The Superhet PDF Print E-mail
Written by Bryce Ringwood   
Sunday, 14 November 2010 21:50

The Superheterodyne Receiver or 'Superhet' was invented by U.S. Army Major Edwin H Armstrong in 1918 in France during World War 1. Almost all the radios that you have ever seen are superhet receivers. (The exceptions being the crystal set and regenerative receiver projects on this web-site. The fremodyne is a type of superhet.)

How the Superhet Got its Name

A 'heterodyne' is another name for the whistle that was observed in early regenerative radio sets when the reaction control was too far advanced. If you build the regenerative receiver – you will know what a 'heterodyne' is without any doubt! A heterodyne is the audible beat note produced when two radio frequencies are close together. This beat note is actually the difference between the two frequencies. For example if you tune the regenerative receiver to 1486 kHz with the reaction control too far advanced and listen to the AM station 1485, then you will experience a whistle at the difference frequency – in this case 1kHz.

What if the whistle is too high pitched for you to be able to hear it ? - Then it will be 'supersonic' - a 'supersonic heterodyne', which is how the superhet got its name.

The Superheterodyne Principle

In a superhet receiver, the signal reaching the antenna is combined with a locally generated frequency in the receiver. The resulting difference frequency is then amplified by subsequent valve (or transistor) stages before being detected and amplified. The locally generated frequency is produced by a local oscillator.

You can easily verify that two different sine waves separated by a small change in frequency combine to give the difference between the two by plotting them out on graph paper, then adding the ordinates. (or maybe you would rather just take my word for it!).

There is a sort of standard for the difference frequency in superhet receivers (by the way its called the intermediate frequency or IF) In US radios, it is almost always 455kHz and in Domestic British Radios, it is usually 465 kHz. In Europe, an IF of 470kHz is common. These figures must not be taken for granted – it is essential to read the service manual because its very easy to align a radio to the wrong IF.

A Closer look at the Design Principle

A very simple superhet radio might consist of only 5 valves. The first would be a double valve which would provide the local oscillator and mixer stage. The next would be an IF amplifier, the third would be an audio amplifier/detector combined and the final valve would be the audio output stage. The fifth valve would be the power supply rectifier. These sets are sometimes referred to as 'short' superhets.

The block diagram might look like this:

superhet block diagram

Each block in the diagram represents an electrical circuit, and since this is a site about the repair of radios, as opposed to the design, the explanation will not be too detailed, and will concentrate on what might go wrong.

The Local Oscillator

In most cases the local oscillator is a triode valve arranged so that there is a tuned circuit in the anode and a feedback winding to the grid, or it could be a tuned grid circuit with feedback in the anode. The local oscillator tuned circuit usually has a coil whose inductance can be varied. There will be a small trimmer capacitor across the coil and a large value padder capacitor in series with the tuning capacitor.

The first superhets used separate tuning capacitors (they would have called them condensers – the old name) for the aerial circuits and the local oscillator. The operator would tune to the station using the local oscillator capacitor and would then 'peak' the signal using the aerial capacitor. This would be followed by further trimming of the local oscillator.

In order to avoid this three-handed procedure, twin-gang variable capacitors were devised in which the vanes of two separate variable capacitors open and close together. This means that the local oscillator and signal frequencies must keep in step or track each other. Now bear in mind that the signal frequency may be 455kHz below the local oscillator frequency and you can see there's a problem. It turns out that mathematically there can only be three points in any waveband where this tracking can be 100% correct. Depending on the shape of the capacitor vanes, these three points are somewhere near the extreme ends of the tuning range and somewhere near the mid-point. This is why its best to read the service manual for a radio – its all too easy to mess-up these adjustments.

See the calculation section - Padder and Trimmer for more detail.

If the local oscillator isn't stable, the radio will drift off frequency over a period of time. Check the resistors in the local oscillator – particularly any that may be across the tuned circuit or in series with it. If the local oscillator isn't working (broken capacitor ?) the the set will be silent, except maybe for a very very faint hiss.

You can check the local oscillator by receiving its signal on another receiver. If you have a set with a digital readout – you will be spot-on! I discovered that this doesn't always work. Some modern single-chip radios have such an inestimably low local oscillator signal,that it just can't be received.

In communications receivers, the local oscillator is provided with a stable anode voltage from a voltage stabiliser valve, such as an 0A2, 0B2 etc.

The Mixer

Although just about any valve can be used as a mixer, most old radios use hexode, heptodes (pentagrid) such as the 6BE6 and DK91 or triode-hexode valves, such as the ECH81. Sensitive radio receivers might use triode valves, and some even use diodes. The “normal” domestic set will probably use an ECH81 (or ECH83 – which as far as anyone can tell is an identical valve). These multi-grid valves are a little like pentodes with an extra control grid. There are earlier octal and UX7 based valves such as the 6k8 and 6A7. You can see a practical design of mixer/oscillator using the ECH81 on my projects page - two valve superhet.

Frequency changer

In a multi-band radio receiver covering long, medium and short-waves, these stages are by far the most complicated part of the receiver because of all the band switching mechanisms. Usually, the layers of dust settle on the top of the set, leaving all the mechanical parts comparatively clean, except for the tuning capacitor.

In order to clean everything up, I usually start by blowing away the dirt with a compressed air duster. The next step is to use a solvent cleanser to remove grease from the switch and capacitor contacts. Once this is done, the contacts will probably need to be deoxidised with a switch cleaner, such as 'servisol' – a little goes a long way and take care none of these things dissolve the plastic. These can leave the contacts rather sticky, so I usually spray more solvent on to the contacts. At this point everything should work. If it doesn't then the switch wafers may have become bent out of alignment. Some very gentle bending of the outside contacts of the switch wafer with a toothpick usually does the tick. Remember that if you overdo it and something falls off, then the chances of getting the set to work again are rather bleak.

The tuning capacitor should rotate very freely – I use very small quantities of sewing machine oil on the bearings.

Otherwise, its the usual components – decoupling capacitors which cause problems. These short out and take the screen and anode decoupling resistors with them. I am going to defer discussion of the first IF tansformer to IF stages, following.

Finally, for some reason that I don't understand, the signal frequency coils in the grid circuit sometimes go open-circuit. Very often, the set will work, but won't be very sensitive and the circuit can't be aligned (brought into tune).

Early transistor sets used germanium PNP transistors, such as the OC44 and OC171 as mixers or mixer/oscillator stages. They are vulnerable to lightning damage and to other forms of failure (such as tin whiskers growing inside the can). They are not very easy to replace, but it may be possible to use modern silicon PNP transistors, provided the biasing is also adjusted.

The IF Amplifier

IF Amplifier 

This is the preserve of valves such as the 6k7 (octal), 6SK7, 6BA6, EF85 and other variable-mu pentodes. As the AGC voltage becomes more negative, the gain of the IF stage decreases, so that subsequent circuits are not overloaded.

transistor if amp 

If we look at transistor sets, there may be conventional transistors with variable-mu characteristics, but they are few and far between. Usually, AGC is applied to the base of a transistor in a direction which will reduce the collector current. It is also possible to reduce the gain by increasing the collector current – but this will increase battery drain in portable radios. Although field effect transistors have a very desirable variable-mu characteristic, they are not used very much in domestic radios.

As can be seen from the circuit diagram, the circuit consists of a valve (or transistor) with an input tuned circuit in the grid and an output tuned circuit in the anode (or collector). If you are making your own IF stage, the problem is to keep the anode and grid circuits from feeding-back and oscillating. Of course this should never happen in a radio you are repairing – but it does. In the transistor circuit diagram, the grey components are a neutralising cicrcuit, intended to reduce the possibility of unwanted feedback. They may not be present in sets using newer transistors.

As far as faults are concerned, the usual culprits are the decoupling capacitors and their associated resistors. Occasionally the screen resistor and cathode bias resistor give trouble and in any case are easy to check. IF valves are more likely to be missing than faulty.

The worst problem are the IF transformers themselves. Provided they have not been “fiddled with”, they are generally fine and if the set is working with reasonable gain and sensitivity, its best they be left alone. If they haven't been left alone, or if the set really has gone out of alignment (see later) then you may find the adjustment cores have become stuck in their former. You may find a previous service person has cracked the core, split it or rendered the trimmer slot useless. This is almost always because incorrect tools have been used to perform the adjustment. Then there are the small Philips transformers. These are no bigger than some of the IF transformers used in transistor sets and have a brass adjustment screws with a small piece of ferrite glued to one end. More often than not, the glue has come unstuck and the brass screw can only push the ferrite inwards. If everything becomes stuck, then the transformer has to be removed and rebuilt (because in the process of trying to dislodge the ferrite, its easy to damage the windings.). IF transformers often have small polystyrene tuning capacitors across the windings. If these are replaced, I would recommend using good quality silver mica capacitors with a high working voltage to replace them. Silver mica capacitors are stable and can take the large voltage that is developed across the tuned circuits on occasion. They are expensive.

Older sets that don't use ferrite cores in the transformers use capacitors to tune the IF.

Some radios have a means of varying the selectivity of the IF. This is so that on short-wave, adjacent stations don't interfere with one another, providing monkey-chatter in the background. On medium wave, the set can be given its full IF bandwidth, allowing the full sound spectrum to be enjoyed. The IF bandwidth is usually broadened by increasing the coupling between the windings of the IF transformer. Most often this is done using a tertiary winding on the secondary – a very small number of additional turns between the main windings can be switched in. This might be done in conjunction with placing additional resistors across the secondary.

Communications receivers also use the above scheme, but may in addition use crystal or mechanical filters. You may also (rarely) find a ferrite filter. Some radios use ceramic filters and may even dispense with IF transformers altogether, simply relying on a ceramic filter to select the frequency. They may also have a ceramic transfilter in the emitter. The mechanical filters used in some communications receivers can give problems. It doesn't hurt to try and repair them, but don't be too optimistic. They contain a fine wire bobbin at each end of an array of plates. You might need to rewind one or other of these. On the other hand, the filter may just contain decomposed foam. You might have to put a modern filter inside the existing filter can.

The Detector and AGC

The detector stage is usually a diode. Very often this is a semiconductor diode, but may also be a valve, such as a 6AL5 (EB91, EAA91) or 6H6. In the majority of radios, the diode is in the same envelope as a pentode forming the last IF stage (EBF89) or as a double-diode-triode (6SQ7), where the triode forms the first audio amplifier stage. The diode detector simply rectifies the voltage at IF and provides the audio frequency (AF) stages with a signal varying in proportion to the audio exactly as in a crystal set radio.

Detector and 1st Audio

Some radios use a triode “anode-bend” detector, such as a 6C5. In this case, the valve is biased on the bend of its anode characteristic curve, so that incoming grid voltage swings will produce an unequal swing at the plate, effectively rectifying the plate voltage to produce an audio output. This might lead to the additional complication of an additional valve to generate an AGC voltage.

In conventional receivers, the voltage generated at the detector is used to provide AGC. This may be done with the detector diode or with an additional AGC diode. In some radios, AGC is “ delayed” until a threshold is reached, so that AGC is inoperative on weak signals. Of course, the AGC should be a gradually varying voltage and should not have the audio impressed on it. This is achieved by using a combination resistor and capacitor filter. The larger the capacitor, the slower will be the response of the AGC. Clearly, it is essential that all the resistors in the AGC circuit have their correct value, otherwise there could be all sorts of strange problems. The same goes for the capacitors.

The above circuit diagram illustrates a typical detector and first audio stage. Note that the cathode of the triode is returned directly to ground - otherwise the diode would have to overcome the valve cathode bias before doing any detecting – it just would not work. The grid bias for the valve, in the region of -0.7 volts is obtained from the action f the grid leak resistor Rg. The resistor/Capacitor combination RC provides the AGC time constant. The value of R is very large – several Meg with R being about 0.01 – 0.025uF. If the resistors in this circuit change value over time, or the capacitors become leaky, all sorts of undesirable problems can occur. The fact that you may be chasing low voltages in circuits having a high resistance leads to measurement problems. The average digital multimeter has an input resistance of 1 MOhm Attempting to measure the grid voltage on the triode of the detector valve with such an instrument will be doomed to failure. The service manuals often suggest that a valve voltmeter be used. Mine has an input resistance of 11Meg- again hardly enough to provide an accurate reading. I am sorry to say, I don't have any simple answer to this problem, other than to make a voltmeter up using an ICL7106 integrated circuit and LCD display. The range will be 0 – 2.0 Volts with an input resistance of 1 Million Meg.

This is just a very basic version of the detector /audio. Usually there will be additional capacitors and resistors designed to reduce hiss and complications to allow a gramophone pickup to be used. Hold a small metal screwdriver by the shaft and carefully so you don't shock yourself,touch the volume control slider with the screwdriver blade. You will hear a very loud hum from the speaker. If not – you have a problem audio stage.

The Audio Stages

In early radio sets, transformers were used to couple the early audio stages to the subsequent ones in the chain. The primary windings of the interstage transformer provided the load for the anode and the secondary was fed to the grid of the next stage. This made for a very simple circuit. A similar principle was used for driving push-pull output stages.

In time these transformer circuits were replaced by capacitor coupled stages,and by valve phase-splitters for driving push-pull. In most cases the only transformer remaining would be the output transformer used to match the high impedance valve circuits to the low impedance loudspeakers. Philips managed to drive a high impedance loudspeaker directly using two output valves in a series push-pull arrangement.

Audio Amplifier and Power Supply

Nowadays, you will find transistor amplifiers that use DC coupling. This means that all the resistors in the circuit must be the value as marked, with any deviation likely to cause problems. Happily, all this is usually encapsulated in on single audio integrated circuit. (These have been around for a long time – certainly long enough to find their way into sets getting on for 40 years old, such as the Barlow-Wadley XCR-30.) If the marked values of the components have become erased – alas, you will have to find the service manual.

Interstage transformers found a resurgence with the first transistor radios, but nowadays transformerless circuits are the norm. Transformers never were too popular because they are bulky, heavy and can introduce distortion.

The likely culprits in valve audio stages are the coupling capacitors, the interstage tranformers (if any) and the power supply capacitors. Symptoms include hum (power capacitors), motorboating (cathode by-pass caps rarely, power supply caps) and severe distortion caused by leaky coupling capacitors.

Transformers can fail because the paper used to insulate each layer of windings contained acid, which eventually eats away the fine wires forming the windings. The transformers then have to be rewound.

The Power Supply

Very early radios used batteries. There would be 2 volt accumulators for the valve filaments, dry batteries for grid bias and high-tension supplies. If you are lucky enough to possess one of these sets, then you need to construct a power supply.

Nearly all the radios I have worked on use a half bridge rectifier using a valve such as a 5Z3, 6X5, 5Y3 and so on. Take care to replace the rectifier valve with the correct part, because although a 5Y3 will fit a 5Z4 base, it may have a limited and unreliable life. Going the other way and putting a 5U4 into a 5Z4 socket may burn out the transformer windings. Note in the circuit diagram that I have placed a fuse and a double-throw on/off switch. Most sets don't have a fuse.

Now we must consider a horrible practice -the “Universal” AC/DC receiver. In these sets, the valves are wired in series – like Christmas tree lights and the high tension supply is derived directly from the mains. Because the mains lead is often twin flex, the chassis is often connected to mains “line”. A single diode valve or selenium rectifier is used for the HT and there may be dropping resistors in the HT line, the filament supply or anywhere else. Sometimes the mains lead is made of resistance wire to reduce the input voltage to manageable levels. As stated elsewhere, an isolation transformer must be used when working on these sets.

Nearly all valve and selenium rectifiers can be replaced with silicon diodes, such as the 1N4000 series, but beware – a silicon rectifier applies HT when the other valves are cold, so you might want to add a thermal delay switch like Tektronix did on its 'scopes.

Quite a few South African radios I have worked on have a vibrator power supply, similar to those a valve car radio uses. The vibrator is almost always broken, and always because the contacts have burned out. The contacts burn out because of a failure of the snubber circuit – the capacitor and resistor in series across the secondary of the transformer. Perhaps there was a dearth of capacitors with a high enough working voltage. New vibrators are available – but only from overseas suppliers Some of these use solid-state circuitry instead of the mechanical contacts. It is possible to repair old vibrators by carefully cutting round the base with a Dremel diamond wheel (or even a fine hacksaw. The top is removed and then the contacts are cleaned using successively finer grades of emery paper. The contacts should look polished, without any pits or hollows. The contacts must then be adjusted until the vibrator works reliably by bending them into position – gently does it. Once everything is working, the top can be replaced and sealed into place with adhesive tape. Yes, it looks ugly, but I have tried aluminium solder + a 120 Watt soldering iron with no success.

There is a school of thought that believes all power supply smoothing capacitors should be replaced finish and klaar. On the other hand, I had a colleague who believed electrolytics could be reformed by slowly (over a period of many hours) bringing them up to their working voltage. I allow them a period of 30 minutes – after that they're scrapped. Modern replacements are about a 20th of the size of the originals. It is possible to make the whole thing look nice by removing the original caps, putting the new ones inside and sealing the result with epoxy. Normally, I just leave the old capacitors in place and put their modern replacements underneath the chassis.

The only other component that can give trouble is the mains transformer. There was a time when I used to get transformers wound to my specifications by firms such as Gardner's and Harper Electric. Now, I usually wind my own, but I have never tried to wind a mains transformer for a valve radio, but then, I haven't had a valve radio mains transformer burn out on me yet. (Update - a set came to me with a burned out transformer. Peter Souris wound me a new one 100% the same as the existing burnesd out version.)

Mains power supplies for transistor radios follow the same design and circuit principles and the same things go wrong. These little power supplies often use a full-bridge rectifier circuit with a massive smoothing capacitor and possibly a voltage regulator circuit of some sort. Quite often the whole thing is built-in to a “wall-wart” power supply. During periods of ESKOM instability, these can and do burn out. Sometimes these small transformers come apart and can be rewound, but don't bank on it.

FM and AM/FM radios

FM radios have the same circuit principles as AM superhets described so far. The big difference is in the “FM Tuner” and the “FM Discriminator” which replaces the detector in an AM radio. AM/FM sets have two IF transformers in series – one for AM (Narrow) and the other operating on 10.7 MHz (Wide). Valve FM communications receivers often use an IF of 5.2 MHz (Too broad for decent AM and too narrow for decent FM). The principle difference is in the tuner, which is designed for Very High Frequency reception, and the detector.

FM Tuners

FM Tuners often use a double triode valve. The first triode acts as a radio frequency amplifier, and the second triode as a self-oscillating mixer. The RF stage prevents unwanted radiation from the oscillator/mixer escaping back into the aerial and causing interference. It also overcomes noise generated in the mixer stage. In domestic radios, the FM tuner is often in a little box on its own placed somewhere near the AM tuner.

Early FM receivers sometimes used almost an entirely separate FM radio joined at the audio stages. These would use acorn valves such as the 956 pentode as an RF amplifier. The pins for these valves stick out of the side and they are generally easy to break. Examples of these include the EH Scott "Phantom"  a domestic radio and the Hallicrafters S27 and S36 VHF communications receivers.

In an FM radio transmitter, the signal frequency varies in sympathy with the sound being broadcast. Since most interference and noise effectively changes the amplitude of a received signal, FM radios do not suffer from noise and interference that spoils the enjoyment of a medium wave AM transmission. Another feature of FM is the "capture effect", whereby interference from a station on an adjacent channel is vastly reduced compared to AM.


The FM Limiter

An FM signal should have no trace of AM, so good quality receivers precede the FM discriminator with an amplitude limiter. The limiter circuit is similar to a normal IF amplifier, but the anode and screen voltages are kept very low, and the vale is biased ear it cut-off region.(The point where any further negative grid voltage will prevent the valve from conducting.) Signals arriving at the grid will make the valve simply switch from conducting to non-conducting, thus producing a constant amplitude output.

FM Detectors


FM Detectors are almost always made up from a circuit called a ratio detector. In this circuit two diodes are arranged in series with the secondary of the discriminator transformer in between the anode of one diode and cathode of the other. A small tertiary winding is used to pick up a phase difference from the primary of the transformer. When the secondary is correctly tuned, there is no voltage developed between the centre-tap of the secondary and ground when an unmodulated carrier is received. When an FM sinal is received, the voltage across the small tertiary winding adds to the voltage across one side of the secondary and subtracts from the other to produce a larger voltage across one diode and a smaller voltage across the other, producing an AF output.

Foster Seeley Discriminator

The Foster-Seeley disciminator is a circuit used in high-quality radios and in VHF communications receivers, such as the Hallicrafters S27/S36 and Eddystone 770R. This is a phase discriminator circuit which relies on the properties of tuned circuits, so although at first glance it seems obvious how it works, its actual operation is far from obvious.

In a nutshell, the small capacitor from the anode of the discriminator is used to unbalance the phase of the signal at the anode of the two diodes. It does this by adding its in-phase component of voltage at one end of the secondary and subtracting at the other in sympathy with the received signal.

Alignment of a Foster-Seeley discriminator is more difficult than it sounds. To align the discriminator, you need a centre-zero 100uA meter and a signal generator accurately tuned to the IF frequency. Begin by detuning both primary and secondary cores. Next peak the primary on the meter. Now tune the secondary core so that the meter reads zero. Job done. Not. At this point swing the signal generator exactly equal amounts either side of the centre frequency - you should get equal deflections on the meter, but this is unlikeley. Now follows a frustrating round of slightly tweaking the primary, the secondary until it happens the way it should. The best plan is to use a device called a sweep generator with an oscilloscope, but this is not equipment you are likely to have (mine is broken.) The best plan is probably to stop when you have completed the first round and the meter reads zero. Of course, if the diodes are unmatched, it will never work - better have one or two spares on hand!

Early transistor radios use the same circuitry as their valve counterparts (e.g. Kurer Radionette), but after the mid 70s integrated circuits, such as the TBA120S began to be used (and still are in South Africa, where there is a big market for basic FM radios.) These use a coincidence demodulator.

Other FM detectors you may read about are the nonode (EQ80) - I know of no set containing this and the 6BN6 FM detector valve. This may have been used in some amateur equipment. There is also the pulse-counting discriminator used in some home-made designs because no alignment is needed. The Fremodyne uses slope detection i.e the sides of the slope of the IF response curve, giving two tuning positions. Maybe with a bit more experimenting, the quench frequency could have been locked to the FM signal, giving a perfect output. Hmmm.

Of course, nowadays everything is digital, so the FM signal can be processed by a signal-processing chip to give perfect reproduction.

Valve FM Stereo Decoder

A number of valve stereo decoders were made. If you come across one of these, give me a call. I believe Grundig made a valve stereo decoder using a single valve in a reflex circuit that is somewhat of a challenge to comprehend. Now that entire FM radios are on a single chip, stereo decoder chips are quite hard to find.

Note - one of my readers has sent me a veritable compendium of valve stereo demodulators - mostly Grundig. Some of these use only one valve!

Last Updated on Thursday, 08 August 2013 15:22
Joomla template by a4joomla