(Reproduced by courtesy of Sony Ericsson)

(a)The BiM 2 433–64 data transceiver operates in the 433 MHz licence free band. Conforming to EN 300 220–3 and EN 301 489–3, it transmits and receives data at up to 64 kbit/s with a range of up to 200m external, 50 m in building.

(b) Block diagram of the BiM 433-64.

(a) (Reproduced by courtesy of Radiometrix Ltd); (b) (Reproduced by courtesy of Radiometrix Ltd. www.radiometrix.co.uk)

Figure 11.1a shows the block diagram of a 1 kW HF transmitter, such as might be used in commercial or military point-to-point communications. The block diagram of a low power solid state VHF FM transmitter, such as might be used as a ‘fill-in’ transmitter where the signal from the main transmitter is inadequate, would be very similar. The baseband signal would consist of the programme input material, speech or music, nowadays often in stereo. Baseband signal processing produces the mono-compatible sum signal, the stereo difference signal which is modulated onto a suppressed subcarrier, and the stereo pilot signal at half the frequency of the subcarrier. Often also, RD (radio data) information at a low bit rate is modulated onto an additional subcarrier. This carries a variety of information such as station identity, other frequencies on which the same programme can be received (useful for auto-searching FM receivers in cars), etc. The composite baseband signal is modulated onto a carrier at a suitable IF frequency such as 10.7 MHz and then, after filtering to the final bandwidth, translated in a mixer stage to the final transmit frequency. In the USA, the serasoidal modulator was at one time popular, but this has a maximum phase deviation less than ± 180°. Frequency multiplication was therefore necessary to obtain the required deviation, making it difficult to achieve an acceptable signal to noise ratio even with a mono signal. In a broadcast transmitter, the transmit frequency is seldom if ever changed, so tuning arrangements are much simpler than those commonly found in receivers. However, sophisticated protection arrangements for safety purposes are necessary, including interlocks to prevent the equipment being accidentally powered up whilst personnel are servicing it, and trips to protect the PA in the event of an antenna fault, etc. In one sense, a good transmitter is easier to design than a good receiver, since the only signal it has to handle is the wanted signal. This is especially true of a transmitter working over only a fairly narrow percentage bandwidth such as the 88-108 MHz VHF FM broadcast band, as it is then easy to arrange that no mixer spurious outputs fall on or close to the wanted output in the transmit band. In an HF communications transmitter covering the band 1.6-29.999 MHz, the problem is more acute. A double conversion scheme would therefore be used with the modulation typically taking place at 1.4 MHz, the signal then being translated to an IF of (say) 45 MHz before down conversion to the final transmit frequency. Low-power UHF transmitters used in walkie-talkies, portable telephones, etc., operating in parts of the 470-960 MHz spectrum usually use complete PA modules from one of the leading manufacturers of RF power transistors, such as Freescale or Philips. These modules accept a drive signal in the milliwatt range, are available in various power output ratings and are ready set up with all interstage matching built in. High power transmitters in this band, e.g. Band IV/V TV transmitters, use valve PAs, although solid state transmitters are currently pushing up to a power level of kilowatts.

Figure 11.2a and b shows single and double superheterodyne receiver block diagrams, such as might be used in a quality short-, medium- and longwave AM radio and an HFcommunications receiver respectively. In the AM single superhet, the IF frequency is typically in the range 455-470 kHz with an IF bandwidth of 8 kHz or even as little as 5 kHz, allowing a modest degree of rejection of stations on adjacent channels (medium wave channel spacing is at 9 kHz intervals in Europe and 10 kHz in USA). However, reception is usually restricted to the lower frequencies in the short waveband, as the image frequency (twice the IF frequency) is only removed by less than 1 MHz from the desired frequency. In a single superhet HF receiver an IF of 1.4 MHz would typically be used, but even this leaves an inferior image performance. Therefore a double conversion system is nowadays always employed in professional HF communications receivers. This moves the image frequency to the VHF band and simple front-end filtering prevents such signals reaching the first mixer.

A high first IF is also desirable for other reasons. If the input at the R port of the first mixer (usually a DBM) includes large unwanted signals, there may be other outputs at IF in addition to that due to the wanted signal. These are all varieties of ‘spurious response’ due to imperfections in the DBM which the mixer manufacturer tries to minimize. There are for example possible spurious outputs due to harmonic mixing. A mixer containing non-linear devices (diodes), will produce harmonics of the frequencies present at its inputs, and these harmonics themselves are in effect inputs to the mixer. So if a single superhet HF receiver with a 1.4 MHz IF is tuned to 25 MHz, the LO will be at 26.4 MHz and the second harmonic of this is at 52.8 MHz. If a large unwanted input at 25.7 MHz is present, its second harmonic at 51.4 MHz may be produced within the mixer and this will beat with the 52.9 MHz second harmonic of the LO to give a spurious output at the 1.4 MHz IF frequency. If the mixer is balanced at the R port, the effect will be greatly reduced but, in practice, not eliminated entirely. The usual double balanced mixer should not result in the production of even harmonics of either the RF signal or the LO, but mixer balance is never perfect. The spurious response due to second harmonics of LO and unwanted signal is variously known as the ‘2:2 response’ or the ‘half IF away response’ since it occurs at a frequency removed from the desired frequency by half the IF frequency. An impractical degree of front-end selectivity would be required to suppress this response to a level where a 100 mV unwanted signal would not drown a 1 μ wanted signal. Further, a double balance mixer offers no such enhanced rejection to the 3:3 response, removed from the tuned frequency by only one-third of the IF frequency, or other odd order responses. This type of receiver spurious response falls off rapidly as higher and higher order harmonics are involved. It can thus be avoided virtually completely by using a double superhet configuration with a first IF well above 30 MHz, since the harmonic orders involved would then be very high. Possible responses at the IF, image and at frequencies as described above are all examples of external spurious responses or ‘spurs’. Most receivers, even professional communications receivers, will have one or more internal spurs. These are frequencies at which there is an apparent CW output even with the antenna input terminated in a resistive load. They are due to spurious spectral lines occurring in the synthesizer and/or interactions between the first and second local oscillator and the frequency standard. Other possibilities are harmonics of the clock frequency of the microcontroller included in all modern receivers.

A superhet is troubled by other types of spurious responses, of which intermodulation is one. Imagine the receiver is tuned to a weak wanted signal and that there are two large unwanted signals, removed by +100 kHz and +200 kHz from it. The lower of the two third-order intermodulation products of the unwanted signals will fall on the wanted frequency: the formation of intermodulation products due to circuit non-linearity is covered in Chapter 6. In a professional HF communications receiver, e.g. Figure 11.7, the third-order intermodulation performance is usually specified with unwanted signals offset from the tuned frequency by ±20 and 40 kHz, at which spacing there will be no assistance from any front-end tuning. However, second-order intermodulation products will not be a problem except in a ‘wide open’ receiver with no front-end tuning of any description: a high quality HF receiver will usually have either a tuned front end or a bank of nine sub-octave band-pass filters covering the 1.6-30 MHz band. The appearance of high dynamic range double-balanced mixers led in the 1970s to a rash of wide open HF receivers, but with the ever heavier use of the HF band and the resulting mayhem against which receivers have to work, the true worth of a tuned front-end is again recognized.

Two other headaches for the receiver designer are cross-modulation and blocking (desensitization). In the former, the envelope modulation on a large unwanted off-tune signal becomes impressed on a smaller wanted signal and cannot therefore be removed by any subsequent filtering. Blocking consists of a reduction of gain to the wanted signal, caused by a large unwanted off-tune signal. Cross-modulation and blocking are usually specified for an unwanted signal offset of 20 or 30 kHz. Like intermodulation, they would not occur in a receiver in which all stages up to and including the final bandwidth defining second IF filter were perfectly linear. It is for this reason that most of the gain is provided in the second IF stages following the final bandwidth filter - by that time the only signal present is, it is to be hoped, the wanted one. Keeping the gain as low as possible in the earlier stages minimizes the size of any large unwanted signals in those stages, minimizing the effect of their inevitable slight non-linearity. However, sufficient gain must be provided to compensate for attenuation in tuned circuits, mixers, etc., so that the signal to noise ratio of a small wanted signal at the input to the receiver does not become noticeably worse at the receiver’s output. As the level of the wanted signal increases, the receiver’s gain must be turned down so as not to overload the last IF stage and/or detector. The operator can do this using the manual RF gain control if provided, but usually it is the job of the AGC (automatic gain control) circuitry, which is ‘scheduled’ so as to maintain the best signal to noise ratio for the wanted signal. The gain at the back end of the second IF amplifier strip is turned down first, to approximately unity. Then earlier stages are successively turned down, until eventually the gain of the RF stage (if fitted) is turned down, or alternatively a voltage controlled attenuator preceding it is brought into operation. AGC which is scheduled in this way provides better performance than winding down the gain of all controlled stages in parallel, or applying full AGC to the IFs and half AGC to the RF stage. It is arranged that the final IF stage is capable of driving the signal and AGC detectors to full output even at maximum gain reduction, either by limiting the gain reduction of that stage or by not controlling it at all. Compared to manual RF gain control, AGC has of course the advantage that it will continually adjust the receiver’s gain to compensate for variations of the strength of the wanted signal due to fading. Typically, sufficient gain is provided in the AGC loop to keep the variation in output signal level to 5 dB or less for a change in input level of 100 dB. AGC is not without its problems: AM signals such as broadcast stations on short wave (and on medium wave, after dark) may suffer selective fading of the carrier, leaving the sidebands unaffected. The AGC will increase the receiver’s gain leading to a large increase in the audio output level, which will moreover be grossly distorted, since in the absence of the carrier, the modulation index is way in excess of 100%. The attack, hold and decay times of the AGC loop will be set to appropriate values for the mode of reception selected. Thus short time constants will be used for AM reception, where there is (normally!) a carrier providing a continuous indication of received signal strength, but much longer hold and decay times are used in SSB mode. Here, the absence of any carrier results in the disappearance of the signal during pauses in speech: a rate of gain recovery (AGC decay) of 20 dB/s is typical. AGC action generally starts at or a few decibels above the receiver’s rated sensitivity level, which for an HF receiver in SSB mode would typically be 1 V EMF for a 10 dB SINAD (signal to noise-plus-distortion) ratio. This corresponds to an NF (noise figure) of about 15 dB, which is usually perfectly adequate for the HF band, where atmospheric and man-made noise levels are very high most of the time. Some HF receivers boast an NF of 10 dB or even lower: there are rare occasions where this can be useful such as when constrained to operate with a grossly inefficient aerial. An example is operating from a nuclear bunker where the antenna is a very short blast-proof whip or is even buried. Some HF receivers have a stage of RF gain which can be bypassed, or switched in to obtain a lower noise figure when no large signals are present, e.g. on a merchant ship alone in the midst of the ocean, although nowadays, maritime communications are commonly carried via satellite services.

The other main class of receiver includes those designed for constant amplitude signals, such as FM and many types of PM. Here, in principle, AGC is not required, provided that the IF strip is designed as described in Chapter 7 so that each stage limits cleanly when fed with an input as large as its output. However, in the more sensitive receivers, AGC is often incorporated to prevent overload of the early stages, when for example a car radio passes by an FM transmitter: AGC of the RF stage will prevent mixer overload. Generally one cannot successfully apply AGC to mixers themselves. In addition to AGC, FM receivers will also frequently incorporate AFC (see Chapter 8). There remain two other classes of receivers, both dating from the earliest days of ‘wireless’: the homodyne and the super-regenerative receiver. The former has in recent years enjoyed renewed popularity, whilst the latter threatens to proliferate also, with possibly unfortunate results.

The homodyne is a single superhet receiver where the LO frequency is equal to that of the carrier of the wanted signal, so that the IF frequency is 0 Hz. One implementation uses an oscillator with a characteristic similar to that in Figure 9.3d as both the LO and the mixer. The loop gain is adjusted so that the circuit barely oscillates and being very susceptible to outside influences, it is easily tuned so as to become phase locked to the carrier of the incoming signal. This arrangement is also known as a synchrodyne. The modulation of the incoming signal is impressed on the local oscillator and may be recovered with a suitably coupled detector. The upper and lower sidebands of an AM signal are in effect translated down to baseband, and as the oscillator is phaselocked to the carrier (and in phase with it), they lie perfectly on top of each other. The circuit will also receive SSB signals, though in this case there is usually insufficient residual carrier power to take control of the oscillator’s frequency, since in SSB the carrier is suppressed by at least 40 dB relative to PEP (peak envelope power). However, as there is only one sideband, the result is quite intelligible provided the mistuning does not exceed about 10 Hz. (Such mistuning on an AM signal would result in one sideband coming out 10 Hz lower in frequency than it should and the other 10 Hz higher, the resulting 20 Hz misalignment garbling the baseband signals.) The homodyne will also receive CW signals, by off-tuning to one side or the other to provide an audible beat. Similarly, it can translate the two tones of an FSKsignal to baseband, where they can be picked out by appropriate narrow-band tone filters to recover the message information. However, when using the simple homodyne receiver off-tuned like this to one side of the wanted signal, interference may be experienced from an unwanted signal on the other side of the LO frequency. For an FSK signal, a better approach is to tune the receiver exactly half-way between the two tones, which now appear at baseband indistinguishable as far as their frequency is concerned. However, one is a positive frequency and one is a negative frequency relative to the receiver’s LO, and they can thus be distinguished if the sense of their phase rotation is taken into account. To do this, it is necessary to compare the outputs of two homodyne circuits with LO signals in quadrature (Figure 11.3a). Now, if the input frequency is above the LO frequency, the phase of the signal in the upper I (in phase) channel will lag that in the lower Q (quadrature) channel, but it will lead if the input is below the LO. Thus as long as a mark tone persists, a 1 (say) will be clocked into the D flipflop every cycle, and likewise a 0 in the presence of a space tone. The bandwidth of the receiver (which is set by the low-pass filters) need only exceed half the tone separation by a modest margin to allow for the data rate and any possible mistuning, so cut-off frequency of the low-pass filters can be set to say 75% of the tone separation. For even greater selectivity and immunity to interference, band-pass filters could be used. Figure 11.3b shows a complete data receiver suitable for a pocket pager working on this principle: the 90° phase shift between the two local oscillator signals to the mixers is provided by the off-chip 45° lead and lag networks C15,R6 and R7,C13. This system works because in an FSK signal only one tone is present at any one time.

The super-regenerative receiver was developed in the early days of wireless to take advantage of the considerable gain in sensitivity which could be achieved by the use of reaction, where a gain of 50 dB in a single stage is possible. With reaction, a proportion of the RF signal at the output of a tuned RF or leaky grid detector stage is fed back to its input. If carried to excess, the stage will oscillate, so it is essential that its characteristic is rather like Figure 9.3d and definitely not like Figure 9.3b. Unfortunately, considerable skill in adjustment was necessary to obtain the full benefit available from reaction, so many listeners could not master the operation. In the super(sonically quenched oscillator)–regenerative receiver, the loop gain of an RF amplifier with feedback is varied cyclically above and below unity at a supersonic rate, typically 100 kHz (Figure 11.4). This is usually achieved by cyclically varying the current drawn by the active device [2]. There is some similarity to the homodyne, but although the sensitivity is increased greatly, the great increase in selectivity achieved with reaction is not obtained. In the absence of any signal from the aerial, the oscillations which build up during each cycle of the quench waveform start from an initial amplitude determined by the noise level in the input circuit and reach an equilibrium value equal to the steady oscillation level which would prevail if the circuit were not repeatedly quenched. (This assumes the circuit is being used in the usual ‘logarithmic’ mode, rather than the alternative linear mode in which the oscillation is quenched before reaching its equilibrium value.) The oscillations die out when the quench voltage reduces the loop gain below unity. For proper operation, the oscillation must decay to a level below circuit noise before the quench waveform again causes the loop gain to exceed unity. If now a signal above noise level is present within the bandwidth of the tuned circuit, when the oscillations start to build up they start from a larger amplitude than before (Figure 11.4). The oscillations therefore reach equilibrium level earlier and the average current drawn by the active device is increased. The signal modulation thus appears as a modulation of the device current, so the device acts as detector as well as amplifier. The equilibrium level of the oscillation and its subsequent decay are not significantly affected by the presence of a signal. A detailed study of this mode of operation reveals that the change in average device current is proportional to the logarithm of the signal amplitude. Thus the reproduction of an AM envelope with a high modulation index is noticeably distorted. However, the logarithmic characteristic exerts a pronounced limiting action, resulting in a much reduced change of output level between large and small signals – a sort of built–in AGC. It also limits the receiver’s response to impulsive interference, which in any case is less of a problem than with other types of receiver, since a narrow noise spike will be ignored completely unless it occurs during the brief period of build-up of the oscillation - a small fraction of each quench cycle. The logarithmic characteristic also results in a capture effect, whereby when two signals are present simultaneously, the larger controls the build-up of oscillations, almost completely suppressing the effect of the weaker signal. The circuit of Figure 11.4 shows a separate quench oscillator, but this can often be dispensed with, by making the time constant CR long enough to cause the oscillator to ‘squegg’. An oscillator squeggs when operating in a mode where it is self-biasing to class C and the time constant of the self-bias circuit is much too long. The last cycle of the build-up biases the device back to a point where the loop gain is just less than unity and due to the excessive time constant it cannot recover to unity or above before the next cycle. The oscillation therefore dies away completely leaving the device cut off, until the charge on C leaks away and the device turns on again to the point where the gain exceeds unity. In this self–quenched mode of operation, the quench frequency increases when a signal is present. The information carried by the incoming signal can be recovered from the frequency modulation of the quench frequency, see Figure 11.5a (the individual cycles of RF are not fully delineated by the digital storage oscilloscope used owing to the large difference between the quench frequency and the RF). The super-regenerative system thus offers a simple, compact circuit with high sensitivity at very low cost, which has reawakened interest in its use at VHF and UHF as a receiver for applications such as remote garage door opening, car central locking, etc. However, if it becomes popular, problems of interference could arise, as it is impossible to design the circuit so that it does not emit energy at the frequency of the oscillator, surrounded by many sidebands at the quench frequency (Figure 11.5b).