Fullscreen HTML5Med Res (??MB)HTML4

SSB Shortwave Receiver Part 1 by Charles Kosina, VK3BAR While there are plenty of cheap radios these days, including software-defined types, I decided to build this analog shortwave radio for the satisfaction of making it myself. I learned a lot about shortwave, SSB and how radios work in the process, which you will not get just buying an ‘appliance’! R adio receiver architectures have changed dramatically in the last few years. Digital techniques have largely displaced the analog techniques from the past. Radio receivers are now available at ridiculously low prices from various internet sources. The simplest ones are the Software Defined Radios (SDR) that are a small module that plugs into a USB port. The typical coverage is from 100kHz to 2GHz; they rely on the processing power of the attached computer to recover the desired signal. They are not ‘communications receivers’, as their noise figure and immunity from intermodulation are quite poor. With no input tuneable filter, a strong signal can easily overload 46 Silicon Chip the front-end circuitry. But for many, the performance is quite adequate. The screen “waterfall display” showing signals is very useful. Still, it isn’t too hard to build an analog shortwave receiver with decent performance, as I shall explain shortly. Performance The performance of this unit is quite reasonable. I set my signal generator to 1µV (-107dBm) output and the signal-to-noise ratio was about 13dB for most of the range (slightly less at 30MHz). Inserting a 20dB attenuator gave me a signal of about 0.1µV (-127dBm) and it was still audible! At that level, communication via Morse Code (CW) Australia's electronics magazine would be possible, but it is too weak for SSB voice reception. Yes, there are the unavoidable birdies, but they do not interfere greatly. What about on-air tests? The only HF antenna I have at present is an endfed half-wave dipole on the 40m band (7MHz). This is fed with 50W coaxial cable and uses a 49:1 ‘unun’ (unbalanced to unbalanced) transformer. The measured SWR (standing wave ratio) is between 1.2 and 1.3. Unfortunately, the ambient noise here is rather high with all the electrical equipment in the surrounding houses, producing heaps of RF hash, so it needs a fairly strong signal to get through. Comparing this receiver with my Bando Technic 5D transceiver, the sensitivity is much the same. siliconchip.com.au Fig.1: a TRF receiver comprises several identical RF amplifiers tuned to the same frequency. Most very early radios used this configuration. Fig.2: the superhet was an early game changer. By mixing the amplified, tuned incoming signal with an oscillator frequency that tracks above it, the signal is shifted down to a lower, fixed (intermediate) frequency. Signals at that frequency are easier to filter out and demodulate. Fig.3: an SSB receiver is a bit more complex as it needs to operate without the carrier wave or half the signal spectrum. The modulated signal is recovered by mixing it with the output of the BFO in a second mixer stage. A list of the features and specifications for this receiver includes: ∎ Covers the shortwave band from 3MHz to 30MHz in two spans ∎ Sensitivity: 1μV (-107dBm) for a 13dB SNR (reception possible <at> 100nV/-127dBm). ∎ 2.7kHz speech bandwidth ∎ Runs from 12V DC <at> 500mA ∎ Digital tuning with frequency display ∎ Analog controls for tuning, volume, RF gain and squelch ∎ USB/LSB decoding ∎ Fast or slow AGC ∎ RSSI display ∎ Built-in speaker and headphone jack Radio receiver types I will start with a brief summary of radio techniques over the last 100 years or so. Back in the 1920s, a typical radio was most likely a TRF (tuned radio frequency) set. They consisted of several cascaded tuned amplifiers, each set to the same frequency, as shown in Fig.1. A detector extracted the audio signal at the end of the chain, which was siliconchip.com.au then amplified to drive headphones or a loudspeaker. The valves used were initially triodes, and with high feedback capacitances, were subject to instability, leading to oscillation. A neutralising system was developed to feed back a phase-shifted version of the signal, minimising or preventing this instability. Later, tetrode and pentode valves were used that had much lower feedback capacitance and obviated the need for neutralisation. Not only were these early radios difficult to set up and use, but they lacked the selectivity to reject unwanted strong signals on different frequencies. However, by using regeneration, where a portion of the amplified signal is fed back to the input grid of the triode and pentode, much higher gain and selectivity could be achieved. Enter the superhet The problems of the TRF receiver were largely overcome by the superheterodyne architecture. Edwin Armstrong is often credited with the invention of this technique, but others filed Australia's electronics magazine patents only months apart. Legal battles followed, and French engineer Lucien Lévy was awarded a patent that included seven of the nine claims in Armstrong’s application. Fig.2 shows the superheterodyne architecture. The incoming signal passes through a tuned filter, followed by an optional RF amplifier. Then follows the mixer, where the incoming signal is multiplied by another frequency from a local oscillator. The multiplication results in two extra signals, being the sum and difference of the frequencies. For example, if the incoming signal is 1000kHz, and the local oscillator is at 1455kHz, the output from the mixer will contain signals at 455kHz (1455kHz – 1000kHz) and 2455kHz (1455kHz + 1000kHz). It will also include the original 1000kHz signal. The following band-pass filter selects just the 455kHz portion, which is amplified by the IF (intermediate frequency) stage(s) and passed to a detector and audio amplifier as before. However, this technique is only suitable for receiving AM (amplitude modulated) signals. June 2025  47 To receive SSB (single sideband) broadcasts, discussed further below, an additional mixer stage is necessary after the IF amplifier(s). This mixes the IF signal with that from a BFO (beat frequency oscillator), and its output goes through a low-pass filter (LPF), as shown in Fig.3. This architecture remained the normal way that radios were built for many decades. Most of the selectivity (ie, rejecting unwanted station signals) came from the IF filter. However, consider a broadcast at 1910kHz with the common 455kHz IF. With the set tuned to 1000kHz, when mixed with the LO at 1455kHz, this signal also will produce a 455kHz output from the mixer. This is termed the image frequency, and it is why there is an input band-pass filter, to attenuate this image. A single tuned circuit was adequate for the broadcast band frequency range of 530kHz to 1,600kHz, but once shortwave broadcasting became commonplace, the single tuned circuit was inadequate at frequencies above 3MHz and resulted in ‘double spotting’ of the same input signal. This was often tolerated, but to get around it, double- and triple-­ conversion superhet sets were used for better performance. This resulted in a higher-frequency initial IF signal, which was then mixed again to obtain a lower-­frequency secondary IF signal. There are some limitations to the superheterodyne architecture. Spurious signals are generated as a result of the mixing process or from non-ideal components (like harmonic distortion). Spurious signals may be produced by the oscillator, mixer, or other components in the receiver. The local oscillator may generate harmonic signals that mix with the RF signal, producing unwanted spurious signals. In nonlinear systems, two or more signals can combine to produce additional unwanted frequencies, known as intermodulation distortion. Fig.4: this Hartley SSB receiver configuration is difficult to implement in hardware as a very accurate 90° phase shift is required across a range of signal frequencies. Fig.5: this alternative configuration is similar to the Hartley type except the 90° phase shift is split into two 45° phase shifts. It’s still difficult to make it work in the analog domain, though. 48 Silicon Chip Australia's electronics magazine Some of these spurious signals are characterised by a rapid tuning rate. The whistle or chirp that is produced changes in frequency much faster than the tuning of the receiver. Hence, they were called “birdies”. Hartley phasing There are alternatives to the superheterodyne receiver. A variation is to use the Hartley phasing method, as illustrated in Fig.4. The incoming signal (ωs) is fed into two mixers. The local oscillator is at the same frequency, and an RF phase shift network of 90° will mix with the incoming signal to produce two signals at baseband, but at 90° apart. These signals are called I (in-phase) and Q (quadrature). The Q signal is applied to an audio phase shift network, which in mathematical terms is a Hilbert transform. This shifts the entire audio spectrum by 90°. However, this arrangement is impractical. A better approach is using two separate phase-shift networks of +45° for the I signal and -45° for the Q signal, as shown in Fig.5. These are then summed and filtered to produce the demodulated signal. This is a simplified explanation of the phasing system; there are plenty of online references that give a detailed mathematical analysis. The phasing method is elegant in its simplicity, but there are practical problems in its realisation. It is relatively easy to have an accurate 90° phase shift, but the audio phase shift network requires an extremely high precision in components to maintain the accurate phase shift over the whole range of frequencies. This is why the analog method has been superseded by digital techniques. In current SDR receivers, the I and Q signals are sampled by analog-­ to-digital converters and the Hilbert transform is done by software. It does require a fast processor, as found in modern computers. I decided to investigate if a phasing receiver was practical using an analog phasing network. There are designs available to implement the Hilbert transform in hardware, but it requires careful matching and selection of components, preferably to within 0.1%. One such design is shown in Fig.6, and I built a test module to test its practicality. I bought about 50 of the 10nF siliconchip.com.au Fig.6: an example of a phase shift network that provides a more-or-less fixed phase shift across a range of frequencies. capacitors and, by measuring them to four-figure resolution, I selected a batch where all were within 0.1% of each other. The exact value is not quite as important as the matching. I originally thought that 8-pin SIL (single in-line) resistor networks with four 10kW resistors each would be closely matched, but found that was not accurate enough. The alternative was to select from lots of 10kW SMD resistors for a matched set. The hardest part is getting the other six resistors to an exact value. For example, a value of a 12,960W is needed, which is realised by two resistors in parallel, 13kW and 3.3MW. But this required measuring and selecting resistors that were close to the nominal Fig.7: the performance of the phase shift network shown in Fig.6. Even with hand-selected matching components, it doesn’t quite hit 90°, nor is it perfectly flat with frequency across the band of interest. siliconchip.com.au value. Some values were difficult to get exactly. Fig.7 shows the measured phase shift of my prototype which, while close to 90°, is not really close enough. The sideband rejection would be 40dB at best. Also, this is not the sort of design that can many readers would bother to build. It is possible to eliminate the Hilbert transform; one solution is the Weaver architecture. Following the baseband low-pass filters, we have another pair of mixers, as shown in Fig.8. The frequency injected into the second pair of mixers is at about half the bandwidth. Again, two signals 90° out of phase are needed, called the pilot tone. A further LPF extracts the wanted signal. There are many articles and papers describing the Weaver method, many with quite complicated mathematics. It is described in detail at siliconchip. au/link/ac51 I built a receiver with this architecture, copying some of the design ideas available on the internet. After many hours of trying to get decent performance, I eventually gave up. Getting the accuracy and balance between the I and Q channels just proved too hard. Overall, the receiver was far too noisy; I suspect because of the multiple mixers, and I could not get rid of the pilot tone in the output. I would be interested to hear from readers who Fig.8: the Weaver receiver configuration has some advantages over Hartley but many more mixers are required, so the resulting noise performance is less than ideal (in the analog domain, anyway). Australia's electronics magazine June 2025  49 may have built a Weaver receiver and find what their results were. Having tried all the different architectures (apart from TRF) over a period of about six months, I decided that the SSB superhet design was the most practical approach for home construction. But before we get to the circuit, here is an explanation of two modulation techniques. Amplitude modulation (AM) is where a ‘carrier frequency’ signal is Fig.9: the basic principle of amplitude modulation (AM). The high-frequency carrier amplitude varies with the instantaneous baseband signal amplitude. Fig.10: the spectra of AM and SSB transmissions. The transmission power of SSB is about ¼ that of AM without significantly reducing the received signal strength. The ultimate design multiplied by an audio frequency (AF) signal, as shown in Fig.9. We get a signal with components in three frequency ranges: the original carrier, plus two ‘side-bands’, being the sum and difference (see Fig.10). To demodulate the AM signal, all that is needed is a diode and a lowpass filter to remove the RF component. This filter may be just a single resistor and capacitor. While AM is easy to implement, is really quite wasteful. The carrier frequency carries no information at all, and the two side-bands at 100% modulation contain half the power of the carrier, with identical information. This is where the single side-band (SSB) method of communication is far more efficient. We essentially get rid of the carrier and one of the sidebands. Instead of a bandwidth of twice the baseband, our filter needs only the baseband bandwidth. The spectrum for SSB modulation is shown at the bottom of Fig.10. However, with SSB, the simple envelope detector will no longer work. To take an example, transmitting an SSB signal modulated at two frequencies, 1kHz and 2kHz, an envelope detector would give us a tone of 1kHz, being the difference between the two frequencies. For a more complex modulated signal, the output of the detector would be quite unintelligible. To recover the audio, we have to multiply the output of the IF amplifier with the signal from a beat frequency oscillator (BFO) with a second mixer. The BFO frequency is set to where the carrier frequency would otherwise be. This results in two signals in the output; one is the original baseband signal, plus another at twice the IF, which is easily removed by a low-pass filter. The filtered signal can then be amplified by an audio amplifier to drive a speaker or headphones. Receiver design Fig.11: the measured performance of the pre-built 9MHz crystal filter module. It combines a flat passband with very steep roll-offs on either side. The receiver presented here covers the frequency range of 3MHz to 30MHz, with an audio bandwidth limited to the frequencies used by human speech: 300Hz to 3kHz. This means that the IF filter bandwidth needs to be 2.7kHz (3kHz – 300Hz). This is best achieved by a multi-pole crystal filter at 9MHz. This is quite a critical item in the design. You can build your own by buying a batch of 9MHz crystals and carefully selecting them for series and Australia's electronics magazine siliconchip.com.au 50 Silicon Chip The front and rear sides of the control board. The five pots, three switches, rotary encoder, LCD screen and headphone socket form the user interface. On the rear of the control board are the Arduino Nano and clock generator modules, LCD adjustment trimpot, two electrolytic capacitors and some connectors. Note that these photos are shown enlarged for clarity. parallel resonant frequencies. But unless you have the equipment and patience to do this accurately, it is not worthwhile. I bought 20 9MHz crystals for about $6, and by selection, managed a reasonable filter after much experimentation. But a complete six-pole filter module is available from AliExpress siliconchip.com.au for about $25, with an excellent bandwidth, as is shown in Fig.11. Next, let’s look at how we deal with image frequencies. If the desired signal fs = 7MHz and the local oscillator fo = 16MHz, producing a 9MHz IF signal, a signal at 25MHz mixed with 16MHz will produce the same 9MHz IF. This is why we have an input tuned circuit. Australia's electronics magazine How sharp does this filter have to be? Using a high-quality toroid, a loaded Q of 100 is typical. There are calculators on the internet that save us the trouble of laboriously working it out; with the above example, the unwanted 25MHz image signal will be attenuated by about 50dB. That is why a relatively high IF is June 2025  51 What is a noise figure? Every device generates broadband noise that will reduce the circuit’s signalto-noise ratio (SNR). The NF is the ratio of actual output noise to that which would remain if the device itself did not introduce noise, which is equivalent to the ratio of input SNR to output SNR. There is another way of expressing the noise performance: the noise temperature, expressed in Kelvins as an equivalent temperature. It is not the physical temperature of a system, but a theoretical value that defines the temperature required to produce a specific amount of noise power. The equivalence between noise temperature and noise figure is shown below. The reference temperature, Tref, is generally 290K (16.85°C). The relationship between noise figure (NF) and noise temperature (in Kelvin). Note that it is not the actual temperature the part is operating at. desirable for higher-frequency signals, as the image frequency is well removed. If an IF of 455kHz were used, the standard for broadcast-band receivers, the image at 7.91MHz would only be 28dB down. Circuit details Figs.12 & 13 show the full circuit of the receiver, which is split across two PCBs, and the circuits correspond to them. One is the control board, while the other is the RF board. At the heart of the control board is the Arduino Nano module, which has the ATMega328 microcontroller. The display is the common 16×2 alphanumeric LCD module; the version with a blue backlight is the best choice. Potentiometer VR6 is the contrast adjustment for the LCD screen. The variable frequency oscillator (VFO) and the beat frequency oscillator (BFO) signals are generated by an Si5351A clock-generator module (MOD2), controlled over an I2C serial bus (SDA/SCL). This module can generate three different frequencies as square waves with amplitudes of about 3V peak-to-peak. In this design, the outputs used are CLK0 and CLK2; the CLK1 output is not used. The 8.2kW pull-up resistors for the SDA & SCL lines are shown greyed out in Fig.12 because they do not need to be fitted as the Si5351A module has onboard pull-up resistors. A rotary encoder (RE1) is used for frequency tuning; the step size for each click can be varied using the integrated pushbutton switch. Pressing the switch cycles through steps of 10Hz, 100Hz, 1kHz, 10kHz, 100kHz and 1MHz. The two poles of the encoder, plus its integral switch, have 33kW pull-up resistors to give defined high/ low levels and 100nF capacitors to ground for debouncing. The I2C serial bus is used to control the input circuit tuning by selecting six capacitors in various combinations for approximate tracking with frequency. Fine potentiometer VR1 is used to change the voltage on a varicap diode to interpolate the approximate values and peak the input circuit to resonance (more on this later). There is audio circuitry on this board as well. Op amp IC1b has a gain of about 5.5, and can drive headphones directly via a 3.5mm jack provided on the front panel. It has internal switching that disconnects the power amplifier driving the speaker when headphones are plugged in. Despite the existence of numerous more modern power amplifier chips, I have used the venerable LM386 (in an SMD package) to drive the speaker. It requires few external components, is cheap and with a 12V supply will deliver over 2W to an 8W speaker Fig.12: the control board circuit. Three main modules are used: the Arduino Nano ‘brain’, an Si5351 digital clock generator that produces the VFO and BFO oscillator signals and the 16×2 alphanumeric LCD module. The dual op amp provides the squelch function (IC1a) and audio gain for driving headphones (IC1b), while IC2 is the power amplifier that drives the speaker. 52 Silicon Chip Australia's electronics magazine siliconchip.com.au The control board and RF board are joined by a 16-wire flat cable between headers CON2. This supplies power to the RF board and carries signals to it as well, including the band change signal and the I2C bus (SDA & SCL). Signals coming into the control board on CON2 include the recovered audio and RSSI (received signal strength indicator) voltage. Potentiometer VR5 is the volume control, while VR2 is an RF gain control modifying the AGC (automatic gain control) voltage. Switch S3 is the SLOW/FAST AGC selection, adding a 10µF capacitor to the RSSI line for the SLOW AGC mode. The squelch control is useful in eliminating background noise from weak input signals. It works by comparing the RSSI voltage level at the inverting input of IC1a (which is used as a voltage comparator) with a DC voltage derived from potentiometer VR4. When the RSSI level is low, Mosfet Q1 is switched on, shorting out the audio. The 1MW feedback resistor provides hysteresis. The mute function is provided by a second transistor, Q2, in parallel with siliconchip.com.au The controls are all labelled on the front panel PCB. The rear of the set only has the BNC antenna terminal, DC power connector and holes so that sound from the the internal speaker can escape. Q1. I found that when the frequency was being changed, there was a loud annoying click in the audio. So during tuning, Q2 is switched on, also shorting out the audio. DC power is via CON1 and diode D1 protects against the wrong supply Australia's electronics magazine polarity. The supply voltage can range from 9-12V DC, with a maximum current drain of about 250mA. An ironcore transformer based plugpack is preferable as it does not generate RF noise, but you can try a switching plugpack; some do have low noise. June 2025  53 Fig.13: this RF board circuit connects to the control board circuit (Fig.12) via CON2. Q1-Q6 and VD1 tune the incoming signal while T1 (3-10MHz) or T2 (10-30MHz) are selected by RLY1 for band switching. Q8 is the RF gain stage; IC1 is the superhet mixer; Q9 is the first IF gain stage; Q10 is the second IF gain stage; IC2 is the BFO mixer; and dual op amp IC3 is the RSSI/AGC signal amplifier. Because the voltage regulation of iron-core plugpacks is poor, a 12V one may put out a voltage that is too high with a light load, so choose one rated at 9V DC and 500mA. Depending on the Arduino Nano, the voltage regulator may not tolerate an input voltage much greater than 12V, so be careful with the choice. You can easily blow up a Nano with excessive input voltage (trust me, I have!). The filtering on this type of plugpack may leave too much 100Hz ripple, which would be heard as hum in the output. That’s why there is a 2200µF electrolytic capacitor after D1. When S1 is switched on, there is a very high inrush current to charge this capacitor, hence a fairly high-­current schottky diode is used for D1. The ideal supply would be a ~12V battery; three 18650 cells in series give just over 11V fully charged. The background noise using a battery is significantly lower than either type of mains-powered supply. A suitable battery holder can be squeezed into the case, although taking out the cells to charge them requires removing a bracket. Still, you could integrate a charging socket. 54 Silicon Chip Most constructors will not need the optional serial debugging interface provided by Mosfets Q3 & Q4; they offer a bidirectional RS-232 compatible serial stream at header CON5. Those components can be left off if not needed. You can also connect a TTL USB/serial adaptor directly to the TXD & RXD pins of MOD1. RF module circuit As shown in Fig.13, the signal from the antenna goes to two tuned toroidal transformers selected by relay RLY1. A high Q is desirable in these transformers for maximal rejection of unwanted frequencies. The toroids are Micrometals T37-17 types with an unloaded Q in excess of 200 at most frequencies. With a 50W source on the primary winding, the loaded Q will be about 100. Transformer T1 covers the range of 3-10MHz and has a secondary inductance of 7.4µH (42 turns). The antenna winding is four turns at the ‘cold end’ of the toroid. This needs a capacitance range from 34pF at 10MHz to 380pF at 3MHz for tuning. Back in the days when valves were used, this capacitance would be part of a two- or three-gang variable Australia's electronics magazine capacitor. These days, such capacitors are relatively rare, expensive and too large. Instead, I used six fixed capacitors selected by the PCF8574 I2C extender (IC4) driving six NPN RF transistors. A BB910 varicap diode (VD1) adds to the tuning capacitance as a fine adjustment to interpolate between the fixed values. The capacitance range of the varicap is from 40pF at 0.5V down to 8pF at 9V (a varicap diode is used in reverse bias, with the voltage across it affecting its capacitance). For the range from 10.1MHz to 30MHz, we switch in transformer T2, with an inductance of 1.1μH (15 turns). Its antenna winding is two turns. Q8 is a low-noise amplifier based on a BF998 dual gate Mosfet. While technically obsolete, this is easily obtainable from many sources. Rather than another tuned circuit in the drain, I have just used a 100μH inductor, which has a reasonably high impedance over the entire 3-30 MHz range. The gain is about 20dB and, while the noise figure (NF) is not given below 30MHz, at 800MHz it is 1dB. The first mixer (IC1) is an NE612 (or SA612) IC. This has a gain of about siliconchip.com.au 17dB and a noise figure of 5dB. The NF of a multi-stage amplifier can be calculated as: NF = NF1 + (NF2 – 1) ÷ G1 + (NF3 – 1) ÷ (G1 × G2) + ... Here, NF is the total noise figure, while NF1, NF2... are the noise figures of subsequent stages, and G1, G2... the gains of the stages. Thus, with a reasonably high gain in the first stage, the overall noise figure is degraded only slightly by the following stages. The final noise figure of our circuit is about 1.5dB. That is more than adequate given the amount of ambient noise in the HF band. Another BF998 (Q9) follows the first mixer, providing another 20dB of gain. The 9MHz crystal filter follows, which has 50W input and output impedances. This is matched to the preceding BF998 amplifier by a pi network with a 3000:50W ratio. The filter introduces a loss of about 5dB. Another BF998 (Q10) is the second 9MHz IF amplifier after the crystal filter. A second NE612 (IC2) is used for the second mixer, with the ~9MHz BFO. There are two outputs available on the chip. One output is connected to op amp IC1b, which has a voltage gain of about 46 times. This is the AGC amplifier. Schottky diodes D2 and D3 rectify this voltage and charge a 1μF capacitor. This voltage is applied to the inverting input of IC3a. With no signal, this voltage is close to zero. The non-inverting input has a voltage from the RF gain control potentiometer on the front panel, and with the resistor values used, the output of IC3a is a maximum of about 4.5V. This is applied to the second gate of the three BF998 transistors for maximum gain. As the RSSI voltage rises, the AGC voltage drops, going down to zero for very strong signals for minimum gain. The assembled RF board - toroidal transformers T1 & T2 are on the left, while the crystal filter module is at lower right. Note that this photo is shown enlarged for clarity. siliconchip.com.au Australia's electronics magazine June 2025  55 The maximum supply voltage for the NE612 mixers is 9V, so 8V is provided by a 7808 regulator. With a 9V main supply voltage, this will drop to about 7.5V, which is quite adequate. The PCF8574 I2C I/O expander driving Q1–Q6 is the only chip that needs a 5V supply, which is provided by an SMD 78L05 regulator (REG2). Parts List – SSB Shortwave Receiver We’ve covered quite a lot in this article, so the construction details will be in a follow-up article next month. It will also cover programming the Arduino Nano, preparing the case, plus calibrating and aligning the Receiver. SC 1 180 × 130 × 110mm blue vented steel project box with feet [AliExpress 1005008418042828] 1 assembled control board (see below) 1 assembled RF board (see below) 1 front-panel PCB coded CSE250204, 165.5 × 97mm, with black solder mask 1 panel-mount DC barrel socket, diameters to match plugpack 1 12V 500mA+ plugpack 1 8W all-purpose loudspeaker (SPK1) [Jaycar AS3025, eBay 7.7cm 5W 226113532195] 5 13mm diameter universal knobs [AliExpress 1005006143033779] 1 25mm diameter universal machined aluminium knob [AliExpress 1005007577048515] 1 10cm male SMA to female BNC panel-mount connector cable [AliExpress 1005003990025513 select “BNC F WATERPROOF 2”] 2 16-way IDC connectors 2 2-way 2.54mm pitch polarised header plugs with matching pins 1 20cm length of 16-wire ribbon cable 4 M4 × 10mm panhead machine screws, nuts & washers (for mounting SPK1) 4 M3 × 15mm tapped spacers 4 M3 × 10mm tapped spacers 12 M3 × 6mm panhead machine screws 4 M3 × 6mm black panhead machine screws Control board 1 double-sided PCB coded CSE250202, 150 × 79.5mm 1 Arduino Nano programmed with CSE25020A.HEX (MOD1) 1 Si5351A clock generator module (MOD2) [AliExpress, eBay etc] 1 16×2 alphanumeric blue backlit LCD module (LCD1) 1 pulse-type PCB-mounting rotary encoder with integral switch and 20mm shaft (RE1) 4 10kW 9mm vertical PCB-mounting linear potentiometers with 20mm shafts (VR1-VR4) 1 10kW 9mm vertical PCB-mounting log potentiometer with 20mm shafts (VR5) 1 10kW multi-turn trimpot (VR6) 3 miniature SPDT toggle switches with solder tags (S1-S3) 3 2-pin polarised headers, 2.54mm pitch (CON1, CON3, CON4) 1 8×2-pin header, 2.54mm pitch (CON2) 1 PJ-341 3.5mm vertical PCB-mounting jack socket (CON6) [AliExpress] 2 15-pin female headers, 2.54mm pitch (for MOD1) 1 7-pin header, 2.54mm pitch (for MOD2) 1 16-pin header, 2.54mm pitch (for LCD1) 4 5mm-long untapped spacers, 3mm inner diameter 4 M3 × 12mm panhead machine screws and matching nuts 2 M2 or M2.5 × 11mm tapped spacers 4 M2 or M2.5 × 6mm panhead machine screws Semiconductors 1 LMC6482IM dual CMOS-input op amp, SOIC-8 (IC1) 1 LM386M audio amplifier, SOIC-8 (IC2) 2 2N7002 N-channel Mosfets, SOT-23 (Q1, Q2) 1 MBR540 40V 5A axial schottky diode (D1) Capacitors (all SMD M2012/0805 size 50V X7R unless noted) 1 2200μF 16V through-hole electrolytic 1 470μF 16V through-hole electrolytic 1 100μF 6.3V M3216/1206 size 4 10μF 25V X5R/X7R 2 1μF 4 100nF 1 47nF 1 1nF NP0/C0G 1 220pF NP0/C0G Resistors (all SMD M2012/0805 size 1% unless noted) 1 1MW 1 22kW 1 3.3kW 1 0W M3216/1206 size 3 100kW 1 10kW 1 68W M3216/1206 size 3 33kW 2 8.2kW 1 10W 56 Australia's electronics magazine Obtaining the components I have been careful in choosing components that are readily available from many suppliers. Virtually all can be purchased from AliExpress (www. aliexpress.com) at quite low prices. For example, the modules on the control board are an Arduino Nano, 16×2 alphanumeric LCD and Si5351a, which can be bought for a grand total of about $10 plus shipping (a few more dollars). Although some components are classed as ‘obsolete’, they are all still readily available. That includes the BF998 dual gate Mosfets and NE612 ICs. The LMC6482 op amp was chosen as it has a very high input impedance, an adequate GBW (gain bandwidth) of 1.5MHz but, most importantly, it is a rail-to-rail input/output type and can be used with a single supply voltage of up to 16V. While the BF998 is easily obtainable, be careful not to use the BF998R, which has a mirror image pinout (mounting it upside-down is not easy!). The most expensive component is the 9MHz crystal filter module, costing about $25. As I mentioned earlier, it’s cheaper to build your own, but it requires the right equipment and is quite a bit of effort. The other expensive item is the case. The metal case that I have specified is available for about $37. It comes with steel front and back panels. The front panel is replaced by a 1.6mm-thick black circuit board that has all the necessary holes and cutouts. Thus, you only need to drill holes in the back panel for the power, antenna connection, and loudspeaker (if fitted). Next month Silicon Chip siliconchip.com.au Ideal Bridge Rectifiers Additional parts for optional debugging interface 1 3-pin polarised header, 2.54mm pitch (CON5) 2 2N7002 N-channel Mosfets, SOT-23 (Q3, Q4) 3 8.2kW SMD M2012/0805 size 1% resistors 1 4.7kW SMD M2012/0805 size 1% resistor RF board 1 double-sided PCB coded CSE250203, 152 × 50mm 1 9MHz/600Hz crystal filter module (XF1) [AliExpress 1005007201667282] 2 100μH axial moulded inductors (L1, L4) 1 10μH axial moulded inductor (L2) 3 4.7μH axial moulded inductors (L3, L5, L6) 2 Micrometals Amidon T50-6 12.8mm toroidal cores (T1, T2) [Minikits T50-6] 1 80cm length of 0.35mm diameter enamelled copper wire (T1) 1 30cm length of 0.6mm diameter enamelled copper wire (T2) 3 red 5-30pF trimmer capacitors (VC1-VC3) 1 vertical SMA connector, female, standard polarity (CON1) 1 8×2-pin header, 2.54mm pitch (CON2) 1 HFD4/5 or G6K-2F-Y 5V DC coil DIP DPDT signal relay (RLY1) 4 5mm-long untapped spacers, 3mm inner diameter 4 M3 × 10mm tapped spacers 1 M3 × 16mm panhead machine screw and matching nut (for REG1) 8 M3 × 6mm panhead machine screws 4 M2 or M2.5 × 12mm panhead machine screws and hex nuts Semiconductors 2 NE612 oscillator/mixers, SOIC-8 (IC1, IC2) 1 LMC6482IM dual CMOS-input op amp, SOIC-8 (IC3) 1 PCF8574 I2C I/O expander, wide SOIC-16 (IC4) 1 7808 8V 1A linear regulator, TO-220 (REG1) 1 78L05 5V 100mA regulator, SOT-89 (REG2) 6 BFR92P low-noise RF NPN transistors, SOT-23 (Q1-Q6) 1 2N7002 N-channel Mosfet, SOT-23 (Q7) 3 BF998 dual-gate Mosfets, SOT-143 (Q8-Q10) 1 BB910 VHF varicap diode (VD1) 2 1N5711 axial schottky diodes (D2, D3) 1 LL4148 75V 200mA signal diode, SOD-80 (D4) Capacitors (all SMD M2012/0805 size 50V C0G/NP0 unless noted) 2 10μF 25V X5R/X7R 1 4.7μF 25V X7R 2 1μF X7R 12 100nF X7R 1 10nF X7R 12 1nF 1 390pF 1 330pF 1 120pF 4 47pF 2 27pF 1 10pF 1 4.7pF Resistors (all SMD M2012/0805 size 1% unless noted) 4 1MW 1 470kW 1 330kW 3 100kW 1 47kW 8 8.2kW 3 150W 2 100W 1 51W siliconchip.com.au Australia's electronics magazine Choose from six Ideal Diode Bridge Rectifier kits to build: siliconchip. com.au/Shop/?article=16043 28mm spade (SC6850, $30) Compatible with KBPC3504 10A continuous (20A peak), 72V Connectors: 6.3mm spade lugs, 18mm tall IC1 package: MSOP-12 (SMD) Mosfets: TK6R9P08QM,RQ (DPAK) 21mm square pin (SC6851, $30) Compatible with PB1004 10A continuous (20A peak), 72V Connectors: solder pins on a 14mm grid (can be bent to a 13mm grid) IC1 package: MSOP-12 Mosfets: TK6R9P08QM,RQ 5mm pitch SIL (SC6852, $30) Compatible with KBL604 10A continuous (20A peak), 72V Connectors: solder pins at 5mm pitch IC1 package: MSOP-12 Mosfets: TK6R9P08QM,RQ mini SOT-23 (SC6853, $25) Width of W02/W04 2A continuous, 40V Connectors: solder pins 5mm apart at either end IC1 package: MSOP-12 Mosfets: SI2318DS-GE3 (SOT-23) D2PAK standalone (SC6854, $35) 20A continuous, 72V Connectors: 5mm screw terminals at each end IC1 package: MSOP-12 Mosfets: IPB057N06NATMA1 (D2PAK) TO-220 standalone (SC6855, $45) 40A continuous, 72V Connectors: 6.3mm spade lugs, 18mm tall IC1 package: DIP-8 Mosfets: TK5R3E08QM,S1X (TO-220) See our article in the December 2023 issue for more details: siliconchip.au/Article/16043 June 2025  57