In a previous section we discussed how the selectivity and sensitivity of a receiver may be improved by adding more tuned RF amplifier stages. In theory half a dozen or more stages might be added until the required performance was achieved, but in practice the need for six sets of tuning coils, six lots of wave-change switches and six tuning capacitors, plus the necessity to keep them in perfect alignment over perhaps three or more wavebands by means of a vast number of trimming capacitors, would make the exercise impossibly complicated. The superhet receiver makes it possible to have eight or more tuned circuits with no more RF tuning components than in a three- tubes (valves) TRE It does this by converting all incoming frequencies to a single one which is passed to pre-tuned amplifier stages that need no adjustment on the part of the listener.
This is achieved by taking advantage of a phenomenon whereby when two different frequencies are mixed together the result is a third, equal to the difference between them. This process is called heterodyning (from the Greek hetero = other and dunamis = force or power). It makes itself apparent in TRF receivers when the reaction control is advanced too far and the detector begins to oscillate at close to the frequency of the incoming station. If this is at 1000 kHz and the oscillation at 999 kHz the result will be a difference frequency of 1 kHz, manifesting itself as a piercing whistle from the loudspeaker.
FIG. 1 Block diagram of the frequency changer stage in a superhet
FIG. 2 Three popular types of frequency changer valve: (a) The triode-pentode;
(b) The triode-hexode; (c) The heptode or ‘pentagrid’.
The full name of the superhet is the supersonic heterodyne receiver, which indicates that the heterodyne frequencies employed are above the range of human hearing; in practice they range from about 100 kHz to 500 kHz. This is called the intermediate frequency (IF). To simplify the workings of a superhet let’s consider one covering just the MW band from 600 kHz to 500 kHz and employing an IF of 450 kHz. Its aerial input and tuning circuitry feeding the grid of the first tubes (valves) would be just like that of the RF amplifier in a three- tubes (valves) TRF, but the tube (valve) involved does more than simply amplify. Associated with it is a local oscillator, the job of which is to generate oscillations over a band of frequencies exactly 450 kHz higher than that of the incoming frequency; for instance, if a station on 1000 kHz is to be received the local oscillator will produce 1000 kHz + 450 kHz = 1450 kHz. When this is mixed with the incoming frequency the result is a heterodyne at 450 kHz. The tuning capacitor for the local oscillator is ganged with that of the RF tuning so that if the latter is adjusted to tune in 1100 kHz the local oscillator will produce 1550 kHz, and so on.
The local oscillator itself may employ a separate tubes (valves) or the RF tubes (valves) may be made to self-oscillate.
FIG. 3 Basic frequency changer stage employing a triode-hexode valve
However, by far the most popular arrangement is to use a special frequency-changer (f.c.) tubes (valves) with yet more electrodes than a pentode. One popular tubes (valves) has no fewer than five grids and is therefore called a heptode or pentagrid. Numbering these conventionally from that nearest the cathode, G1 acts as control grid of a triode with G2 as its virtual anode; these are employed as a local oscillator. G3 and G5 are a pair of screen grids connected together and spaced either side of G4, the main control grid.
The oscillations produced by G1 and G2 modulate the electron stream from the cathode on its way to G4 and thus frequencies received from the aerial mix with those of the oscillator to produce at the anode the IF.
Another tubes (valves) working on this principle is the octode, in which an extra grid (G6) is placed between G5 and anode, where it acts like the suppressor grid in a pentode.
An alternative type of frequency changer is in fact a double tubes (valves) incorporating a triode section to act as local oscillator. It shares a common cathode with the receiving section, which in early examples was a pentode, where the mixing of oscillations takes place via the common cathode. A later improvement was the triode-hexode, the latter section having four grids. Of these, G2 and G4 are screen grids. According to individual design either G1 or G3 may be the control grid, leaving the other to act as injector grid; connected internally to the grid of the triode, it modulates the electron stream as in a heptode or octode. The triode-heptode has a fifth grid, G5, which acts like the suppressor grid in a pentode.
Keeping the local oscillator in step
FIG. 4 Simplified diagram of how trimmer and
The most popular method is to use a standard two gang variable capacitor for tuning both RF and padder capacitors are wired in a local oscillator. The trimmer is always adjusted at the high frequency end of a band and the padder at the low frequency end local oscillator coils, with series and parallel capacitor added to the latter to effect adjustments at the low and high frequency ends of each band. An alternative is the use of a two-gang tuning capacitor with one normal set of plates for RF tuning, and for the oscillator a smaller, specially shaped set which automatically keep the two in step. Tracked tuning capacitors, as they are called, appeared in only a few sets, probably because they were dearer to produce. Whatever type of frequency changer is employed, the end result is always an IF appearing at the anode of the mixer section. This is coupled by special intermediate- frequency transformers (IFTs) to the grid of an IF amplifier. The transformers, having to handle only one frequency, may be made to work with higher efficiency than ordinary tuning coils which have to cover a wide range of frequencies. Two main types are to be found, using either air-cored or iron-dust coils. The first type have small trimmer capacitors to bring them into alignment, whilst the second have small fixed capacitors shunted across them with the dust core movable to effect the tuning. In either case the response curve may be made to approach the ideal for good station separation consistent with adequate sideband coverage.
Another 1FF couples the IF amplifier to either a second amplifier in expensive receivers or, more commonly, to the detector. Although early superhets used a triode in either grid-leak or anode-bend mode, they were very soon supplanted by the diode, with its ability to rectify strong signals without distortion. The diode may be a separate tubes (valves) or, more likely, it will be incorporated into another, often a triode used as first AF amplifier.
A specimen diode detector circuit is shown. C3 is an IF by-pass capacitor and R4 acts as an IF filter with C1. Typical values for the two capacitors would be 50 pfd and for the resistor 47 k-ohm Although C3 returns directly to chassis and C1 via the cathode of the valve, the effect is the same since the impedance of the cathode by-pass capacitor is close to zero at IF R1 is the load resistor for the diode. This is a necessity for any diode detector and its value will be 500 k-Ohm or more. Note that it does not return to chassis but to the cathode of the valve. The reason for this is that the diode anode is then held at the same potential as that of the cathode. If it were to return to chassis it would receive an effective negative bias equivalent to the voltage appearing across R3. This in turn would prevent it from conducting until the voltage of the incoming signal exceeded that of the bias. The effect would be to restrict reception to strong stations only, with all others suppressed.
FIG. 6
FIG. 5 Air-cored IF transformer
However, bias is necessary for the triode section of the valve, so its grid must return to chassis, which it does via the potentiometer R2, which also acts as volume control by enabling any desired proportion of the total AF voltage across it to be passed to the grid. C2 permits the AF signals from the diode to travel freely to the top of R2 but blocks the DC voltage from the cathode.
The circuitry of a superhet following the detector and first AF amplifier is precisely the same as in a TRF and needs no further discussion.
Automatic volume control
Even a simple superhet with one stage of IF amplification following a mixer can have greatly improved sensitivity over a TRE This could make it rather uncomfortable for the listener when turning along a wave band as, if the volume control were turned up to receive weak stations, strong ones would be deafeningly loud. Thus tuning had to take place with another hand ready to operate the volume control, not a particularly convenient or attractive proposition as a sales point. Another problem was that distant, weaker, stations tended to fade, that is their signal strength varied, especially at night on MW. With automatic volume control (AVC) the gain of the receiver is controlled by the strength of the signal received and thus automatically adjusts itself to weak or strong stations or to fading. In the 1940s there was a tendency to replace the term AVG by automatic gain control (AGC) and both may be found in vintage radio literature; they may be taken as synonymous. Once again the humble diode carries out the work. We have seen previously how the effect of rectifying a signal results in a steady DC voltage plus the AF modulation; in TRF sets only the latter is required and the DC voltage dumped. However, since this voltage is both negative with respect to chassis and is directly proportional to the strength of the incoming signal it can be used as grid bias for the mixer and IF amplifier tubes (valves) when these are variable-mu types. In fact, this is exactly what was done in many cheap sets, AVG bias simply being tapped off from the signal diode load resistor. Simple AVC, as it is called, works well enough on local stations but it is far from ideal for long distance reception. The reason for this is the fact that any level of signal at the diode will start to generate AVG bias and thus reduce the gain of the set, even though this is undesirable for weak stations. Delayed AVC is designed to prevent this happening by preventing AVG bias from being developed until and unless the signal strength of a station is adequate to give full loudspeaker volume. This entails the use of a separate diode to rectify the carrier wave purely for bias purposes; known as the AVC rectifier or rectifier, it is again usually incorporated in the envelope of the first AF amplifier, as shown in FIG. 8.
FIG. 7 Simple AVC derived from the negative DC voltage appearing on the detector
diode load resistor
The anode of the AVC diode is fed with the IF signal via C1, typically 100 pfd. At one time it was common practice to obtain the signal from the detector diode anode, as shown, but research showed that this tended to introduce AF distortion, so the preferred method is to couple the AVG diode to the anode of the IF amplifier. R2 is the usual cathode bias resistor for the triode section, whilst R1 is the load resistor for the AVG diode anode, typically about 1 m-Ohm. As the resistor returns to chassis, the diode anode is biased negatively by the amount of voltage existing across R2. The set designer knows how much signal voltage is required to obtain full volume at the loudspeaker, so he arranges for the AVG rectifier anode to be biased to the same amount; e.g. if the required signal voltage is 3 V the bias also will be 3 V. Thus the value of R2 is chosen to give that amount of cathode bias. In some cases a high degree of delay is required which cannot be obtained in this way, so it is ‘borrowed’ from elsewhere in the set, such as a negative smoothing circuit (see previous section). We shall look at advanced AVG systems in a later section. R3, typically the same value as the load resistor, filters out any remaining IF on the diode anode, with C2, typically 0.1 uF as a decoupling capacitor. In this circuit only the IF amplifier is shown as receiving AVG bias but in practice it would usually be fed to the frequency changer and RF amplifier, if used.
An alternative delay method
Instead of using a separate diode for the AVG with bias applied to ‘delay’ its operation, some firms, especially in the USA, took the AVG bias from the detector diode, as in simple AVG systems, and then used a ‘clamp’ diode to prevent it from coming into effect before a certain level was reached. Basically, the clamp diode is simply shunted across the AVG line and a small amount of positive bias is applied to its anode to make it conduct and thus effectively to short out the AVG until its negative voltage exceeds that of the positive voltage on the diode anode. When this occurs the diode ceases to conduct and the AVG takes control of the tube (valve). The positive bias for the clamp diode is usually obtained from a voltage divider across the HT line, with the resistor connected to HT+ being of high value, typically between 4.7 M-ohm and 22 M-ohm. One of the advantages of this system is that the delay bias, no longer depending on the cathode voltage of a double-diode-triode, readily can be set at any voltage the designer wishes.
A further refinement of the method was designed to save the use of an extra diode to provide the clamp action. It was found that the suppressor grid of an RF pentode in the IF amplifier stage could be used as a virtual diode, and one instance of this being exploited appears to have been in the well-known ‘Wartime Givilian Receiver’ (1944). A year or two later Mullard produced a miniature RF pentode incorporating a single diode (EAF42IUAF42) expressly designed for this kind of work, the diode being used as detector and to provide AVG and the pentode suppressor grid used to provide the delay.
The ‘short’ superhet
By the mid 1930s advances in tubes (valves) technology had made possible the production of steep-slope mains output pentodes having Gm figures of up to 10 mA/V. Their high sensitivity enabled them to be driven directly from the detector in a superhet without an intermediary AF amplifier (or, at least, this was the intention, not always realised as we shall see later). A natural extension of this was to incorporate the detector diode, plus an AVG rectifier, into the same envelope, thus producing the double-diode-output pentode (DDP). Superhets using just three tubes (valves), frequency changer, IF amplifier and DDP, plus the necessary rectifier, soon began to be called short superhets to distinguish them from more conventional designs. The chief exponent of the genre was the Ultra Electric Go., and the circuit of a typical example of one of its many models is shown in the appendix to this section.
A good basis for study
Although many hundreds of different superhet designs must have appeared through the vintage years, some of them with extra features such as an RF amplifier preceding the frequency changer, or two or more IF amplifiers, probably about 85% of them will differ only in detail from the two basic types discussed above. As long as you have a good grasp of the principles involved you should be able to work your way without difficulty through almost any circuit.
FIG. 8
Reflex amplifiers
Despite the claimed ability of double-diode-pentodes to work directly from the detector the lack of an AF amplifier did make itself apparent in some receivers. To make amends for this a device called reflexing was adopted. This makes use of the ability of a tube (valve) to amplify both HF and AF simultaneously without mutual interference. The usual arrangement is to have the AF signals from the detector fed back to the grid of the IF amplifier via the secondary of the first IFT, along with the AVG bias. Amplified AF signals may be taken either from the anode of the IF tubes (valves) or from its screen grid acting as a virtual anode. When the latter method is used the value of the decoupling capacitor on the screen grid has to be chosen carefully to ensure that it bypasses currents at IF effectively without affecting the AF signals.
Part superhets
This description refers to sets which have a frequency-changer stage as in a normal superhet, but instead of an IF amplifier have a grid-leak detector usually with reaction. They do not have AVG and use the same sort of volume control system as in TRF receivers. It was a cheap way to be able to make a set that could be sold as a superhet but which cost little more than a TRF, with a marginal improvement in performance. A.G. Cossor Ltd was the main exponent of this genre, some acting as a superhet on long, medium and short waves but with others operating as a superhet only on short wave and as a TRF on medium and long. The part superhet appeared first around 1936 and continued to crop up now and again over the next 20 years, virtually to the end of the tube (valve) era. It is most likely to be found in cheap ‘midget’ sets (see T43DA at the end of section).
Battery superhets
Equivalent battery tubes (valves) were produced for all the mains types discussed above (with the exception of the high slope double-diode-pentode) and most receiver manufacturers produced battery powered superhets. They follow exactly the same principles as their mains equivalents except, of course, that the LT, HT and GB supplies conform to those already discussed for battery TRF sets. There is thus no need to enlarge upon their design in the RF amplifier, frequency changer, IF amplifier and output stages, but we need to look at the detector and AGG stages, where some unusual tubes (valves) may be encountered.
From about 1935 onwards the vast majority of battery superhets used diodes for detection and AGC, mostly in the same double-diode-triode package as found in mains sets, except, of course, that they lacked a cathode. The circuitry is very similar apart from the methods of biasing the triode grid and the diode anodes. The Ever-Ready 5030 affords a typical example. The grid is returned, via the bottom end of the volume control R22, to a battery tapping 1.5V negative with regard to chassis. The signal diode is isolated from the volume control and this voltage, by C12 and Rio, returns it to LT+. The AGC diode returns to chassis via R14 and Ri5 and so receives a modest amount of bias by virtue of the filament being slightly positive. This would delay the AGC action to a small extent, but not enough to satisfy some designers. For their benefit some tubes (valves) manufacturers introduced an almost unique device — a battery powered double diode which was indirectly heated.
Two alternative circuits taking advantage of this tubes (valves) are those of the Cossor 376B and the Decca PT/ML/B (or PT/B). Cossor used a simple potential divider across the HT supply to provide approximately 6 V on the cathode of their 220DD valve. The signal diode was returned to the latter via, again, the volume control, and the AGC diode to chassis via a 1 M-ohm resistor. The delay voltage was thus that of the cathode, permitting better reception of weak stations.
Decca plumped for cathode bias derived from an ordinary grid bias battery, the connections of which are a little confusing at first sight. It is, in effect, split into two sections, the chassis being connected to the —6V socket and the cathode to the positive. The —9 V tapping goes to a potential divider network consisting of R15, R16 and R17, which provide different levels of bias for the various tubes (valves). The diode cathode returns to the junction of R15 and R16, so the actual delay voltage is that found at the junction, some 0.75 V added to the 6V from the battery, giving rather more delay than in the Cossor receiver. The tube (valve) used in the Decca was the Mullard 2D2, equivalent to the 220DD, but with a filament current of only 0.09 A as against 0.2, this being kinder to the LT supply.
Appendix
--- GEC Model BC4050
The GEC Model BC4050 was released in March 1939. It is an absolutely typical circuit of a four tubes (valves) plus rectifier AC only superhet of the latter half of the 1930s, a design which continued right through until the end of the tube (valve) era 20 years later and as such it contains many of the features discussed in the text. It covers long, medium and short waves; note that the aerial input and RF tuning coils are little different from those of a TRF with the exception that the bottom of the long wave coil does not return to chassis but to the AVC line.
The frequency changer is a triode-hexode and the local oscillator has been simplified by the use of a single anode coil for all three bands. Note the padding and trimming capacitors associated with the oscillator grid coils.
The IF is 456 kHz, and the pentode IF amplifier is followed by a double-diode-triode action as detector. AF amplifier and AVC rectifier. Note the very comprehensive IF filtering in the diode load circuit. Further filtering takes place at the anode of the triode, which is resistance-capacity coupled to the grid of the pentode output valve.
There is a fixed tone correcting capacitor across the primary of the output transformer and a continuously variable tone control consisting of C32 and P31. The power supply section uses a conventional full-wave rectifier with a mixture of choke and resistance smoothing. Note that one side of the LT winding is connected to chassis.
All the tube (valve) have octal bases, which first appeared in the USA Liz 1935 and swiftly spread across the Atlantic. This type of base nominally has 8 pins for electrode connections (although not all installed for the simpler tubes (valves)), set around a locating spigot which ensures that it is inserted into its holder correctly. The receiving tubes (valves) have 6.3 V heaters and that of the rectifier is rated at 5 V
--- Cossor Model 57
The Cossor Model 57 appeared in 1942, which was rather surprising since not only had domestic radio production virtually ceased by that time, A. C. Cossor Ltd was a major contractor to the Government for radar equipment. The 57 is a four- tubes (valves) plus rectifier AC/DC having many of the features discussed in the text, and may be considered as very typical of this class of receiver, with the sole exception that as long wave broadcasting by the BBC had ceased on the outbreak of war the set covered only medium and short waves.
Following the growing trend for transportable receivers, the medium wave RF tuning coil was actually wound as a frame aerial, attached to the back cover for the cabinet. A triode-hexode frequency changer is used and the IF is 465 kHz. The IF amplifier is an RF pentode.
A double-diode-triode is used as detector, AF amplifier and AVC rectifier: note that its cathode is strapped to that of the output pentode. This was a fairly common device to have both tubes (valves) HT current combined to produce a larger voltage drop than would have been the case with just one. This voltage is fed back via the detector load resistor to the diode anode, thus preventing it from being negatively biased.
The delay bias for the AVC rectifier is obtained in a very interesting manner. The two resistors R15 and R16 are in series between the HT negative line and the chassis, and carry the entire HT current drawn by the tube (valve). The resulting voltage drop across the two resistors is used to bias the grid of the output pentode, whilst the reduced negative voltage at their junction is fed back to the AVC diode anode and also along the AVC line to the grids of Vi and V2. This means that cathode bias is unnecessary and that their cathodes may be returned directly to chassis, saving the cost of two resistors and decoupling capacitors.
Although V5, the rectifier, has two sets of anodes and cathodes these are strapped and the tube (valve) operates in the usual half-wave mode associated with AC/DC sets. Note the two-stage resistance smoothing for the HT R2i, the main dropping resistor for the heaters is in this set incorporated in the mains lead. By itself it is suitable for mains inputs of 200/225 V while the addition of R20 permits supplies of 225/250 V to be used.
--- Ultra model 25
The Ultra model 25 is absolutely typical of the ‘short’ superhet and of Ultra’s products in general, for the firm used this type of circuitry extensively. The sets were not usually very exciting but they were well designed and made.
Note the use of band-pass tuning preceding the triode-pentode frequency changer, V1. The local oscillator is well designed with iron-dust cored coils and fixed padders (C6 and C7). The IF is 456 kHz and air-cored IFTs are used. The IF amplifier (V2) is an RF pentode; ignoring the usual practice its suppressor grid is returned to chassis and not to cathode. Resistors R9, R12 and R13 in the anode circuit feed the neon lamp tuning indicator.
The detector diode load includes an HF choke to filter out any residual IF and is returned to the cathode of the pentode section to avoid its anode receiving negative bias. Note that the cathode resistor is made up of two (R1 and R20) in series; this is because the bias voltage required for the pentode grid is lower than that needed for the delayed AVC. Thus the grid is returned via the volume control to the junction of the two resistors whilst the AVC diode load returns to chassis. Ultra was ahead of many other makers in feeding the anode of the AVC diode from that of the IF amplifier, a much better idea than feeding it from the detector diode.
Due to the omission of an AF amplifier before the output stage ‘short’ superhets were always plagued with low gain when a gramophone pick-up was used. In this set a step-up transformer (T1) is fitted to mitigate this problem.
The load for the AVC diode also consists of two resistors (R21 and R22) in series, with the bias for V1 being taken from the top and that for V2 tapped off at their junction. The effect of this is that V2 receives about three-quarters of the bias voltage whilst V1 has the full amount.
Note the use of two ‘top cut’ capacitors (C15 and C16, the latter in series with R17) across the primary of the output transformer (T2), with a third capacitor (C14) able to be brought in by closing S8 if desired by the listener to decrease still further the treble response.
An energized loudspeaker is used, the field of which smooths the HT provided by the indirectly heated full-wave rectifier. An unusual feature was the inclusion of a separate LT winding on the mains transformer for the dial lamps. Capacitors C2i and C22 in series across the mains transformer primary with their junction taken to chassis, suppress mains-borne interference and ‘modulation hum’.
--- Marconiphone model T43DA
The Marconiphone model T43DA was one of the very last receivers to be made by EMI before it closed its radio and television division. A part-superhet transportable AC/DC set, it was arguably the worst design ever produced by the firm, a sad and uncharacteristic ending to a glorious era. However, it stands as an example of many of the design features mentioned in the text, and of how not to make a radio set.
A ferrite rod aerial is used, carrying the medium and long wave tuning coils which are coupled by C4 to the control grid of the triode-heptode frequency changer: its grid is returned to chassis by R3. The tuning knob is attached directly to the spindle of the variable capacitor VC1/VC2 without benefit of slow-motion drive and no dial lamp is fitted.
The triode section of this tubes (valves) is employed as the local oscillator designed on the cheapest lines possible — just one HF transformer basically for medium waves but with the grid coil shunted by C9 to give LW coverage. Vi has to operate without any bias at all, there being neither cathode resistor nor AVC.
The single IF transformer feeds signals to V2, an RF pentode used as a grid leak detector with reaction, TC3 being a preset control for the latter. The resulting AF at its anode is resistance-capacity coupled to the grid of the output pentode, V3, with RF filtering by C15 and C14. No proper AF volume control is fitted: instead, using technology from a by-gone era, this is affected by varying the HT voltage on, simultaneously, Vi anode, screen grid and oscillator anode, and on the screen grid of V2. It is a minor mystery how the volume control potentiometer, RV1, managed to carry all this current without over-heating.
V4 is the half-wave rectifier with HT smoothed by the over-wind on the output transformer TR1, resistors R12 and R11 and capacitors C21, C18 and C17.
The big question about this set is, why would anybody do it on purpose?
--- THE BUSH MODEL AC1 (post-war version)
The Bush model AC1 (post-war version) was a part AC-only set using an auto-transformer to provide the voltage for the rectifier anode, tubes (valves) heaters and dial lamp. It would thus work only on AC mains but the heaters and dial lamp were operated in series and the chassis was connected to one side of the mains. The use of a single-pole on/off switch meant that even with the mains plug inserted correctly, so that the neutral was connected to chassis when the set was working, the path through the auto-transformer and heaters back from the live main to chassis meant the latter became ‘live’ when the set was switched off Thus this receiver had all the disadvantages of an AC/DC set without its advantages, and none of those of a true AC-only set. As such it was arguably an exercise in futility because the auto- transformer must have been considerably dearer to produce than a mains dropper. It did run much cooler than an AC/DC set, however, for what that was worth in a fairly spacious table cabinet. Perhaps Bush considered that potential customers would believe that a set that would work only on AC was intrinsically better than an AC/DC type. To cater for people on DC mains, however, a companion set, the DAC1, was fitted with a mains dropper and thus became suitable for either AC or DC. Strange are the workings of salesmen’s minds. The two alternative power supply stages are shown side by side (the frequency-changer stage is omitted for clarity). In either case only resistance-capacity HT smoothing was employed.
Bush exploited this bizarre fantasy about ‘AC-only’ and AC/DC models in a series of other receivers, so presumably it must have been commercially successful.
Another feature of the AC1/DAC1 was something Bush called ‘bi-focal tone’, a device which was supposed to give the same fidelity of reproduction at high or low settings of the volume control. It was in fact negative feedback from the cathode of the output tubes (valves) (note the absence of a by-pass capacitor) to the bottom of the volume control via R16 and C24. It was an interesting idea but it was soon quietly dropped.