniques were used to force the local oscillator into phase and frequency synchronism, and in which inter-station muting was incorporated, is shown in outline in FIG40. A practical receiver based on this structure was described in Wireless World. (JLH. Jan. 1986, pp. 51-54, Feb. 1986, pp. 53-56 and March 1986, pp. 58-61).

FIG39 Circuit arrangement of homodyne receiver. Limiter and narrow-band

Frequency modulation (FM), systems:

In this type of transmission system, the frequency, rather than the amplitude, of the transmitted signal is modulated to carry the program content. This idea, also, was due to the same remarkable Major Edwin Armstrong who invented the concepts of positive feedback, regeneration, the super-regenerative and the superhet types of radio receiver.

As a broadcasting technique, FM has many advantages, of which the major ones are that, since the amplitude of the received signal is now no longer important, all of the incoming signals can be amplified sufficiently to allow their amplitude to be limited by a clipping stage, such as the simple back-to-back diode limiter shown in FIG41. This has the immediate practical benefit that signal fading is eliminated, and that all incoming signals are received at the same volume level. In addition, 'frequency discriminator' demodulation systems are, at least potentially, very much better in terms of linearity, and freedom from signal distortion, than comparable AM demodulators.

They will also, in good designs, give a high degree of rejection of 'impulse type' noise, such as that caused by electrical switch contact interference and motor vehicle ignition noise. Rejection of interfering adjacent channel signals is also assisted by the 'capture' effect, a feature of the demodulator system employed, in which the presence of a stronger signal, at the demodulator, will completely suppress a somewhat weaker one: the extent to which this occurs is known as the 'capture ratio'. Good FM demodulator circuitry can offer capture ratios better than 1dB. The major drawback of FM broadcast systems is that, for optimum modulation bandwidth and linearity they require a very wide transmission bandwidth. Peak modulation bandwidths of +/-75kHz are typical of contemporary FM broadcast transmitters, for which good quality reception requires a receiver bandwidth of at least +/-120kHz. Clearly, this amount of broad cast bandwidth is not available in the existing and crowded long- or medium-wave commercial transmission bands, so a part of the frequency spectrum between 88MHz and 108MHz has now been set aside, by international agreement, for domestic FM broad cast transmissions. The use of this part of the RF spectrum offers both advantages and disadvantages, in that the general absence of significant ionospheric reflection means that reception will only be possible, normally, over line-of-sight distances. So, for general reception, a network of local stations is necessary, and the choice available to the listener is usually restricted to the programs broadcast on his or her own national network. On the other hand, the limited range of transmissions in this frequency band eliminates over crowding, and the likelihood of unwanted adjacent channel interference.

FIG40 Synchrodyne receiver

FIG41 Simple diode amplitude limiter circuit

FM receiver design FM demodulator systems The design of receiver used for FM reception usually follows conventional radio practice, only modified as necessary to suit the 88-108MHz signal spectrum, and the required IF bandwidth. The frequency normally used for the IF is 10.7MHz, and a wide range of piezo-electric ceramic surface acoustic wave (SAW) filters is available, allowing substantially flat-topped response curves, with a range of bandwidths, for different applications.

Contemporary 'tuner head' design practice favors either dual-gate MOSFET, or neutralized junction FET stages, with similar devices for the mixer stage.

Since varicap diode tuning allows additional tuned circuits to be added without practical difficulties, it’s common, in better class tuners, to find three or four tuned circuits preceding the mixer, to reduce cross modulation effects, and improve the effective s/n ratio.

A typical layout for a varicap tuned FM RF/mixer circuit is shown in FIG42, and a 10.7MHz IF stage, using ceramic SAW filters, is illustrated in FIG32.

FIG42 Commercial FM tuner head design.

The slope detector:

Various methods have been adopted, over the years, as a means for converting a constant amplitude RF signal which varies in frequency into a varying DC output voltage level. Of these, the simplest, and most straight forward, is to feed the signal to a diode detector, coupled to a tuned circuit, whose natural resonant frequency is displaced somewhat to one side of the signal frequency, as shown in FIG43. This will somewhat distort the demodulated signal, as shown, because of the curvature of the tuned circuit resonance curve. It will also offer no immunity from AM signal breakthrough or impulse-type interference.

FIG43 Simple FM 'slope' demodulator

FIG44 --- b Output voltage from Round-Travis circuit

The Round-Travis circuit:

An improvement in this arrangement is offered by the 'Round-Travis' layout, shown in FIG44, in which two such demodulator circuits are connected with their outputs in series, with one tuned somewhat above, and the other tuned somewhat below the mid point signal frequency. The effect of adding the two resonance curves is to improve the overall demodulator linearity.

The Foster-Seeley discriminator:

The most important improvement in the effectiveness and linearity of the FM demodulator was the introduction of the 'Foster-Seeley' circuit shown in FIG45. The operation of this is based on the fact that the phase of the voltage developed across a tuned circuit, which is loosely coupled to an input winding, will be at quadrature (90° or 270°), to that of the input signal when the tuned circuit is at resonance. This leads to the possibility that, if an input signal is injected into the centre tap of the resonant circuit, it will either add to the output voltage or oppose it, depending on the input frequency, and this circuit will give a very linear output voltage/input frequency relationship for relatively small input signal voltages.

The Foster-Seeley discriminator ; Frequency/voltage characteristics of Foster-Seeley circuit.

FIG45

The major practical drawback with this layout is that it offers very little discrimination against impulse-type AM breakthrough, a persistent problem with VHF reception due to the prevalence of motor vehicle 'ignition noise'.

The ratio detector:

The circuit known as the 'ratio detector', shown in FIG46, minimizes this problem by connecting the two demodulator diodes in opposition, so that the output is the ratio between the two signal voltages rather than their sum. It’s not, however, either as efficient or as linear in its demodulation characteristics as the Foster-Seeley discriminator.

Contemporary design practice favors, almost exclusively, the use of a quadrature coil 'gate coincidence' demodulator (GCD) combination, shown in FIG47, because this can easily be fabricated as part of a monolithic IC structure. Since these ICs, such as the RCA CA3189, also offer a host of other useful functions, such as gain control, automatic frequency control, signal strength indication and off-station noise 'muting', it’s understandable that they are widely used.

The GCD is a true phase-detector, in that there will be no output voltage change if an input signal voltage is applied to either of its signal ports, on it own, or if two input voltages arc applied in phase quadrature --which would be the case if the quadrature coil, L1/C1 is resonant at the incoming frequency. If the input frequency is altered, there will be a change in the output voltage level which, over a limited voltage range, will be as linear as the output from a Foster Seeley discriminator, but with improved AM rejection. Typical THD values for a single quadrature coil GCD design lie in the range 0.8-1.5% THD for a +/-75kHz modulation level. By comparison, an AM demodulator would give 2-5% THD for a 50% modulation level.

FIG46 --- The gate coincidence demodulator---The ratio demodulator, Frequency A cottage characteristics of ratio demodulator

FIG47 --- Circuit layout of gate-coincidence demodulator system; Performance of gate-coincidence demodulator:

It’s claimed by the manufacturers of gate coincidence demodulator ICs that the linearity of the de modulator system can be improved to allow better that 0.1% THD distortion levels, by coupling a further resonant tuned circuit to the quadrature coil. This configuration remains more of a theoretical possibility than a commercially valuable system because the out put distortion given by this arrangement depends critically upon the coupling factor between these two tuned circuits. If they arc 'under coupled' little benefit is given, while if they are 'over coupled', the demodulator output will show a pronounced crossover distortion type kink at the mid-point frequency (JLH, Wireless World, March 1991, p. 220).

Phase-locked loop demodulator systems:

The circuit layout shown in FIG37 can also be used to provide a very efficient, low distortion FM demodulator system, in that if the loop is locked, the frequency of the voltage controlled oscillator will be identical to that of the input signal. If this varies in frequency, the DC output from the loop (point 'B' in the figure) will be that which is required by the voltage controlled oscillator to cause it to oscillate at that frequency. If there is an accurately linear relationship between the VCO control voltage and its output frequency, then an input FM signal, applied to point *A\ will also be demodulated, by this means, at point '?', with very little modulation distortion. Such a PLL demodulator will also give very good AM rejection and an excellent capture ratio. As a matter of personal interest, I have designed a number of experimental FM receivers using this system, of which the most recent was offered as an amateur constructional project in 1987 (JLH, Electronics Today, March 1987, pp. 34 38).

Automatic frequency control (AFC):

One of the inherent advantages of any FM demodulator is that the signal output is a DC voltage which is related to the incoming signal frequency. If this varies, the output voltage will change. With tuner circuits using varicap diodes, this DC output voltage can be applied to the oscillator tuning system so that un wanted drifts in the tuned frequency can be corrected.

This is a particularly valuable feature in receivers operating at such relatively high signal frequencies, so that most commercial tuner head designs provide an AFC input control point, for direct connection to the AFC output from the demodulator IC.

Inter-station noise muting:

With FM tuner designs in which a high signal amplification level is coupled to an amplitude limiting clipping stage, all signals, including the thermal noise input from the aerial, or that due to the input stages to the receiver, will be amplified to the clipping level.

Moreover, since the thermal noise will have a random distribution in both amplitude and frequency it will be demodulated into a wideband (white), noise signal, and this is a disconcerting feature in inexpensive FM receivers without automatic inter-station muting facilities.

Various techniques are used to mute this noise, of which the simplest is to monitor the IF signal strength, at some point prior to the limiter stage, and arrange the circuit to switch off the audio output if the measured signal level falls below some predetermined value. A circuit for this purpose is shown in FIG48.

FIG48 Inter-station muting circuit for FM receiver

A rather more sophisticated approach, employed by RCA in their CA3189 IC, is to monitor the extent of signal frequency deviation, and mute the output if this exceeds a certain value, as it will do with a wideband noise input.

FM Stereo broadcast transmissions:

The extra modulation bandwidth available on the Band ? region of the VHF allocated for FM broadcasts has been exploited to allow stereo broadcasting, using the GE-Zenith 'pilot tone' system, for which the band width distribution is as shown in FIG49. In this a normal 'L+R' mono signal is transmitted, using the 30Hz-15kHz part of the modulation spectrum, while an additional 'L-R' signal is broadcast as a double sideband supersonic signal based on a carrier frequency of 38kHz, giving a maximum modulation for the composite (L+R/L-R) signal of 90% of the per mitted +/-75kHz deviation. This 38kHz carrier is suppressed before transmission to avoid unwanted breakthrough into the normal audio passband, and is then reconstructed, in the 'stereo decoder' of the receiver, from a phase synchronous 19kHz pilot tone broadcast at 10% effective modulation depth. This technique is 'mono compatible' in that, on FM receivers without stereo decoding facilities, the signal is received as a normal mono broadcast, whereas, on suitably equipped receivers, a pair of separate, low distortion, 'L' and 'R' channel outputs, having up to 40dB channel separation, are provided. Decoding this composite signal can be effected by treating the L-R signal as a normal LF radio signal and, after separating the carrier from the L+R signal, demodulating it in a conventional way to give a L-R output, from which the separate L and R signals are obtained by subtraction or addition. Alternatively, the L and R outputs can be obtained by synchronously switching the composite signal between the two channels, at a 38kHz frequency -- a process which effectively does the same job.

These alternative decoding techniques are illustrated schematically in FIGs 50 and 14.51.

FIG49 The carrier modulation characteristics used in the GE-Zenith FM stereo transmission system

FIG50 Matrix stereo decoder system

FIG51 Switching type stereo decoder

Appendix.

Transmission and reception data

There are a number of terms which are conventionally used in the classification of radio broadcast signals.

Those relating to the frequency spectrum are listed in TBL. 1.

TBL. 1 Classification of radio frequency bands

TBL. 2 Broadcast band allocations

Waveband Allocation Wavelength

(* denotes amateur transmitter frequency allocations).

Within these frequency bands segments have been allocated, by international agreement, for domestic and amateur usage, as shown in TBL. 2.

Note. It’s observed that not all national/commercial broadcasting authorities adhere strictly to these frequency allocations.)

In addition to these allocations there are other VHF/UHF band designations, listed in TBL. 3.

TBL. 3 VHF/UHF Frequency band designations

Waveband Frequency allocation.

Recent changes in the internationally agreed transmitter frequency separation in the LF and MF bands have imposed constraints on the broadcast bandwidth, in order to limit adjacent channel sideband interference. VHF broadcasts are not subject to these constraints, and can therefore provide a wider program frequency range.

Details of the British broadcast standards, published by courtesy of the BBC, are listed below.

Audio bandwidths

Peak deviation level (100% modulation), corresponds to +/-60.75kHz deviation


Prev. ------- Next

top of page  Article Index  Home



Home | Glossary | Books | Links/Resources
EMC Testing | Environmental Testing | Vibration Testing

Updated: Friday, 2014-10-24 23:11 PST