Chapter 24: The Superheterodyne Receiver

It cannot reject an image on the other side of its LO

A DC receiver turns signals on both sides of its LO into audio, so a signal where the other sideband would be cannot be filtered out. Its selectivity, from audio filters, is actually very good.

Preselector, RF amplifier, mixer, IF filter

The preselector filters the antenna signal first, the RF amplifier lifts it a little, the mixer converts it to the IF and the IF filter selects. The manual puts the filter, not the amplifier, straight after the antenna.

The first mixer

The first mixer produces sum and difference frequencies, one of which is the IF. The product detector is a mixer too, but it combines the IF signal with the BFO to give audio.

Rejecting signals on the image frequency

An image signal reaching the mixer would produce the IF just as the wanted signal does, so the preselector must stop it first. Keeping out broadcast stations was the job of the direct-conversion receiver’s input filter.

Strong signals could otherwise overload the mixer

Too much gain ahead of the mixer lets strong signals overload it, causing intermodulation; some designs omit the RF amplifier. In a direct-conversion receiver the RF amplifier feeds the product detector instead.

Higher, as IF stages have improved

Improved filters and IF stages now work well at higher IFs, which also put the image further from the wanted signal. A lower IF would bring the image closer.

A filter using quartz crystals

Modern superhets typically use quartz crystal filters at the IF, which can be made very narrow with steep sides. LC filters of coils and capacitors suit the preselector, where the signals to be rejected are far away.

3,145 MHz or 4,055 MHz

The LO must differ from the signal by the IF: 3,600 − 0,455 = 3,145 MHz or 3,600 + 0,455 = 4,055 MHz. Twice the IF away, 2,690 and 4,510 MHz, are where the two images would lie.

A VFO is easy to make stable at that frequency

A stable VFO is easily made around 5 MHz. The 5,2 MHz LO actually puts the image closer, at 3,8 MHz; a 23,2 MHz LO would put it at 32,2 MHz, 18 MHz from the signal.

9,0 MHz and 16,4 MHz

A mixer gives the difference, 12,7 − 3,7 = 9,0 MHz, which a 9 MHz IF filter keeps, and the sum, 3,7 + 12,7 = 16,4 MHz, which it rejects. 21,7 MHz is where this receiver’s image lies.

14,060 3 to 14,063 0 MHz

Each signal mixes down by 5,060 0 MHz, so 14,060 3 to 14,063 0 MHz lands on 9,000 3 to 9,003 0 MHz: the upper sideband of 14,060 MHz. The lower sideband mixes to just below 9 MHz.

It mixes to 8,998 0 MHz and is rejected

14,098 0 − 5,100 0 = 8,998 0 MHz, below the 9,000 3 MHz edge, so the IF filter rejects it. Landing on 9,002 0 MHz would need a signal on 14,102 0 MHz.

The bandwidth of its filters, mainly at the IF

Selectivity is set by bandwidth: the narrower the passband, the closer an unwanted signal can be and still be rejected. In a superhet the IF filter does most of this; gain makes signals louder, not better separated.

Most of the receiver’s selectivity and gain

At one fixed frequency the IF filter can be made very narrow and the IF amplifier given high, stable gain, so the IF stages supply most of both. The image must be rejected earlier, by the preselector.

A narrower bandpass filter at the IF

Selectivity comes from a narrow passband, such as a crystal bandpass filter passing only a few kilohertz around the IF. A lowpass filter passes everything below its cutoff, so it cannot reject signals just below the wanted one.

After mixing, it lands on the IF just like the wanted signal

The image produces the same IF as the wanted signal, so once they are mixed no filter can tell them apart, however narrow. That is why the preselector, ahead of the mixer, must remove it.

The back end of a direct-conversion receiver

After the IF filter come an IF amplifier, a product detector fed by the BFO and an audio amplifier: a direct-conversion receiver’s chain, fed from the fixed IF. A TRF receiver has no BFO or product detector.

A carrier insertion oscillator

The beat frequency oscillator (BFO), which replaces the missing carrier at the product detector, is also called a carrier insertion oscillator. The local oscillator is the tunable one that feeds the first mixer.

A product detector

A product detector mixes the IF signal with the BFO, replacing the missing carrier, so SSB comes out as clear speech. An envelope detector relies on a carrier in the signal, so it suits AM but garbles SSB.

SSB and CW signals

Beating the IF signal with the BFO turns SSB back into speech and makes a CW carrier into an audible tone, so one product detector serves both. FM needs a discriminator instead.

To recover the modulation, such as speech

Detection, or demodulation, recovers the information, such as speech, that was used to modulate the carrier. Amplifiers raise the level, filters select and the local oscillator sets the tuning.

Down by 700 Hz

Moving the LO down by 700 Hz puts the CW signal on 9,000 7 MHz, inside the passband and 700 Hz from the BFO. Moving it up gives 8,999 3 MHz, also 700 Hz away but outside the filter.

Switch in a second filter, 8,997 0 to 8,999 7 MHz

With the LO and BFO left alone, an LSB signal lands just below 9 MHz, from 8,997 0 to 8,999 7 MHz, so a second filter there passes it. A filter above 9,003 MHz would miss it entirely.

LO 5,246 7 MHz, BFO 9,003 3 MHz

The LSB, 14,247 0 to 14,249 7 MHz, must land in the passband: LO 5,246 7 MHz. The carrier then falls at 9,003 3 MHz, where the BFO goes; a 9,000 0 MHz BFO inverts the audio.

Such filters are expensive to buy

The narrow crystal filters used at the IF are costly, so one filter serving both sidebands, with the LO and BFO moved, saves money. One preselector serves the receiver whichever IF filter is in use.

Treat it as SSB, using one sideband

Unchanged, the receiver treats AM like SSB: the IF filter keeps one sideband, drops the carrier and the other sideband, and the BFO supplies a carrier. Without the BFO, a lone sideband would give no proper audio.

The audio stays faithful even if slightly mistuned

With the carrier and both sidebands passed, envelope detection reproduces the original audio frequencies even when the receiver is a little off tune. A 6 kHz filter is wider than the SSB filter, so selectivity gets worse.

451 to 459 kHz

AM has a sideband on each side of the carrier, each as wide as the audio, so it needs 455 kHz ± 4 kHz: 451 to 459 kHz, 8 kHz wide. Passing only one side would suit SSB, not AM.

Tracking

Tracking keeps the LO one IF from the RF circuits’ tuning at every dial setting; ganging is just the mechanical link that turns them together. The manual names tracking only in revision question 10.

8 010 kHz

The LO is on 7 100 + 455 = 7 555 kHz, and the image lies one IF beyond it, at 8 010 kHz: twice the IF above the signal. 6 190 kHz would be the image with the LO below.

A signal and its image

The settings are 910 kHz apart, twice the 455 kHz IF, so at one the station mixes to the IF directly and at the other as an image. Intermodulation needs two strong signals mixing together.

3,250 MHz

The LO is on 21,250 − 9 = 12,250 MHz, and the image is the other frequency 9 MHz from it: 12,250 − 9 = 3,250 MHz. 39,250 MHz would be the image with the LO above the signal.

The RF and IF amplifiers

The block diagram shows the AGC detector sampling the IF signal and setting the gain of the RF and IF amplifiers. The RF and audio amplifiers are the ones controlled in the direct-conversion receiver.

It reacts faster to sudden changes

IF-derived AGC reacts more quickly than audio-derived AGC, so a sudden strong signal is tamed before it blasts the listener. S meters stay only roughly calibrated, however the AGC is derived.

A high IF eases image rejection; a low IF eases selectivity

A high IF puts the image far from the wanted signal, where the preselector can reject it, while narrow filters are easiest to make at a low IF. Swapping the two gets both wrong.

A high intermediate frequency

The image lies twice the IF from the signal; at VHF a 455 kHz IF would leave it under 1 MHz away, too close for the front-end filters. A high IF moves it many megahertz off.

Such an IF already keeps the image far enough away

With a 9 MHz IF the image lies many megahertz from an HF signal, which a simple preselector rejects easily, and crystal filters there give the selectivity. A higher IF moves the image further away, not closer.

24 times

The image is twice the IF from the signal, so its distance grows with the IF: 21,4 ÷ 0,91 = 10,7 ÷ 0,455 = 23,5, nearest 24 times. Doubling only one of the two IFs gives 12 or 47.

A dual-conversion superhet

Two intermediate frequencies, 45 MHz and 455 kHz, mean two frequency conversions: a dual-conversion superhet with two IF stages. The detector’s final step down to audio is detection, not counted as a conversion.

It tunes any frequency over a wide range, not just the bands

General coverage means tuning continuously over a wide range, typically 500 kHz to 30 MHz and sometimes up to 1 GHz, including the frequencies between the amateur bands.

An imperfectly balanced first mixer

An imperfectly balanced mixer lets some incoming RF straight through to its output, so a signal already on the IF passes the IF filter. More RF gain makes a leak louder but is not its cause.

Its IF can avoid every band, so RF on the IF can be filtered out

With an IF such as 8,5 MHz, well away from every amateur band, the preselector, perhaps helped by a notch filter, keeps signals on the IF out of the mixer. A leaking signal on the IF passes any IF filter.

It might need to receive signals on that very frequency

With the IF inside the coverage, notching it out at the input would deafen the receiver there, so such receivers put the IF outside their range. An input notch removes incoming RF, not the converted IF signal.

A spurious response

Any unwanted response, such as to a signal leaking through at the IF or to an image, is a spurious response. Poor selectivity means failing to separate close signals; leak-through can come from far off.

Its first IF lies above its whole tuning range

The signal is first converted up to an IF above the coverage, such as 60 MHz for 0,5 to 30 MHz, which keeps both images and IF leak-through well away. A lower second IF then provides the selectivity.

45,5 to 75 MHz

The synthesiser sits one IF above each signal: 0,5 + 45 = 45,5 MHz up to 30 + 45 = 75 MHz. Adding twice the IF gives 90,5 to 120 MHz, the range where the images lie.

12 to 15 kHz

A roofing filter must pass every mode the receiver handles, so with FM included it is about 12 to 15 kHz wide. The narrow 2,4 kHz SSB filter comes later, at the second IF.

To keep strong nearby signals out of the second mixer

The roofing filter rejects strong signals close enough to pass the preselector that could cause intermodulation or images in the second mixer. The final SSB selectivity comes from a narrower filter at the second IF.

At the second IF, after the roofing filter

Narrow filters are easier to make at the lower second IF, so the mode filters go there. The first IF has only the wider roofing filter, which lets every mode through.

53,3 MHz

With the oscillator below the first IF, it must sit one second IF lower: 64 − 10,7 = 53,3 MHz, as the manual’s receiver uses 60 − 51 = 9 MHz. 74,7 MHz would also give 10,7 MHz, but from above.

The circuits from the antenna up to the IF filter

The manual uses front end for everything between the antenna and the IF filter: preselector, RF amplifier, mixers and oscillators. The stages after the IF filter are much like a direct-conversion receiver’s.

170 MHz

The LO is on 144,3 + 10,7 = 155,0 MHz and the image another 10,7 MHz up, at 165,7 MHz, so 170 MHz is nearest. 155 MHz is the LO itself, only one IF from the signal.

150 MHz

The LO is on 7,1 + 70 = 77,1 MHz and the image a further 70 MHz up, at 147,1 MHz, nearest 150 MHz. 75 MHz is close to the LO itself, only one IF from the signal.

A noise limiter

Ignition interference arrives as short, strong pulses, which a noise limiter clips. A low-noise preamplifier lowers the receiver’s own hiss but amplifies the pulses along with the signal.

Thunderstorms and badly suppressed car ignition

Natural sources such as thunderstorms and man-made ones such as poorly suppressed ignition systems produce brief, strong spikes. Thermal noise is a steady hiss, which a limiter or blanker cannot remove.

Its brief spikes spread energy over a wide range of frequencies

A very short spike contains energy over a wide spread of frequencies, so it is heard across a band at once, with no mixing needed. Its large amplitude gives it a lot of energy despite its short duration.

0,6 V

The 0,4 V signal is too small to make the diodes conduct, so it passes unchanged, but the pulse is clipped where they conduct, at 0,6 V. The clipper cannot pull the pulse down to the signal’s 0,4 V.

24 dB

The voltage falls by 9 ÷ 0,6 = 15 times, and 20 log 15 = 23,5 dB, nearest 24 dB. Using 10 log, which is for power ratios, gives the wrong 12 dB.

One clips positive peaks, the other negative ones

Each diode conducts on one polarity, so together they clip both halves of a pulse at about 0,6 V. Neither can tell noise from signal: anything smaller passes, anything larger is clipped.

It never reaches the 0,6 V the diodes need to conduct

Below about 0,6 V the silicon diodes stay off, so the wanted signal passes untouched; only peaks that try to go higher make them conduct and are clipped. Conducting diodes would shunt the signal, not pass it.

It silences the receiver for a very short, set time

The blanker senses each pulse and blanks the signal for just a moment, whereas a noise limiter clips it; the manual says the audio is muted for a few milliseconds, though real blankers gate the IF for microseconds.

The ear is insensitive to very short gaps in sound

Human hearing barely notices very short interruptions, so losing the wanted signal for that moment costs far less than hearing the spike. The AGC only sets the gain; it cannot fill in missing audio.

A ratio detector

The manual names the Foster-Seeley discriminator and the ratio detector. An envelope detector responds only to amplitude, which FM does not vary, so on its own it cannot recover FM.

Frequency changes become amplitude changes on a filter’s slope

On the sloping side of a filter’s response the FM signal comes out stronger as its frequency rises and weaker as it falls, frequency becoming amplitude; a diode detector recovers this AM. Real Foster-Seeley circuits use phase shift.

Its amplitude decreases

On a rising slope the response grows with frequency, so as the signal moves down from FCentre towards FLow its output falls from ACentre to ALow. Swinging upwards gives a larger output.

The discriminator also responds to amplitude changes

A discriminator responds to amplitude as well as frequency, so noise and fading would reach the audio; the limiter clamps the IF to a constant level first. It limits amplitude, not deviation.

The limiter cannot yet hold the level steady

A weak signal is too small for the limiter to clip, so its amplitude noise reaches the discriminator and is heard as hiss. As the signal strengthens the limiter takes over and the hiss fades.

The limiter

The order is IF filter, IF amplifier, limiter, discriminator, audio output: the limiter must level the signal before the discriminator sees it. The product detector belongs to SSB and CW reception.

Quieting

As the limiter takes hold, amplitude noise no longer reaches the discriminator and the hiss dies away: quieting. A squelch is different: it mutes the audio completely when there is no usable signal.

A squelch circuit

Below the squelch threshold the squelch mutes the audio, sparing the listener the hiss of an empty channel. The AGC does almost the opposite: with no signal it turns the gain right up.

It may be fixed, or adjustable by the user

The squelch threshold is the signal level below which the audio is muted; the manual says it may be fixed or set by the user with a squelch control. The AGC controls gain, not muting.

Phase noise on the LO mixing with strong unwanted signals

The synthesised LO carries phase-noise crud around its frequency, which mixes strong nearby signals into a spread of noise, some of it landing in the passband. The BFO works only on the already-filtered IF signal.

Synthesised rigs from the early 1980s

Chapter 20 says synthesisers in early-1980s amateur rigs had significant phase noise; modern designs are much better. A free-running LC oscillator is not a synthesiser, and good ones have far less phase noise.

Its dynamic range is too small

A clean, strong signal nearby filling the passband with noise is reciprocal mixing, which limits dynamic range: hearing weak signals beside strong ones. Too wide a filter would let the station itself through.

Raise an off-tune signal until it matches its on-tune passband power

An off-tune signal is raised until it gives the passband as much power as it did on tune; the rise in decibels is the figure. A signal on the image frequency measures image rejection instead.

Around 100 dB

The manual treats this as a working measure of dynamic range, at least 100 dB in most good receivers. A figure of 20 dB would let quite an ordinary nearby signal bury a weak one.

−115 dBm

The spread-out noise lies 95 dB below the strong signal: −20 − 95 = −115 dBm, enough to bury a very weak station. Dropping the minus sign of −20 dBm gives the wrong −75 dBm.