How your HF receiver gets its selectivity

In my previous article I discussed sensitivity requirements for HF receivers and this month I will be discussing selectivity. Selectivity is the ability of a receiver to discriminate between signals having different frequencies. In a receiver specification it is defined as the bandwidth, and is usually measured in kHz. If several signals are on the same, or overlapping frequencies, no amount of selectivity adjustment will enable the different signals to be resolved properly. If your receiver had no selectivity at all, it would be in effect a broad band amplifier, and all signals would be received at once. Many different methods are used to determine the bandwidth, the most commonly encountered are tuned circuits, ceramic elements, crystal elements, mechanical resonance and resistor capacitor networks.

Nowadays, selectivity is almost universally done at IF frequencies. This is because the IF frequency is the same, no matter what RF frequency the receiver is tuned to, enabling the selectivity to be effected at one common frequency. In very early receivers, no IF frequency was used, the selectivity being done at RF frequencies, hence the name TRF (tuned radio frequency). For best receiver performance, the main selectivity should be determined as close to the antenna input as possible.

The further back in the receiver circuits it is placed, the greater the signal handling requirements of the pre-selectivity amplification become, because it will have to handle signals at all frequencies, some of which might be of very high level. In some ways the TRF met this requirement, but it has too many other disadvantages as compared to a superhet, with it’s fixed IF and wide frequency coverage, but TRF receivers are still used for specialised fixed frequency applications. In a high quality receiver, the main selectivity is done at the first IF conversion, this commonly being 45MHz or 70MHz in modern receivers. (Older designs converted directly down to a low IF frequency, which makes the requirements of the RF tuned circuits very stringent to get good image rejection, and nowadays virtually all general coverage HF receivers convert up, and then back down again). Good filters at high frequencies are very expensive, and most amateur receivers use a cheaper filter at these frequencies to provide a reasonable amount of selectivity (roofing filter), the main selectivity being done later on in perhaps a second or third IF. Frequencies commonly used are 10.7MHz and 455kHz, where good filters can be more cheaply made.

All early receivers used tuned circuits to determine the bandwidth. The disadvantage of this is that a high number of these had to be cascaded to make steep sided filters, and to make a filter with similar characteristics to even a modestly specified crystal or ceramic filter designed for SSB use, would be very complex and impractical. At low frequencies it is easy to make narrow LC filters, but at higher frequencies, obtaining a narrow bandwidth becomes increasingly difficult, hence the conversion down to a low frequency IF in early receivers. To change bandwidths became complex, and clever switching arrangements had to be devised. One of the most novel is the continuously variable selectivity used on the early Eddystone receivers. A mechanical arrangement was devised which physically moved inductive components in the IF transformers to vary the coupling, using a mechanical linkage to a small lever control on the front panel.

From the 1930’s, practical crystal filters were developed, notably in the USA. Early filtering arrangements were made up of several discrete crystals often referred to as a lattice. Because of the difficulty in manufacturing high frequency crystals, early crystal filters were made at low frequencies, typically around 460kHz. Over the years, crystal technology has improved considerably, and modern units comprise multiple crystal elements in a single metal can, with frequencies at least up to 70MHz.

Ceramic filters are a more recent innovation, and uses a ceramic material much the same way as a normal quartz crystal. These devices are less well specified than their crystal counterparts, but they are a lot cheaper and are used widely in communications and consumer equipment. One of the disadvantages of a crystal filter is it’s power handling, typically in the region of a milliwatt for in band signals. In high performance receivers with a crystal filter at the first mixer, the signal levels presented to the filter may easily exceed this level, and such filters have to be specially designed and manufactured so as not to be damaged or suffer from non linearity effects.

Mechanical filters have been developed (notably by Collins in the USA), as an alternative to crystals, and can give very high performance. Essentially it is a line of mechanical resonators made of a specially selected material and shaped to resonate at the IF frequency. The input of the filter is to a transducer (not unlike a loudspeaker), which excites the resonators and the output of the filter is from another transducer (not unlike a microphone). Frequencies will only pass between transducer to transducer, if it is equal to that of the resonators. RC filters are used in direct conversion receivers. In this type of receiver, the IF is at zero frequency. Direct conversion as regards HF receivers still has somewhat of a novelty status, and although not in common use, it does have the great advantage that effective filters (because they are at audio frequencies) can be easily and cheaply made, and unlike the normal superhet, it does not suffer from so many spurious responses.

You might be thinking why not just make a good audio filter, and put it in the receiver audio stages, such as a narrow CW filter. This breaks the golden rule of having the main selectivity as far forward in the receiver as possible. Yes it will reduce broad band  noise and increase the effective sensitivity due the narrower bandwidth, but it will not be effective in rejecting strong adjacent frequency signals.

This is because AGC action is determined by signals in the bandwidth before the add on AF filter, so whenever there is strong signal within this bandwidth, the gain of the receiver will be reduced and your weak wanted signal may be lost. Don’t try and be smart and switch the AGC off. If you do this, all the receiver stages before your AF filter will have to instantaneously handle (instantaneous dynamic range) the strong and the weak wanted signal. Most receiver IF stages are not designed to do this, because they rely on the AGC to turn the gain back to prevent overload. Even if you turn your manual gain control back, it will not increase the receiver’s dynamic range, it will just reduce the strength of all signals, including any wanted ones.

The bandwidth of a typical HF receiver will be 1.8kHz or 2.4kHz for SSB, 9kHz for AM, several hundred hertz for CW and about 15kHz for FM. These bandwidths are usually specified at the -6dB points, with the -60dB points also commonly given. The ratio of the bandwidths at these points is known as the shape factor. For example, if the -6dB bandwidth is 2.4kHz and the -60dB bandwidth is 4.8kHz, the shape factor is 2:1. A good filter will have a shape factor of 2:1 or less. Some receivers are fitted with a system that can continuously vary the bandwidth, which is commonly known as passband tuning (PBT). If you can imagine two identical bandwidth filters in series, one with a fixed centre frequency, and the other with a variable centre frequency, when the two centre frequencies are the same, the overall response will that of an individual filter.

If the centre frequency of one of the filters is moved, the overall response will be narrowed, because signals will be passed only where the filter responses overlap. In practice, two filters with a fixed centre frequency are used, and a frequency mixing arrangement is used to effect the amount of overlap. The centre frequencies do not have to be the same, and typically 10.7MHz and 455kHz frequencies are used, the PBT control varying the frequency of the local oscillator used in the frequency mixing arrangement.

It is often assumed that it is the IF filter which ultimately decides how good signal rejection is on adjacent frequencies, but this is not the case. The characteristics of the local oscillator are critical as regards rejection. Although we think of the local oscillator signal as being comprised of one frequency with perhaps some harmonics, in practice this is not strictly the case. The  harmonics are unimportant, because any responses produced by these can be effectively filtered out, but what is important is what is often referred to as the purity of the signal.

All signals contain a certain amount of residual FM, this being generated by noise inherent in the oscillator components, notably by transistors or varicap diodes. The deviation of the FM is very small, but enough to produce sidebands, which being caused by noise, will appear as a broad band of low level signals extending from a few hertz to many tens of kilohertz either side of the centre frequency. This is known as local oscillator sideband noise. Because the LO signal consists of a broad range of signals, it will mix adjacent unwanted signals into the receiver passband. This effect is known as reciprocal mixing, and it doesn’t matter how good your IF filtering is, if your oscillator has noise sidebands, it is the level of these that will determine the adjacent frequency rejection and not your filters! The old valve receivers had very pure local oscillators, and this effect was negligible. Since the advent of synthesised receivers this problem has become evident, because to design a really pure synthesiser is very difficult. In a good modern HF receiver, you may find that half the circuitry is taken up with the synthesised local oscillator in order obtain the required purity.

When you look at those really sharp filter responses in your receiver specification, it doesn’t mean you are going to obtain this in practice. In a well specified receiver, a figure will be given for adjacent frequency rejection which will take into account the local oscillator sideband noise. For you serious CW operators there is no substitute for a good crystal filter at the beginning of the IF, and even if you have a fancy DSP AF filter, although often useful, it will not solve strong adjacent frequency problems. I hope my explanations of what is quite an involved subject, have given you some insight into HF receiver selectivity considerations.

Tony Richardson.

Return to the index of Vital Spark articles.

G3MGQ’s Month on the Air

Prepared by the clubs RSGB trainer, G3MGQ, you will find the latest DX contests including the ones to shoot for as well as ones to give a wide berth. Why not download the latest edition of Month on the Air and enjoy your DX just that little bit more.

Become a member of HERC

Join the Hastings Electronics and Radio Club.

Why not join one of the largest and most established Radio clubs in the South East of England? Very low joining cost, and free for a year to new licencees.

Vital Spark Archive

Vital Spark newsletter articles

Take a look through a large selection of articles written by club members over the years which have been published in the monthly Vital Spark newsletter .

Used Ham Radio Equipment

View HERC's Used Ham Radio Equipment for sale list..

Every four weeks, HERC's Used Ham Radio Equipment for sale list is updated on the site. Bookmark the gear for sale page to re-visit easily and take advantage of the used equipment on sale through the club.

Club Photographs

HERC Image Galleries.

Here is the official HERC photograph archive which contains multiple image galleries spanning several decades since the club was formed many years ago. Enjoy the images!

UK Amateur Radio Repeaters

UK repeaters

Click button above for full list, or a local repeater callsign below for info.

GB3EB 2m in Uckfield- Active
MB6EB 2m DStar Node in Eastbourne - Active
MB6RY Wires-X DigiGate in Broad Oak - Active
GB3HE 70cm in Hastings - Active
GB7HE 70cm DStar in Hastings - Coming soon
GB3ZX 70cm in Eastbourne - Low Power
GB3JT 23cm ATV in Hastings - NoV cleared
GB7RY 70cm X-Wires Repeater Rye - Active
GB7ES Eastbourne - DSTAR Rpt. - Active
GB3ES 2m in Hastings - Active

For a complete list of repeaters, head over to
the UK Amateur Radio repeaters list.

Popular pages

Get your amateur radio licence - Find out more about amateur radio licence training.
Month on the Air - G3MGQ's popular monthly DX contest/expedition list.
Wilf Gaye Memorial Cup - The clubs annual operating event in the memory of Wilf Gaye M0GYE.
St. Richard's College Buildathon/STEM/ARISS - HERC attends St. Richard's Catholic College for their various events surrounding the Tim Peake ARISS contact.
G3BDQ - John Hey's Rare QSL Cards.
Sussex Electronics Radio Fair - SERF Sussex Electronics Radio Fair 2016.
Vital Spark - A selection of articles re-published from the Vital Spark.
RSGB News - Find out how to get RSGB news on your mobile or PC.
Experimenters Corner - A selection of Proteus projects by Bob Gornal (G7DME)
BBADL - Bath Based Distance Learning Course.
Conquest Hospital Radio - Presented by HERC member Antony (G4CUS).
Radio Rallies 2016 - An up to date list of radio rallies scheduled for 2016.
Club QSL Cards - A selection of QSL cards the club has received over the years.
Other Newsletters - Excellent newsletters and magazines from other clubs.
TX Factor episodes - Take a look at the TX Factors YouTube videos.
John Taplin - A bio of the late John Taplin.

Amateur Radio Resources

Other Radio Clubs & RAYNET

BSARS - Brede Steam Amateur Radio Society

RAYNET - The Hastings and Rother RAYNET Group.

HERC members sites

Sigord - Gordon Sweet
Hastings Radio Comms - Andrew Haas-Campbell
Hoofbags - Liz Costa