Rob's web

Home - Techniek - Electronica - Radiotechniek - Radio amateur bladen - Radio Communication - Improving direct conversion receiver design


Several years ago, the Shefford and District Amateur Radio Society started a club project to build a 2m direct conversion (DC) transceiver. This article is the one I would have liked to have had when we started work. It explains the various ailments that can afflict the DC Rx, and suggests some cures. DC receivers are not toys: G3WRJ managed to work Finland, 2000km away, using the club project rig on 144MHz. The KISS principle(1) has much to recommend it, and many designs for DC receivers work well. However, given some attention to design, the DC receiver can rival the performance of the best communications receivers.

Please do not look at this article's length and think "If there's that much in the DC receiver that can go wrong, I'm not building one." DC receivers are fun and simple to build. However "you don't get owt for nowt", and, for a given level of performance, the fewer components you use, the more critical to the design each component becomes. Quite apart from this, the superheterodyne receiver is not without its own selection of complex ailments. A quick look at a dual conversion superhet's spurious signal mixing chart will be quite enough to convince you of that. The problems of DC receivers can be divided into the following categories: bandwidth, local oscillator radiation, hum and microphony. Each of these will be treated in turn, and finally some suggestions for mixer designs and their attendant diplexers will be made.

Bandwidth

For simplicity's sake, the single mixer detector is the best choice for the radio amateur DC receiver. This gives an 'IF' bandwidth twice that of the equivalent superhet; RF input signals both higher and lower than the local oscillator frequency cause an audio output. This degrades the noise figure of the Rx by up to 3dB when receiving SSB or CW signals, and gives a rather peculiar effect when tuning through a crowded HF band. However, this is not a problem on VHF where the bands are usually sufficiently clear.

The double 'IF' bandwidth also means that Double Sideband (DSB) signals such as AM are difficult to receive, though it is possible to phase lock the local oscillator (LO) to the incoming carrier if the receiver is DC-coupled. A quadrature phase-locked DC Rx would work well for AM broadcast reception on HF.

By using complex phasing techniques, 'genuine' SSB reception is possible, and the best demodulator choice for this seems to be the Weaver method(2). Unfortunately, this demodulator puts a narrow notch in the middle of the audio passband, and this, with its attendant phase distortion, means that the best demodulator for CW reception probably uses the phasing method.

Making simple DC receivers for use on FM is not easy. I once built one which was quite successful at demodulating local 2m repeaters, but it definitely did not like fluttery signals from mobile stations; the PLL would drift off and sulk for long periods during the contact. My advice is to stick to SSB. And that's my personal bias out in the open.

Hum due to the af amplifier

In the superhet, most of the gain is in the IF stages which pick up no hum. By contrast, the AF amplifier of a DC Rx is very sensitive to mains hum (Fig 1) because it has a voltage amplification of 100,000 (100dB). It should be designed with its own 300Hz - 3kHz filtering which will prevent the amplifier from picking up much of the 50Hz fundamental, but will not filter out harmonics above 250Hz. Power supply rectifiers generate harmonics of 50Hz, so good supply decoupling is essential around the first few stages of the audio amplifier. The mixer and the first audio amplifier should be mounted on the same board, and the connections between them should be as short as possible.

This is how to find the source of hum in receivers with diode mixers. First stop the local oscillator; either pull the crystal out and check that the oscillator stops, or disconnect the LO supply. With no LO drive, the mixer will go high impedance which will leave the audio amplifier without an input load. Simulate the mixer in operation by connecting a 50Ω resistor across the mixer AF output. If the hum stops, the fault is RF hum and does not lie in the AF amplifier. If the receiver carries on humming, add more supply decoupling components and try to eliminate any earth loops.

Suspect any inductors in the low pass filter between the mixer and the first AF amp, especially the ones with inductances in excess of 0.5mH. Try rotating inductors for minimum hum. As a last resort, try Mu-metal screening.

The easiest way out of hum problems is to run the receiver from an external power supply or, preferably, a battery. Toroidal mains transformers tend to have less flux leakage than the conventional types, so they are preferable for internal mains power supplies. If the AF output from the receiver is to be connected to any other equipment, for example a modem, use an isolating audio output transformer. This will help avoid earthing troubles.

fig 1
Fig 1: Gain distribution In VHF receivers: (a) Superhen (b) DC

Microphony due to the af amplifier

Microphony is the tendency of component parts of the Rx to act like a microphone. A typical DC Rx, when tapped lightly on the chassis with a screw-driver, will give a loud crackle or a booming noise in the headphones. This is why most DC Rxs use headphones rather than a loudspeaker: the vibrations in the receiver cabinet caused by a loudspeaker are too great for feedback to be prevented. This would manifest itself as a loud whistle from the speaker if the volume control were turned up to anything louder than a whisper. Using headphones eliminates the mechanical feedback path and has the secondary advantage of requiring less AF gain, making it easier to keep the audio stages stable. The other problem is that loudspeaker coils produce quite a high AF magnetic field which can couple directly back to the first AF stage and so cause feedback by a magnetic path. Microphony on DC Rx audio boards is easy to track down. Stop the LO and put a 50Ω resistor across the mixer as above, and tap each component in turn while listening to the Rx output. If none of the components is microphonic, the cause is RF microphony, and the fault does not lie with the audio board. Otherwise, the components usually at fault are the largest value capacitors or inductors between the mixer and the input of the first AF amplifier transistor or IC.

Ceramic capacitors are completely unsuitable for this job; the dielectric is slightly piezoelectric. This makes the capacitor work as a surprisingly good microphone; in fact this is exactly how crystal microphones work. Moulded polycarbonate capacitors (eg the PMC2R series available from STC) are more suitable, being only very slightly microphonic. Tantalum bead capacitors perform quite well, but have wide tolerances on the nominal capacitance. Ferrites are slightly magneto-strictive; this means that their magnetic field varies when they are vibrated. So, to avoid microphony, do not use cores with high inductance factors.

For inductors of more than 0.1 mH, use screened air-cored coils if the magnetic hum field is low enough. RM series inductor cores can be used with caution, and are less likely to pick up hum (but are relatively expensive). Do not attempt to use toroids to make inductors of 0.5mH or greater; they are either too microphonic or else require an impossibly large number of turns.

fig 2
Fig 2: (a) LO leakage, the superhet receiver; (b) LO leakage, the ring mixer DC receiver

Local oscillator radiation

Local oscillator radiation is the big problem of DC receivers. In a superhet, the LO is prevented from reaching the antenna by the selectivity of the RF input stage (Fig 2a). In a DC receiver, the LO is running at the same frequency as the incoming RF from the antenna, so it is impossible to filter it out (Fig 2b).

Here's an example of what can happen. I knew one of two amateurs who were friends, they both built 80m DC Rxs using kits of a popular brand. The LO radiation from these was so strong and distinctive that each could tell when the other was listening to the band and, of course, they could tell what the other chap was listening to. You may look on this with an indulgent smile, until I tell you that they lived over a mile apart.

On VHF, the problem is just as bad; a powerful and continuous carrier radiated within 2kHz of a calling frequency will not be popular with local radio amateurs. Apart from a requirement to be kind to your neighbours, a radiating LO will degrade your own receiver's performance; more about that later.

There are two ways the LO can escape to the antenna: by conduction or by radiation.

Conduction

No mixer is perfect, and manufacturers usually specify 30 to 40dB isolation between the LO and RF connections. If there is no preamplifier, this signal is conducted back through the filter to the antenna, as in figure 2b. If there is a preamp, the leakage is attenuated. Unless special designs are used, the attenuation of the leakage depends on the feedback circuit determining the amplifier's gain.

This can be either an actual circuit, or just the internal feedback capacitance of the active device (transistor or IC). In either case, the reverse attenuation is usually only a few dB greater than the amplifier's forward gain. This relationship is important, and will be mentioned later. Reference 3 discusses the reverse signal leakage of various types of amplifier. One way to reduce conducted leakage would be to insert an isolator between the preamplifier and the mixer. As ferrite isolators are expensive at VHF and unobtainable for the HF bands, the best solution is to use a grounded gate high-current FET stage (Fig 3).

fig 3
Fig 3: A broad-band 'isolator' to reduce LO radiation from the antenna.

With this circuit, instead of the forward gain being determined by feedback, it is determined by the severe mismatch between the transistor's high drain impedance and the drain load of about 100Ω. The result is that the gain of this circuit is zero forwards and -40dB backwards. This allows the RF from the antenna to reach the mixer, but will reduce the level of conducted mixer leakage reaching the antenna. Another advantage of the circuit is that it offers a broadband 50Ω termination to the mixer's RF input connections up to 300MHz.

The circuit's third order intercept point of +20dBm/tone is high enough for most applications. Note that the specification for the FET allows thedrain current to be between 24 and 60 mA. At 60mA, the power dissipated by Ti would be 0.72W, which exceeds the 0.5W maximum specified for the device. I have not blown a FET this way yet, but the squeamish can, if they like, put a resistor in series with Li to reduce the drain current to 42mA if necessary.

Radiation

Some local oscillator power will be radiated from the LO compartment, the connections to that compartment, and the connections to the mixer. Some of this power will escape directly from the receiver, and some will be received by the high Q tuned circuit(s) at the Rx input. From there, the signal can be conducted to the antenna. The only cure for this is sensible receiver layout and attention to screening.

The LO must be mounted in a separate box. Use feed-through filters or capacitors for all non-RF connections to the LO board. The LO connection to the mixer must be kept short. If RF connectors are not used on the LO box, the outer of the LO output coax should be connected to the LO compartment wall at the point where the cable passes through. It may be necessary to use double-screened or solid-outer coaxial cable for the RF connections. To make extra screening for coax cable: remove the outer braid from a thicker coax cable and thread the thinner cable inside, stretch the extra screen tight and solder it to the connector body at both ends. This works for RG174 (use UR43 outer) and UR43 (use UR67 outer).

RF microphony

Most of the microphony problems of DC receivers stem not from the audio board, but from the RF components, and specially any high Q tuned circuits in the path between the antenna and the mixer. Some of the LO power arriving here by either conduction or radiation will pass back to the mixer. This LO leakage signal is mixed with the original LO signal, and this gives a standing DC (direct current) voltage on the output of the mixer. Typical values for the leakage power and resultant DC voltage at the mixer output are shown in Fig 4a. Note that the values of DC voltage are for the reflected leakage component only; the actual voltage measured at the mixer output will generally be larger than the values shown.

fig 4
Fig. 4: (a)High LO leakage gives RF microphony; (b)Poor RF microphony supression by a preamp; (c)RF microphony suppression with an isolator.

Suppose that the front-end tuned circuit is now knocked with a screw-driver. The wobble of the coil in the tuned circuit will cause its resonance frequency to wobble, and so the amplitude of the leakage signal at the preamplifier input will change in sympathy. This causes a wobble in the standing voltage at the mixer output, which is passed through the audio amplifier and appears as a loud clang in the headphones. Because mixers also work as phase sensitive detectors, microphony is caused by both phase and amplitude variations of the LO leakage.

An alternative way of looking at the problem is that the LO leakage appears to the Rx as a carrier at exactly the LO frequency, and with a signal strength 60dB or so above the noise floor. Knocking the input tuned circuit will phase and amplitude modulate this carrier slightly, and, though the resulting side-bands are many decibels below the carrier level, the carrier is so strong that the side-bands have enough energy to rise well above the receiver noise floor, and thus appear as a clang in the headphones. The effect is the same as tuning an AM receiver to a strong signal from a crystal calibrator, switching the BFO off, and turning the RF/IF and AF gain controls fully up. Those that have used AM on HF will remember that this generally results in a deafeningly loud shriek from the receiver due to mechanical feedback from the loudspeaker. The crystal calibrator output of the AM receiver was giving the same effect as the LO leakage of the DC Rx.

The only way to reduce this effect is toreduce the LO leakage somehow. For micro-phony due to LO radiation, screening is the only answer. For microphony due to conducted LO leakage, use the FET 'isolator' shown in figure 4, from which it can be seen that, although a preamplifier reduces LO radiation from the antenna, it is of little use in reducing RF microphony.

If there is a significant radiated component to the LO leakage, any modulation of the RF field inside the DC Rx will affect the LO leakage at the mixer, so any of the wiring inside or near the Rx can seem to be micro-phonic. Intermittent connections of any sort will cause a loud crackling noise in the headphones. Examples of intermittent connections are loose circuit-board fastening screws, or the intermittent connection of the screw to the chassis due to the paint-work not being scraped away underneath the screw-head. It is better to earth circuit-boards with a proper earth wire and to use insulating circuit-board supports.

Tuned circuits are most prone to micro-phony when they resonate at the LO frequency. Other contributory factors are high local oscillator leakage, high circuit Q, and poor mechanical stability of the components. In a DC receiver it is very noticeable that the sensitivity to microphony reaches a peak at the resonance frequency of the tuned circuits. The phase variation due to a mechanical shock is at a maximum here, as is thecircuit's ability to receive or radiate energy from the LO. (Tune for maximum microphony, and you won't be too far off!).

The electro-mechanical performance of a tuned circuit can be improved by providing the coils with a tight-fitting former; for VHF inductors, wind the coil on a mandrel with a slightly smaller radius than the final former. Use plenty of varnish to hold the turns in place and fix the coil former firmly to the board. Air-spaced trimmer capacitors are preferable to the foil or ceramic types. (Remember micro-phony can also be caused by an oscillating pre-amplifier!).

If all that I have said about RF microphony sounds a bit far-fetched, do the experiment I did to convince myself of its reality. Dig out your old DC receiver. Make a series resonant tuned circuit for the band that the receiver works on. Connect the tuned circuit by a length of coax to the receiver's antenna socket. Switch the receiver on and tap the tuned circuit. The higher the receiver frequency band, the more probable it is that the receiver will make the tuned circuit microphonic. I expect that over 50% of all DC receivers will have this fault.

RF hum

The origins of rf hum are explained in Reference(4), which lists the characteristic faults of DC Rxs. The LO radiation leaks from the Rx into nearby wiring. Because this wiring is connected either directly or indirectly to power rectifiers that are switching on and off, the RF impedance of the wiring varies throughout the 50Hz mains cycle. This modulates the LO radiation, giving it 50Hz sidebands. The radiation then finds its way back into the receiver front-end. The signal is then demodulated, causing a 50Hz related buzz in the headphones.

The cure for this is to stop LO signal radiating from the receiver cabinet. All connections to the Rx, such as the power supply and the headphones, should be RF grounded to the receiver case using decoupling capacitors appropriate to the LO frequency.

fig 5
Fig. 5: Probe for tracing RF hum.

To find the source of RF hum, make a probe as in Fig 5. This is a screened loop. When the diode is conducting, the loop is made. When the diode is not conducting, the loop, to RF eyes, does not exist. It therefore makes RF hum at the AF oscillator frequency. Wave the probe around in the receiver. Where the whistle is loudest, the leakage is greatest.

One other possible problem is included here because its mechanism is the same as RF hum, even though it sounds different. If the decoupling of the headphone lead is insufficient, its RF impedance can vary with the audio waveform, and this can cause the Rx to oscillate at audio frequencies (Fig 6). This effect is easily mistaken for AF amplifier instability. In receivers with AGC, it manifests itself as a slow plopping noise. An isolating audio output transformer will help to cure this.

fig 6
Fig. 6: RF feedback causing apparent AF instability.

On VHF, LO radiation from the antenna feeder can be a cause of hum, so use a high quality low leakage feeder from the Rx to a point well away from mains supplies; from that point to the antenna, cheaper cable can be used. If cable screening is a problem, the coaxial socket on the Rx should be an N-type orTNC. Because of the RF hum problem, DC Rxs do not work well on the higher frequency bands (20m and up) with either set-top antennas, or long-wire antennas connected directly to the back of the set. If the mixer balance is poor, hum can be caused by direct demodulation of LO hum sidebands; make certain that the LO power supply is ripple-free.

Mixers for dc receivers

Use a broad-band mixer to avoid microphony: mixers using high Q tuned circuits should be avoided. In addition, the mixer should have a high second order intercept point. If this is not high enough, high-powered HF AM signals will be rectified directly to AF, and the Rx will operate as a broad-band AM detector(5) of both the RF and LO inputs. Our first VHF receiver was very good at this; it picked up all the stations on the 7MHz broadcast band at the same time.

This limits the choice of mixer to those with some inherent balance. Do not even attempt to use a mixer with a single bipolar/FET/ diode. Trying to make a satisfactory DC receiver with a self-oscillating mixer would probably be a passport to madness. Circuit balance cancels out second order distortion; push-pull power amplifiers cancel out their second harmonics using this effect.

Of course, the broadband AM detector effect is not apparent in the superhet because the AF component is removed by the IF filter. Beam deflection mixer valves are certainly worth some experimentation. One way of avoiding LO radiation is not to make the frequency in the first place.

This the main advantage of the sub-harmonic anti-parallel diode mixer as propounded by RA3AAE(6). This mixer has its oscillator drive at half the standard LO frequency; the LO leakage can then be stopped by the input bandpass filter. Of course, the second harmonic of the oscillator is now at the RF frequency, so the sub-harmonic mixer reduces LO radiation and all its attendant problems by an amount equal to the suppression of the second harmonic with respect to the fundamental.

Unfortunately, the RA3AAE sub-harmonic mixers do not have as good a strong-signal handling performance as ring mixers. The other disadvantage is that the ratio of RF-in to wanted-product-out, known as the conversion loss, is very dependant on LO drive level, and also depends slightly on the shape of the LO waveform. The rectifying action of the diodes in a sub-harmonic mixer generates harmonics of the LO. So, unless care is taken with diode balance, the level of leakage at twice the LO frequency appearing at the RF port can be similar to that of a ring mixer. A variant(6) of the RA3AAE type of switching harmonic mixer is shown in Fig 7. Its conversion loss is about 6.5dB. This is similar to the figure obtained with the narrow-band mixers of Reference 7, which operate in a similar way. The third order input intercept point of the mixer in Fig 7 is about +3dBm/tone.

fig 7
Fig 7: A broad-band sub-harmonic mixer; conversion loss is shown in Table 1.

Using transmission line transformers gives the mixer a wide bandwidth, and a high degree of balance is achieved by using the Siemens surface-mounted BAS40-04 dualdiode (the size is only 3 by 1.4mm, so be careful not to lose them).

With a perfectly balanced circuit, all even order harmonics of the LO cancel, and there would be no LO leakage at the RF frequency. In practice, the leakage at the RF frequency due to the LO is smaller than -50dBm. The 'AF' bandwidth is 0 to 1 MHz.

The rise in conversion loss at high frequencies is mostly due to the diode capacitance: this can be reduced by using four BA481 s; by doing this, and by using higher frequency transformers, the conversion loss at 70cm is reduced to 7.5dB, but the circuit balance is not as good as with the BAS40-04s. One of the quad (meaning four diodes in a single package) mixer diodes, such as the HP 50822830 from Farnell, can also be used in this circuit.

Table 1
Frequency (MHz)Conversion Loss (dB)
1.96.5
306.9
707.4
1447.8
43211.6

fig 8
Fig 8: A broad-band sub-harmonic mixer using the SBL1 or SRA1H

For those that do not wish to wind transmission line transformers, or solder surface-mounted devices, help is at hand. It is possible to make a half-frequency sub-harmonic mixer using standard diode ring mixers of most types, eg SBL1 or SRA1 H. They work well, provided that the maximum reception frequency is less than 200MHz. The circuit is shown in Fig 8, which can be used as a replacement for Fig 7.

The circuit works because pins 1 and 2, which would normally be connected to the RF input, are shorted to earth. The shorted primary winding has the effect of shorting all the connections of the transformer secondary together. A little thought, and mental shuffling of diodes, then suffices to show that the mixer is now the equivalent to the circuit of Fig 7.

Above 200MHz, the mixers in Fig 8 work to a degree, but the amount of LO drive required varies wildly from one frequency to the next. This is probably due to the fact that the shorting of the transformer primary does not produce a very good short on the secondary at high frequencies, and some resonance effects start to creep in.

For use above 200MHz, the mixer in Fig 7 is much more reliable. As has been mentioned before, the LO drive power for sub-harmonic mixers must be carefully optimised to provide minimum conversion loss. Fig 9 shows a graph of the drive levels required to achieve a conversion loss of 8dB for the mixer in Fig 7, and for either an SRA1 H or an SBL1 when used in the circuit of Fig 8. Within the upper and lower contours, the conversion loss will be less than 8dB.

fig 9
Fig. 9: LO driver power requirement of the various sub-harmonic mixers.

Af filtering for dc receivers

The af filtering in a dc receiver provides the selectivity, in effect performing the same function as the crystal filter in a super-het. To protect the AF amplifier from large unwanted signals above 4kHz, this filtering should be done as soon after the mixer as possible.

When viewed from this aspect, the ideal place to put the AF filter is between the mixer and the first audio amplifier. In this position, the filter must be a passive inductor-capacitor (LC) type, because resistor-capacitor (RC) filters are too lossy. Unfortunately, high value inductors & capacitors in this position tend to pick up AF hum and microphony.

Filters after the first AF amplifier can, of course, be either active or passive. There is thus an engineering tradeoff to be made, based on the number and power of the signals within about 100kHz of the wanted signal. On the lower HF bands, adjacent signals are large, so good filtering is important. On VHF, unless the next-door neighbour uses the same band as you, adjacent signals are smaller. This allows the designer to concentrate on achieving lower levels of hum and microphony by putting more of the filtering after the first AF amplifier. For similar reasons, either the SRA1 H or SBL1 used as a conventional ring mixer is a good choice for the lower HF bands; both have good strong signal handling capacity. In contrast, the sub-harmonic mixers with their lower leakage, but lower signal handling capacity, are good choices for the VHF bands.

Diplexers for the mixers of dc receivers

In good receiver design, all the connections to the diode mixer are properly terminated. This prevents signals being reflected back into the mixer which would spoil its inter-modulation performance. Although a completely broadband termination is desirable, it is especially important that two frequency bands at the mixer IF/AF output are terminated. They are LO + RF and LO - RF.

Take, as an example, a 7MHz DC Rx with an RF filter having a 0.5MHz bandwidth centred on 7MHz. For this receiver, the two important frequency bands for termination at the output of the mixer would be 13.5 to 14.5 MHz (LO+RF) and 0 to 0.5 MHz (LO-RF). Unfortunately, we wish to put our audio filter in this part of the circuit, and the filtering falls within the mixer's termination sensitive zone of 0 to 0.5 MHz. Consequently, if a filter is to be used between the mixer and the first AF amplifier, it must be of the sort called a diplexer. This is a network of various lossless filters which, if correctly terminated, gives a constant input impedance at one or more of the network's connections. This constant input impedance side is connected to the mixer output.

Each of the sub-harmonic mixers in Figs 7 and 8 uses a diplexer to separate the 'AF' out from the RF in. This is the simplest form of diplexer, and consists of the 8.2µH RF choke and the 3.3nF capacitor. More complicated diplexers can be designed from the information in Annex A.

fig 10
Fig 10: Mixer, diplexers and first AF stage for an HF receiver.

Fig 10 shows a mixer/first AF stage for use on HF. It has an SBL1 mixer, two diplexers and an AF amplifier. The first diplexer uses the same components as the sub-harmonic mixers. It ensures a good match for the LO + RF frequencies and has a crossover frequency of 1 MHz. The second diplexer has a crossover frequency of 4kHz to provide AF filtering. The capacitor between L10 and the 43Ω resistor is included to provide extra filtering. It does causes an impedance mismatch at 5kHz, but this is not too serious when compared to the difficulty of winding the extra inductor that would be required to give a perfect diplexer. The transformers on the LO and RF ports of the SBL1 reduce the LO leakage by providing extra mixer balance. As long as the AM noise on the LO is low enough, the LT1028 (Maplin) will provide a lower noise figure than the OP27.

Conclusion

This article has a moral for the designer of direct conversion receivers: "Keep your local oscillator under control on a short lead at all times, and never let it out in public." The higher the Rx frequency, the more the effort that must be made to achieve this. This frequency dependence is due to two effects. Firstly, the amount of signal radiated by short wires increases with increasing frequency, so causing increased LO leakage at high frequencies. Secondly, the noise power from the antenna decreases with increasing frequency; so that, for a given level of LO leakage, hum and microphony sidebands will be more apparent above the noise at higher frequencies, because the antenna noiise is less there.

To reduce LO leakage, use a separate well screened LO compartment with properly filtered supply and control lines. Use a non-resonant balanced mixer and an isolator stage to keep the mixer's LO leakage away from tuned circuits in the pre-amp or input filter. To reduce the hum level, use an external power supply unit and try to stop LO leakage from radiating from the receiver's case.

To help stop microphony, use a solid style of construction, and mount all RF circuitry on double sided printed circuit board. You will find that the DC receiver is one of a very select group of radio circuits that actually works better in a box than when it is bird's-nest built. Finally, there is the coward's way out; use a preamp with 40dB gain. This will drown microphony and hum in a flood of front-end noise. It will also degrade your receiver's dynamic range by 30dB or so.

Acknowledgment

The author thanks the other members of the club project team for their help. They are Dick G3WRJ, Hugh GOLGV and Pete G8EMJ.

References

  1. 'Keep it Simple: Direct-conversion HF receivers', Pat Hawker, G3VA, IERE Conference on Radio Receivers and Associated Systems, IERE Proceedings No.40 July 1978, oo137-148
  2. 'Direct conversion SSB receivers', SR AlAraji & W Gosling, The Radio & Electronic Engineer, March 1973, pp209-215
  3. 'A new negative feedback amplifier', Victor Koren, RF Design, Feb 1989, pp54-60
  4. Handbook for the Radio Amateur, ARRL, 1986, p12.8
  5. Radio Receivers, W Gosling, pp177-178, Pub. Peter Peregrinus Ltd. 1986 for IEE
  6. Technical Topics, G3VA, Radio Communication, April 1977, pp290-291, based on article by RA3AAE
  7. "Twin-diode mixer - a new microwave mixer", Jim Dietrich, WAORDX, Ham Radio, October 1978, pp84-86
  8. 'Matching Circuits for Schottky Ring Mixers', J.Kestler, DK1OF, VHF Communications, 1/1976, pp13-18

Annex A

The design of HP/LP diplexers with more than two components

fig 11
Fig 11

The standard formulae used until now by the radio amateur in designing diplexers are given in Reference 8. This gives the design data for LP/HP diplexers using one inductor and one capacitor. These have the disadvantage that the transition between the passband, through the crossover frequency, and into the stopband is rather slow. For example, if a simple diplexer has a crossover frequency of 4kHz, the loss due to the filtering of the lowpass side is still 1 dB at about 2kHz.

Much more rapid transitions can be achieved by using HP and LP filters with more than one component each. These diplexers can be designed from the standard tables of LC filter element values contained in Al Zverev's Handbook of Filter Synthesis (1967) which have been reproduced by almost every subsequent book on filters. The trick is to usethe values for Butterworth T filters with a source impedance of zero Ohms. Some of these data are shown below; from them HP/ LP diplexers of up to 10 components can be made. Bandpass/bandstop filters could also be made by applying the usual transformations.

To denormalise, multiply all capacitors by C and all inductors by L.

eg. Design a perfect four-component diplexer to replace the second diplexer in Fig 10.

Choose fc = 3.3kHz and Ro = 50Ω, so calculate C = 0.965 F and L=2.41 mH from the table. C1 = C2 = 0.7071C=0.68µF and L1 = L2 = 1.4142L = 3.4mH so the circuit for the diplexer is:

fig 12
Fig 12

G4TXG, Nic Hamilton