In the history of shortwave receivers, the Wadley loop played an important role. It was invented by the South African engineer Trevor Wadley in 1946 and it was applied in professional shortwave receivers from Racal in the 1950s and 1960s. In the 1970s it was popular in amateur shortwave receivers until it gave way to PLL synthesizer circuits.
It is a very clever circuit that made it possible to construct shortwave receivers that covered the entire HF spectrum with good image rejection, good dial accuracy and good frequency stability (adequate for SSB). Circuit complexity and cost were modest compared to other receiver designs that met the same requirements.
The Wadley loop utilizes two tuning knobs that tune two different continuously variable frequency oscillators (VFOs). The first tuning knob tunes to the required megahertz band from 0 to 29. Though it controls a continuously variable oscillator, it appears to act as a band switch. Due to drift cancelation, any deviation from an integer multiple of 1MHz is canceled out. The second tuning knob tunes to the exact frequency within one megahertz band.
The very first shortwave receivers used regenerative circuits. They typically had one or more RF amplifiers (vacuum tubes) with tuned circuits between them. All of these tuned circuits were tuned to the received frequency. This is called a TRF (Tuned Radio Frequency) receiver. The final RF amplification tube had positive feedback, such that it nearly oscillated but not quite. This provided much gain and selectivity, but setting the feedback level just right was critical. Too much and the circuit would oscillate. Satisfactory results could already be had with just the regenerative tube without any other amplifiers before it. But if you made such receiver to oscillate, the whole neighborhood would be treated to a howling sound. This was a problem especially in the normal broadcast bands (mediumwave and in Europe also longwave).
After 1930, the superheterodyne receiver would be used nearly universally. It would provide uniform and good nearby selectivity across the tuning range, something a TRF receiver could not do. Plus that you did not treat the whole neighborhood to a howling sound if you did not know how to tune properly. In a superheterodyne receiver, a variable oscillator is used to mix the received signal to a fixed frequency, the Intermediate Frequency of IF. Further amplification and filtering is performed at that fixed frequency. By varying the oscillator frequency you tune to a different station. As the oscillator is at a frequency different from the received frequency and the signal is not very close to the antenna input. radiation from the local oscillator is typically not a problem for the neighbors.
Around World War 2, even professional communications receivers were still standard superheterodynes. They may have used a higher than standard IF and an extra tuned RF stage to improve image rejection, but the basic system was the same.
In the 1950s and 1960s many professional communications receivers would use double conversion (two different IF frequencies). Further they would use one carefully calibrated Variable Frequency Oscillator (VFO) and mix its output with numerous different crystal frequencies to obtain the local oscillator signal for all bands. This way the VFO could have only one tuning range at a comparatively low frequency, so its stability and dial accuracy could be good enough for SSB and RTTY. Double conversion for these receivers still meant IF frequencies of 1 to 3MHz, certainly not 40MHz or higher.
For example the German Rohde & Schwarz EK07 receiver of the 1960s has a first IF of 3.3MHz (for the higher bands) and it has a VFO at a 3.4-6.4MHz, a 3MHz wide range. To obtain the required stability and dial accuracy for such a comparatively wide range it relied heavily on mechanical precision engineering, rather than clever circuit design.
One brand of professional communications receivers however, used the Wadley loop. This brand was Racal, producer of the famous RA17 receivers. And yes, it does use a first IF of around 40MHz, hence it does not require special treatment of the lower bands (which would need single conversion if the first IF was lower). Now the RA17 did use mechanical precision engineering for the VFO, but the requirements were less strict than for the EK07 (having only a third of the tuning range). And the number of switches and band-dependent filters was drastically reduced compared to the EK07.
In 1973 the same Wadley loop was introduced in a consumer grade receiver, the Barlow Wadley XCR30. Although its performance was much less than that of the Racal, it cost only a small fraction (around 1/25th of price of the Racal). Its performance was much better than that of any general coverage receivers you could buy in this price range. The Yaesu FRG-7 was introduced in 1977 and it performed considerably better than the Barlow Wadley.
The Wadley loop had its limitations though and in the 1970s, professional communications receivers moved to PLL synthesizers to generate their tunable local oscillator signals to mix the received signal to a high first IF. Amateur receivers followed suit after 1980. In the meantime there were some very interesting designs that combined a VFO (tuning a 1MHz wide range) with a PLL synthesizer to generate 1MHz multiples. This system was used by the Kenwood R1000 and Yaesu FRG-7700.
Modern receivers use direct digital synthesis (DDS) instead of (or combined with) PLL synthesizers. In the 1990s some high end communications receivers digitized the final IF signal and did the IF filtering and demodulation digitally in a DSP (digital signal processor). This technique is now commonplace, even in cheap receivers. Instead of a dedicated DSP chip one can use the power of a modern personal computer to do the job. There is now a trend to use a low first (and only) intermediate frequency, typically less than 48kHz, in the range of normal sound cards. To suppress the image frequency, one can use two mixers and two AD converters, each fed by oscillator signals of the same frequency and 90 degrees apart. This is called quadrature mixing.
As of 2016 it is now even possible to directly digitize the entire HF spectrum from 0 to 30MHz and do all further processing in a computer. No more mixers, no more image frequencies (but you still need good anti-aliasing filters).
Let's start with an ordinary superheterodyne circuit for a standard LW/MW radio, whose diagram is shown below. Such radios were made from the 1930s to the 1990s. The received signal (either in the longwave range 150-350kHz or in the mediumwave range 530-1620kHz) is fed through a preselector, typically a parallel resonant LC circuit of a fixed coil and a variable capacitor. The coil could be wound on a ferrite bar, so the coil itself would act as an antenna. The received signal is mixed with a variable oscillator to obtain the desired 455kHz intermediate frequency. The oscillator also contains a resonant LC circuit containing a fixed coil and a variable capacitor. The variable capacitor is typically a dual variable capacitor where both capacitors are on a single shaft so they are varied both by rotating the shaft. Both the preselector and the oscillator have two tuning ranges. The coils for both ranges can be switched using a band switch.
The 455kHz IF signal is fed through a series of narrow band filters and amplifiers. The filters select the 455kHz signal with a well defined bandwidth (around 10kHz in an AM broadcast receiver). The amplified signal is then demodulated using a diode detector. The demodulated signal is further amplified and fed to a speaker.
In traditional radios these filters were constructed from transformers and capacitors. In more modern radios you would find ceramic resonators. Communications receivers would contain several ceramic filters with different bandwidths, that can be selected. Communications receivers would also contain a BFO, a Beat Frequency Oscillator that oscillates near the 455kHz IF. The diode detector could then be replaced by a product detector, which is basically another mixer.
Finally we would have an AGC (Automatic Gain Control) circuit. The amplified IF signal is detected by another diode and then smoothed using a much larger capacitor to eliminate all modulation and just retain the carrier level. This signal is used to control (reduce) the gain of the IF amplifiers. This reduces the effect of fading and it reduces the difference in audio level between strong and weak stations.
The standard superheterodyne works surprisingly well for mediumwave and longwave. But if we move on to shortwave, we will run into the problem of image frequencies. In a superheterodyne circuit, two frequencies will mix to the desired IF (f1) if you tune the oscillator to a given frequency f2. f2+f1 and f2-f1. So if you tune the oscillator to 1455kHz and the IF is 455kHz, both an input signal of 1000kHz (f2-f1) and an input signal of 1910kHz (f1+f2) will mix to the IF. We do have a tunable preselector and 1910kHz (unwanted) is far removed from 1000kHz (wanted), so there will be no problem.
But on shortwave the situation is different. If you tune to 15100kHz, the image frequency is at 16010kHz, well within the range of a single resonant LC circuit used a a preselector.
There are two straightforward ways to reduce the problem. Professional shortwave receivers of the 1940s would typically use both methods combined:
It could be desirable to increase the IF even further to improve image rejection, but then two other problems popped up:
But other receivers used double conversion, as shown in the diagram below. In the example we have a first IF of 2MHz (making the image frequency 4MHz removed from the received frequency). It has a first IF of 2MHz and it tunes three shortwave ranges: 5.8-10MHz, 11.5-15.8Mhz and 17.4-21.9MHz, covering all broadcast bands from 49m to 13m where each range is just over 4MHz wide. The 2MHz IF is mixed with a fixed frequency oscillator of 2455kHz to the second IF of 455kHz. At this frequency you can get the desired selectivity much more easily and cheaper than at 2MHz (let alone 9MHz).
With suitable band switches you can configure the same radio to operate in single conversion mode in lower bands (longwave, mediumwave and low shortwave, for example 1.6-5.2MHz). You would bypass the first IF section and the second mixer and go straight from the first mixer to the 455kHz second IF section. But what if we make double-conversion superheterodyne whose first IF is way above the highest shortwave frequency, for example at 40MHz? Image rejection could be achieved with just a low pass filter at 30MHz. The first IF filter would have to be narrow enough to reject the image for the second mixing (910kHz from 40MHz), but this is a fixed frequency filter, so this could be done. We do not need awkward configuration switches for different bands, so what keeps us from doing just that? Unfortunately it is impossible to achieve adequate frequency stability if the first local oscillator runs from 40 to 70MHz. At least not with traditional VFOs. It would be inadequate for AM and completely disastrous for SSB.
But the idea of a general coverage shortwave receiver with such a high first IF was too good to be false, so people kept trying to find a way out. In 1946 Trevor Wadley did find a solution to the stability problem and this was applied in the 1950s and 1960s by Racal in the RA17 receivers.
In the example the first IF is the band 45.5-44.5MHz (but practical examples could have the first IF band anywhere from 40 to 60MHz). The first VFO is tuned from 45.5 to 74.5MHz. When the VFO is tuned to 45.5MHz, the band 0-1MHz will be converted to the band 45.5-44.5MHz. The band is turned 'upside down' because the received frequency is subtracted from the oscillator frequency, so the low frequency of that band is converted to the high end of the IF band and the high frequency of the band is converted to the low end of the IF band. When the VFO is tuned to 46.5MHz, the band 1-2MHz will be converted to 45.5-44.5MHz, again 'upside down". Finally, when the VFO is tuned to 74.5MHz the band 29-30MHz is converted to 45.5-44.5MHz.
The second IF is the band from 3-2MHz. We mix the first IF with an oscillator signal of 42.5MHz. The oscillator frequency is subtracted from the received frequency, so the band will not be turned upside down again. As the band was already upside-down at 45.5-44.5MHz, it will still be upside-down at 3-2MHz.
The band from 3-2MHz can be tuned by an ordinary superheterodyne circuit with an IF of 455kHz in the example. The IF stages and later are not shown in the block diagram, but these are similar to other diagrams and they will include selectable filter bandwidths and a BFO. The kilohertz dial is marked 0 when the circuit is tuned to 3MHz and 1000 when the circuit is tuned to 2MHz.
So far I left out a very important detail of the circuit. All would be fine if the first VFO is tuned to an exact multiple of 1MHz plus 0.5, but if it isn't, wouldn't disaster strike? No it would not strike and it is caused by the way the 42.5MHz signal is generated. It is not just generated by a fixed 42.5MHz crystal. So how is it generated then?
First we have a 1MHz crystal oscillator. This is tuned to 1MHz and it is rock stable. The harmonics generator turns the sine wave from the 1MHz oscillator into narrow pulses. These pulses contain all the harmonics of the sine wave. We need all harmonics up to and including the 32th harmonic. These harmonics are mixed with the output of the first VFO (which tunes from 45.5 to 74.5MHz). When the first VFO is tuned to 45.5MHz, the third harmonic (3MHz) will mix with it to give 42.5MHz. When the first VFO is tuned to 46.5MHz, the fourth harmonic (4MHz) will mix to give 42.5MHz. When the first VFO is tuned to 74.5MHz, the 32th harmonic (32MHz) will produce 42.5MHz. All other harmonics are present simultaneously, so the mixer output will contain lots of other frequencies 1MHz apart, but a narrow filter selects just the desired 42.5MHz mixing product.
Now what if the first VFO is not exactly at a 1MHz multiple plus 0.5? Suppose we want to tune it to 46.5MHz but tune it to 46.6MHz instead. The intended frequency band 1-2MHz will now be at 45.6-44.6MHz (instead of 45.5-44.5MHz) and it could still be passed by the first IF stage as this filter is slightly wider than 1MHz. But the when 46.6MHz is mixed with the fourth harmonic of 1MHz, it produces 42.6MHz instead of 42.5MHz. Again this will still be passed by the 42.5MHz filter. Now the 1-2MHz band (which ended up 100kHz too high in the first IF passband) will be mixed with 42.6MHz instead of 42.5MHz and it will end up at exactly 3-2MHz. So at the second IF it is exactly where we want it.
So any small deviation of D kilohertz (up or down) in the first VFO will cause the received megahertz band to have a deviation of D in the passband of the first IF filter. The signal with which we mix down to the second IF will have a deviation of D too. As this frequency is subtracted from the first IF, the deviations will cancel each other out and the received signal is at the second IF in exactly the band we want to have it. This technique is called drift cancelation and this is the essence of the clever idea of the Wadley loop.
All Wadley loop receivers that I know of use a second IF band of 3-2MHz. Also they always have a first IF centered around a whole megahertz, for example 44.5-45.5MHz. The filter after mixer 4 (which selects the desired mixing product as input signal for mixer 2) is then always 2.5MHz below the center of the first IF band. The VFO is always exactly between two multiples of a megahertz when it is properly tuned. It would be possible to design the receiver for instance with a first IF band of 44-45MHz and the filter after mixer 2 at 42MHz and the VFO tuned to a whole MHz, but this would not work so well. For one thing, the 42nd 1MHz harmonic could get through the 42MHz filter directly. The mixing products between the 1MHz harmonics and the VFO would all be close to a multiple of 1MHz and they could intermodulate with each other to give mixing products that are all very close to a multiple of 1MHz, possibly some very close to 42MHz. The problem is mitigated if the VFO and the filter after mixer 4 are both offset by 0.5MHz There could still be intermodulation products at multiples of 1MHz, but they would end up outside the filter passband.
It all started with the Racal RA17, produced in England between 1955 and 1970. The first IF is from 39.5 to 40.5MHz and the second IF is from 2 to 3MHz and the third IF is 100kHz. This requires the tunable filter in the 2-3MHz band to have two tuned circuits instead of just one. The kilohertz scale is a stretch of 1.5m of photographic film, which provides excellent dial accuracy of less than 1kHz. The radio is built like a tank and all sections are carefully shielded. It contains a tunable preselector and an attenuator, so you can prevent overloading under all circumstances. It contains more than 20 vacuum tubes and no semiconductors. Drawbacks: the wider IF filters (3kHz and up) are LC filters and are not as good as modern ceramic filters. It is not optimized for SSB reception (no real product detector). You could purchase an external SSB unit, but these are rare nowadays.
Between 1973 and 1980 the Barlow Wadley XCR-30 was produced in South Africa. I know of no other electronics products that were widely exported from South Africa. The first IF runs from 44.5 to 45.5MHz, the second IF runs from 2 to 3MHz and the third IF is 455kHz. Dial markings are at 10kHz and you could easily determine the frequency to 5kHz. It does have a product detector and selectable LSB and USB settings and it works well with SSB. It has very low battery consumption. There is a preselector of some sort, whose band selection is controlled by the MHz tuning knob (as well as the VFO). It can be further tuned with a knob marked Antenna Trim. It is also very sensitive, but combined with a poor dynamic range this is mostly a drawback. As it has only a rudimentary preselector and no attenuator, this radio could easily overload. Further the 1MHz harmonics are not shielded well enough, so they could get into the signal path, making tuning of stations at exact megahertz multiple problematic.
In 1977 Yaesu (Japan) produced the FRG-7 receiver. Its first IF is 54.5-55.5MHz, the second and third IF are 2-3MHz and 455kHz respectively. It comes with a tunable preselector with its own dial scale marked in MHz. It was succeeded by the FPG-7000, which added a digital frequency counter. The Yaesu receivers perform better than the Barlow Wadley.
Around 1980 Radio Shack had the Realistic DX300 and DX302 receivers. They also have digital readout and a manually tunable preselector. Apparently they are not as good as the Yaesu products. The DX-302 has several IF filter bandwidths, while the DX-300 does not. Otherwise the radios are very similar.
Is there another way of building a general-coverage receiver with a high first IF that is nevertheless stable across the entire range? Yes there is. First use a VFO at a comparatively low frequency and with a fixed 1MHz range. Next mix it with a crystal oscillator with 30 different selectable crystals. So imagine we have the VFO running from 4.455 to 5.455Mhz. Now mix this with with the output of a crystal oscillator, with selectable crystals with frequencies from 36 to 65MHz, each an exact multiple of 1MHz. We will use only the sum frequencies, so we have to filter out the lower difference frequencies. As the sum frequencies of the lower ranges overlap with the difference frequencies of the higher ranges, we need to switch several filters across the range. The result will be that we can cover the range 40.455-70.455MHz in 30 selectable 1MHz ranges. This can be used in a general coverage receiver with a first IF of 40.455MHz. A 40MHz oscillator mixes it down to the second IF of 455kHz.
To avoid the need for switchable filters, we could mix the VFO in two stages: first mix it with a crystal oscillator to around 20MHz, then mix it with another crystal oscillator with selectable frequencies to arrive at the desired range. But if we use two mixers and two crystal oscillators, why not have selectable crystals in both crystal oscillators, thereby reducing the total number of crystals from 30 to 11?
Example: Suppose the VFO runs from 4.455 to 5.455MHz. First mix the VFO with a first crystal oscillator, which has six selectable frequencies: 16, 17, 18, 19, 20 and 21MHz. A high-pass filter selects only the sum frequency. Then the resulting signal covers 20.455 to 26.455MHz in six ranges. Next mix this signal with a second crystal oscillator, which has five selectable frequencies: 20, 26, 32, 38 and 44MHz. Again only the sum frequencies will be used by passing the mixer output through a high-pass filter. The resulting signal now covers the range from 40.455 to 70.455MHz in 30 1MHz wide ranges, one range for each combination of crystals. This arrangement can already be called a frequency synthesizer. Did I say synthesizer? I will come back to it in the next section.
Some general coverage professional communications receivers do in fact use the mix VFO approach to cover all the bands with a single VFO range. But it was also especially popular in non general coverage receivers, for instance for the amateur bands only. These often have a single VFO range of 500kHz wide, for instance from 5 to 5.5MHz. For each 500kHz band section the VFO is mixed with a dedicated crystal. Most amateur bands fit nicely into a single 500kHz segment, except for 10m, which would need four crystals.
Advances in IC technology made the PLL synthesizer practical for amateur receivers. The first radios with a PLL synthesizer use it only to generate 1MHz multiples and still use a VFO to tune around an any 1MHz range. We could use a PLL synthesizer in a mix VFO arrangement to replace the battery of 30 crystals required for each band. But a more clever approach is to mix the VFO signal into the feedback loop of the PLL to translate the VCO signal to an exact multiple of 1MHz, which is then divided by the programmable divider of the synthesizer to arrive at 1MHz. The 1MHz output of the divider is compared with a 1MHz reference signal and the phase comparator result is filtered and it controls a VCO, just as in any other PLL synthesizer. The Kenwood R1000 and the Yaesu FRG770 use arrangements like this.
But this arrangement was short-lived and the local oscillator signal would soon be fully generated by a programmable synthesizer without a VFO. The tuning knob was replaced by an encoder that sends pulses to the microcontroller to instruct it to tune the frequency up or down.
Both the mix VFO approach and the PLL synthesizer approach have the advantage that the first IF filter can be much narrower than 1MHz (it can be made as narrow as you can practically get at this frequency). This is a big win compared to the Wadley loop, which needs to amplify a full 1MHz band in the first IF amplifier and run it through the second mixer as well. This and advances in mixer technology largely eliminated the need for a manually tuned preselector. A few fixed band filters would suffice.
So far we have only seen the Wadley loop in general coverage HF receivers with a range of 0-30MHz. But could we get even more use out of this circuit? I think that today any of the proposed extensions would be better achieved with PLL synthesizers, but imagine what could be done with the technology of the late 1970s.
These are a few loose ideas:
The Wadley loop is a very interesting circuit that played an important role in the history of shortwave receivers. In particular in the 1970s it allowed the construction of affordable general coverage HF receivers for amateurs that offered good dial readout (before there were affordable frequency counters) and that offered sufficient frequency stability before PLL synthesizers were affordable.
The circuit could potentially be used for other receivers than just HF receivers and also for signal generators or even transceivers. The invention of affordable PLL synthesizers made the system obsolete in these applications though.