There’s a reason why most homebrew transceiver kits and scratch-built projects are monoband and single mode — theres a chance you’ll finish it, or at least, get it working for a while. Building a multiband HF transceiver is a big job, as any homebrewer who has attempted it will tell you. It may take years.

My build of Eamon EI9GQ’s transceiver is no exception. It was started in 2018, the first rush of enthusiasm resulting in a working superhet receiver on 160 to 40m, and boxed up in a custom solid aluminum case. This video shows it off circa 2018.

With a heap of work ahead, and a list of unresolved minor problems, other projects took priority, and the rig ended up spending the next four years in a carton (with all my notes, schematics, assembly and PCB sketches, and unused components). Until recently.

Schematics

It’s my normal practive to include kicad schematics, but in this case the EI9GQ designs are copyright RSGB. So go and buy Eamon’s book from the RSGB Bookshop, or on Amazon, you won’t be disappointed.

Resumption

E-mail discussions with another keen maker (Neville ZL2BNE) about his build of this transceiver gradually tickled my interest to resume. Nev was making good progress, had his rig transceiving, and was working QRP DX. I dug mine out of its carton and fired it up — it had all the appeal and problems that I remembered from 2018. I resolved to kick it along the road for a bit, to see how much interest I could reconstitute. Upon resuming, a number of issues needed addressing, some easy, some repetitive, some more difficult but not impossible. Here are those that I can remember…

Firstly, the 9MHz IF amplifier oscillated with the manual AGC turned up, and was generally unstable. To fix this, I put a 5k5 resistor in parallel with the drain tuned circuit — a well known technique for taming high gain stages. I figured that of the three identical 15 to 20dB gain stages, it made sense to damp down the gain of the first stage. This did the trick, and all three stages exhibited a nice resonance peak using the trimcaps, and overall stability.

The next thing was to calibrate the si5351, a simple job I’d never bothered to do, resulting in the display showing odd and fractional frequencies for the property resolved SSB stations in the 2018 video. If you want to know how this is done, go to your si5351 library’s README file, it is quite easy to do, and once done, lasts for years, or for ever.

Next problem was the LCD backlight. The large font 20×4 LCD I had chosen for this rig was a bit of a novelty back in 2018. I liked that it’s huge characters could be read from half a room away. I used to joke with myself that I’d still be using this rig in my 90s, when all the compact rigs with fiddly little OLED displays were beyond my failing eyesight. And the four lines of 20 characters gave me extra display space for luxuries like a UTC clock and metering. But that big display had an equally oversized backlight which could light up a darkened room, but drew more than half an amp. Although this rig was intended for the shack bench, I was not used to my receivers pulling over an amp.

I decided to multiplex the LED array with a simple PWM LED dimmer, which uses one half of a 4093 quad gate as a variable duty cycle oscillator running at around 70kHz, driving an IRF540 FET switch. This worked a treat and the dim potentiometer was mounted right on the front panel. It controls the backlight from off to 90% on, when it drawn around 500mA.

The next mini-project was an AM detector. I was not interested in full AM transceive capability– I have other homebrew projects for operating VK legal limit AM — but I did want to be able to enjoy decent AM reception using the high quality 6kHz filter in this receiver’s set of three KVG crystal filters. I chose an infinite impedance AM detector, successfully used in a prior AM receiver project.

This mod necessitated making a PCB to overlay the existing product detector and audio preamp, with a plug-in board containing the AM detector, it’s own preamp, a miniature relay to steer the incoming IF signal to either detector, and a shared 4046 quad bilateral analogue switch which selects the product or AM detector’ s output, and additionally does receiver muting and Sidetone routing when in CW transmit mode. This assembly worked well. One hickup — I originally left the BFO powered up in AM mode — and even though there was no direct BFO coupling anywhere, there was more than enough stray coupling to resolve sideband. This was fixed by switching the BFO board’s DC power off in AM mode.

The next task involved coming up with a mechanism to select one of the three KVG (ex TelRad) filters. These high quality 9MHz crystal filters were common on eBay a few years back, and are a feature of this receiver. These filters came in a set of three. In a slight departure from superhet convention, a separate filter is used for USB and LSB, with a 6kHz AM filter in the middle. The BFO runs permanently on 9Mhz. I wanted to have filter selection under software control, so that the correct sideband filter would be selected for the current band. One of the front panel pushbuttons was used for a mode control — pushing it cycles thru the sidebands and an AM setting. A small daughter board was added to the filter assembly containing a PCF8574 demux IC on the I2C bus. A few additional lines of code implemented a simple control function.

The 2018 receiver’s front end board included space fo r an RF amp block, with a pair of miniature relays to bypass it. The original PCB was layed out for a MMIC which probably would have worked fine but in the end I built up one of EI9GQs broadband RF gain blocks using a parallel pair of MPSH10 transistors for around 20dB of gain. The EI9GQ amp was built on an overlay board sized for the available space.

A few minor additions followed. A 41MHz Low Pass Filter was added on the VFO buffer input. On the highest band (28MHz) the VFO is 9MHz higher, on 37MHz. This low pass filter cleans up any VFO harmonics, probably low anyway, but a safety precaution.

The diplexor was added, a balanced tee arrangement, with series and parallel 9MHz tuned circuits arranged to pass through 9MHz energy, but sink all other frequencies into 50 ohm resistors. This ensures proper termination of the receiver mixer and keeps unwanted mixing products, particularly at the image frequency, out of the first IF amplifier.

One of the nore repetitive jobs (which just has to be done!) is making and tuning up the remaining Band Pass Filter modules. Three mire were built, for 17, 15 and 10m, using the EI9GQ design. This gave eight bands, 160, 80, 40, 30, 20, 17, 15 and 10m. My approach to the last three boards was mostly as I’d used in 2018, other than improving the use of right angle 0.1″ header pins inserted through a row of holes drilled through the PCB for mechanical strength.

Next job was a small board mounted on the rear panel next to the two SO239 sockets. This board has the transmit-receive relay, and a second relay that switches between two SO239 for an Antenna A/B switch. Three 7812 regulators supply the three supply rails, 12v always on, 12v receive, 12v transmit. All three unregulated DC supplies are available on headers as well, to avoid heavy current being drawn thru these regulators, such as the transmitter PA and the LCD backlight.

The rear panel has two antenna sockets (for A and B antennas, switched from the front panel), a panel XT60 socket for DC 12V, and a set of 3.5mm and RCA sockets for connections (external muting, paddle, external speaker, and an auxiliary RCA, as yet unassigned). These 3.5mm switched stereo sockets are very useful pieces but unfortunately are not long enough to go through a 3mm aluminum panel. The best solution is to mill out a recess larger than the socket, but that requires a milling machine that I don’t have. So the workaround was to cut a rectangular hole in the 3mm rear panel, and bolt on a 1.2mm aluminum plate to hold the sockets. Its easy to paint and label this small piece.

Preview of the Exciter

To turn the EI9GQ receiver into a transceiver I needed a microphone amplifier, balanced modulator, transmit mixer, T/R switching, a PA stage and LPF set. Back in 2018 I built Eamon’s LM324-based microphone amplifier (a design out of EMRFD) and another hand made diode balanced modulator. All of these mixers use 1N5711 Shottky diodes individually matched to within a millivolt on the diode test setting on my digital multimeter. The transmit mixer is another hand made diode ring mixer, this time an unconventional triple balanced mixer.

The PA chain after the transmit mixer will be MPSH10 x 2, 2N5109 and a pair of RD16HHF1s for about 10 watts 160 to 10 meters. The LPFs will be by Eamon, derived from the original W3NQN designs, each filter on its own plug-in board, relay-switched at either end, filter selection via another PCF8574 on the I2C bus. There will be six LPFs (for 160, 80, 40 and 30, 20 and 17, 15 and 10 meters).

What’s next?

If you’ve made it this far, thanks for reading, feel free to leave a comment, and stand by for the third and final post on this transceiver project, which will address completing and commissioning the SSB and CW transmitter stages.

Weak Signal Propagation Reporter is a global radio propagation monitoring and reporting network comprised of thousands of low power beacons operating on the amateur radio bands. WSPR beacons can be detected from the lowest of Medium Wave frequencies (137kHz) all the way through the HF spectrum (all the bands from 160m to 10m are popular) to the VHF bands, 50 and 144MHz. WSPR receivers decode the tiny beacon packets and upload them to a central database, at WSPRNet.org, where anyone can literally ‘see’ the propagation paths that are currently open.

Equally, you can go back and revisit the radio frequency propagation conditions during any previous time window. Running a WSPR beacon from your home allows you to ‘watch’ the propagation paths open, peak, and close each day under the influences of solar radiation, sunspots, and other ionospheric conditions. Arduinos and a few common accessory boards that can be had for tens of dollars make a beacon accessible to just about any experimenter (with an amateur radio license).

I’m late to the WSPR party. I’ve wanted to try a beacon project for a few years. A while back, I took a copy of the ZachTeck script and experimented with it and a Ublox GPS, but after getting the NMEA strings decoded from the GPS unit at roughly one second intervals, the rest of my code was over-engineered and bloated, and did not fit into the small memory constraints of the Arduino Nano. I put is aside.

Recently, I did a much needed upgrade to my Arduino IDE and libraries. The thought occurred to me that improvements to both IDE and libraries may give me a fighting chance of getting that old WSPR script fitting. When I opened it up, and started to work through it, I saw some obvious ways of reducing memory usage. I had too many String objects (memory-hungry). And my code was writrten to parse each NMEA message string and tokenise it. This allowed me to get to discrete data fields a long way down the messages, like the number of detected satellites. In a simple WSPR beacon, all you really need is the UTC timestamp at the very start of a number of the NMEA messages. I ditched the superfluous stuff and got it uploading, and more to the point, not hanging!

WSPR dataset applications

WSPR is brilliant for teaching you about rare and exotic places that you feel compelled to Google when they turn up on your map in the morning, places like Orlygshafnarvegur (TF4AH, Iceland) or Fuerteventura (EA8BFK on the Canary Islands).

The database of historical propagation across the HF spectrum is widely used by amateur researchers to learn about propagation and has some more serious applications as well. Experimenters have used the data to support ideas or research questions about how symmetrical propagation is at opposite sides of the globe in the same period, and to test antennas. More seriously, a theory was proposed that impressions in the WSPR dataset may indicate the path of the lost flight, Malaysian Airlines flight MH370.

Script

The script is here: https://github.com/prt459/WSPR_GPS_Beacon

Schematic

The schematic is so simple it really doesn’t need a kicad. The Ublox 6M GPS connects to Arduino D2 and D3 for serial data transfer. It also needs GND and +5V. The si5351 breakout board uses I2C and so goes to Arduino A4 (SDA) and A5 (SCK). Connect the si5351’s CLK0 to whatever low power HF amp you like. Mine is from Experimental Methods in RF Design (EMRFD), Fig 12.32, but I could have chosen any number of similar two-transistor stages.

WSPR works on truly tiny power levels. If you connect the bare si5351 clock output to an antenna, you will get decodes! (You should add a Low Pass Filter if this is anything more than a quick test). So use a single 2N3904, or anything with gain, up to a full 5 watt QRP PA with an IRF510 or Mitsubishi RF FET, which is a ‘big gun’ in the WSPR world. Mine uses a 2N3904 and 2N4427 in common emitter feedback configuration, delivering around 10 volts peak to peak into 50 ohms, followed by a W3NQN Low Pass Filter for the band of interest.

Gallery

Acknowledgements

Thanks to Harry from ZachTek for making his code open source. And to Jason Milldrum NT7s for his si5351 and JTEncode libraries.

SPRAT #192

After publication of this project in the RSGBs SPRAT #192 Autumn 2022 a number of builders commented below the post or contacted me with build reports and questions.

First, the 12v DC series decoupling resistor was not labeled in the published schematic, it can be anything between 47 and 100 ohms (see corrected scheatic below).  

Secondly, Ian McCrum did an LTSpice model of the PA which revealed that I omitted  a resistor from the original design in EMRFD (in parallel with the 100n coupling capacitor between the stages) which forms part of the biasing of the PA transistor. If omitted, the PA works, but with reduced power.  Adding the resistor increases the RF power output to around 0.5W, although this will vary depending on drive level, the PA transistor, and the DC supply voltage. Thanks Ian for taking the time to LTSpice this QRP PA.
 
Steve K8SDK got more power out of his PA by increasing the si5351 drive level from 2MA (my value) to 4, 6 or 8MA.  There is a general understanding and some analyses that report the si5351 clock phase noise gets dirtier as you go up in drive level, and for this reason I left it at 2MA.  But with sub 1 watt power levels and a LPF on the output I don’t think it would make any practical difference.

On the question of WSPR beacon power, I don’t think 0.5W is preferable to 0.2W under curent band conditions. With 0.2W the decodes at DX WSPR receivers get marginal and the spread of the band opening can be seen as the grey line moves across the globe. Higher beacon powers can result in saturation which looks like a more rapid ‘lighting-up’ of the remote receivers on the map, and the subtleties of propagation can be lost. This is another way of saying that you should really use a minimum power necessary to get decodes at the far end. However my preference for low power is a personal preference, and some may need slightly higher power to overcome antenna losses.

Several people reporting having to use the set_correction() call with the second parameter (around line 320). I added a comment note immediately above the call in question:

// NOTE: There was a library change to the signature of this method. If you get a compiler error, try this:
// si5351.set_correction(19100, SI5351_PLL_INPUT_XO);

I did it this way to allow for compatibility with subsequent versions of the NT7S si5351 library. But a few people got the compile error and did not trace it back to the offending source line, or perhaps read the comment. I should have just made this change in the repo code to make it foolproof for all comers.

Neil G0ORG emailed with a question about si5351 calibration, his calibration attempt resulted in a value of 149850 for the call to set_correction().  I personally have not seen a calibration offset that large, however, the tiny crystals on the Adafuit and clone breakout boards must vary considerably. Neil reported that this offset pulled his si5351 outside the WSPR passband — we are not sure what went wrong with the usual calibration process for this to happen, bit Neil got his beacon back on the spot by experimentally reducing this large offset value.

Stephen G3ZNG had problems compiling the sketch, and after a bit of investigation discovered he had a previously installed JTEncode library. Upgrading the library fixed the problems and Stephen reported his beacon was working.

Chris G4BMW had the same problem. When he installed JTEncode v1.3.1 and Si5352 v2.1.4 it worked. Chris then discovered that GPS units do not always work inside, and had to move his unit next to a window to give the GPS receiver a sniff of the overhead statellite’s signals.

John G8CHP emailed me a photo of his completed WSPR beacon in a sandwich box. John reports spots in western Canada from his QTH on the east coast of the UK. John used a QRP Labs QLG1 GPS unit, mainly because it was on hand, and a 2N3866 for the PA.

Jonathan G5LUX got his beacon working on a breadboard, and it worked first time.

Aaron K5ATG emailed to discuss his build options, saying that most commercial WSPR beacon products are reasonably pricey for what they are, and my design is made from just a handful of commonly available components, most of which he already had.

Nigel, G4ZA emailed me to say that he had also come up with his own homebrew WSPR beacon, and that Harry’s code (Zachtek) saved him a ton of work too.

Steve K8SDK used three FETs (presumably in the conventional Class E PA configuration as used in many QRP CW PAs) (see comments below post).

And finally, Dave AA7EE has done a superlative job of building his own WSPR beacon using my script, and of course, his blog write-up is amusing, informative and a celebration of the finest amateur radio homebrew spirit. Dave set up his beacon on 10m but had si5351 stability issues which he describes in detail. He solved the drift problems with patience and experimentation, eventually settling on 20m. His post shows remarkable QRPp WSPR results. Thanks Dave for the acknowledgments peppered throughout your post. (See also Dave’s comments below).

A number of builders commented below the post, please read for further discussion. As well, I did receive other emails so if I have missed you please leave a comment below.

The bottom ends of 80, 40 and 20m are not what they used to be. For starters, the busiest part is the digital segment where computers talk to computers – listening to this segment is like eavesdropping on a bunch of dialup bulletin boards having a party in 1983. Then there’s the CW segment. When there are CW signals to listen to, all are frequency stable, chirp and click-free, generated by more computers from deep inside rigs that are more computer than radio. These shining examples of digital CW perfection have traded efficiency and quality… for personality.

In 1979 as a teenager I spent countless hours scanning these frequencies on my FT200, and the sounds from those days are indelibly printed on my memory. The CW segments were a managerie of good, average and bad sounding fists, warbles and tones. There were the regular VK DX men with polished 599 emissions, frequency stable and chirp free, some with curious hand-keyed idioms and flourishes. There were JAs by the dozen, banging out formulaic patterned QSOs, and Eastern Europeans, some on their creaky old ex-WW2 equipment that had to be heard to be believed — sloppy, chirpy, drifty, and gloriously messy CW. And of course the Americans with their kilowatt CW reaching out half way around the globe, some with the self-assurance of a military comms man, armed and dangerous.

It was possible back then to tell where a station was from before hearing the callsign by the combination of signal quality and the operator’s fist. The key to this ability was variety. The CW segment was a rich technical and cultural melting pot of sounds and styles, like the marketplace in some kind of global village populated only by fanatical radio enthusiasts, the ham equivalent of the bar scene from Star Wars 4. In those days, sending a CW CQ gave me butterflies–you could be answered by absolutely anyone, or anything!

The loss of analogue CW struck me again when reading the comments under Peter VK3YE’s video on a two transistor CW transmitter, in which he tries different values of the VXO to PA coupling capacitor. 1nF gives ample drive but pulls the oscillator when keyed. 470pF gives lower drive but less chirp. The CW sound was evocative. Several viewers commented that they would always answer a chirpy CQ because it was likely to be a boat anchor or homebrew rig, something more exotic, perhaps something to discuss during the QSO. The thought of a chirpy CW signal being irresistible to some, a feeble flickering flame to which the morse moths are helplessly drawn, began to form.

Using digital technologies to simulate analogue predecessors for continuity or nostalgia is not unusual. Society has recognised the impact of the digital transition on those of us who are living through it. Melbourne Trams sound a digital facsimile of a 1920s bell, a sound synonymous with the city’s central shopping district for 120 years. Electric vehicles include computer controlled devices to generate a petrol engine’s sound to alert pedestrians, or in the case of the new Dodge, maintain faith with the rusted on fan base.

So, why not do the same for CW? It occurred to me that the power of microcontrollers and multisynth PLL clock generators could easily make this happen. It was a simple coding task to modify my keyer code to make small frequency shifts during dot and dash formation for chitp, and to consistently increment or decrement the oscillator’s base frequency to simulate drift. Time constants were #defined as labels for tuning. With a bit of experimenting, a reasonable approximation of both chirp and drift was found.

These first simulation attempts may be overly simple, as a typical analogue oscillator’s chirp does dot pull the frequency as a linear function of time, but rather, might pull hard, then ease off as the oscillator and subsequent amplifier stages settle down. The corresponding mathematical function is probably a complex polynomial. The same with drift. Most of my analogue oscillators have drifted in one direction, then reached some kind of equilibrium, thereby stabilizing. I’m not a big boat anchor guy, but I wouldn’t mind betting that certain old transmitters have their own chirp and drift signature. How else would you recognize them when you hear them?

I leave more sophisticated simulations of bad CW as an exercise for the reader. Same for the many ways you could use this CW party trick. For example, why not arrange for a switch that disengages the chirp and drift. You could call a chirpy CQ to attract attention, then when you snare someone, turn it off for computer-perfect CW. The options are endless.

If you are brave enough to use this, I ask only three simple things: use digichirp mode sparingly, don’t drift out of the band or segment, and don’t blame me for your bad signal quality reports!

Chirp gallery

Here’s a rogue’s gallery of dubious CW signals. Enjoy!

This AM solid state Class D single band transmitter was assembled over a three year period. Started in 2018, it’s first configuration used a 100 watt push pull RF module published by Drew Diamond VK3XU in Amateur Radio magazine, modulated by a 200 watt linear power amplifier driving a reversed mains transformer, available as a kit from local supplier Jaycar. I built up the RF board, 50 volt power supply (using a stock 300VA toroids mains transformer, no regulator) and proceeded to destroy half a dozen power FETs (STW20NM50) in the RF power stage. Realising I didn’t really know what I was doing, I wisely put it aside.

Not long after I connected with Laurie VK3SJ and Wayne VK3ALK, who coached me along the twisty path of switching technologies for RF, power and modulation– class D H-bridge topologies, 300 watt buck regulators, and Pulse Width Modulators. I quickly learned that switching technology was dramatically smaller, lighter, and more efficient than old school linear approaches. Two homebrew 200 watt transmitters followed, as well as various built and tested AM transmitter modules. For most of this period the 7MHz band had been in the sunspot doldrums, but in 2021 a pulse returned, and so did VK AM stations on 7125kHz. The time had come to finish this project.

The transmitter is comprised of the following modules/PCBs:

• A 300 watt 0..100 V DC linear power supply, consisting of the original toroid mains transformer with an additional hand-threaded 50 VAC winding, a 50A rectifier block, 9,000uF 250v capacitor bank, and a buck regulator to provide continuously variable power from 0 to 100VDC
• Two regulated linear 12V DC 1 amp supplies
• A digital VFO comprised of an Arduino Nano, 16×2 Liquid Crystal Display, various transmit sequencing lines and si5351 triple multisynth PLL
• A Pulse Width Modulator using a crystal clock divided down for 125kHz sampling, IR2110 gate driver, IRFP260s in push pull, followed by a four pole Low Pass Filter with hand-wound RM10 inductors, delivering the modulated DC supply to the RF module
• A Class D H-Bridge PA using a single IXDD614 gate driver and four FETs delivering up to 120 watts carrier
• A 7MHz Low Pass Filter using T106-2 toroids and 1kV glass mica capacitors
• An unbalanced high input impedance microphone amplifier using an audio JFET and a TL071.

Cabinetry, socketry

A transmitter like this involves mains power, and many kilograms of metal and copper. Physical rigidity and having everything bolted down is paramount. I considered repurposing several surplus rack boxes and settled on my favourite, a nice aluminium 3U number, formerly some kind of video switch, that I picked up from Rockby disposals a few years ago. Most of these disposal rack boxes are steel which is difficult for an amateur metalworker like me to drill or file. So if you see an aluminium one like this… grab it. As a bonus, this box included a 240V IEC mains socket and two nice side mounted fans. It also had a front panel bevelled cutout that was cut for a 16×2 LCD, including welded-on mounting risers for the popular 1602 LCDs — perfect!

I left the original labels on the front panel, for provenance, and because they did not annoy me. I cut out the middle centre rectangular hole and backfilled it with 1.5mm aluminium plate, sprayed matt black and labelled. Coincidentally, The new white DecaDry labels I had on hand matched the original labeling nicely. White DecaDry label sheets are almost impossuble to get these days.

Power supply

The power supply consists of a 300VA toroidal mains transformer with 40-0-40 secondary; I wound on another 20V AC winding to get a series total of 100VAC, as well as another 45 turns for 15VAC for 12 and 5VDC regulated supplies. So as not to load down one of these with the in-built fans I added a fourth winding (26 turns for 8VAC), rectified and regulated (via two 7805s), just for the fans. As it all worked out, these fans were not required, due to the 90% efficiency of the modulator and RF board!

I’d never threaded enameled copper wire through a power toroid before. The trick is to use a bobbin as per traditional hand weaving.

Switching regulator

A switchmode voltage regulator (buck converter) regulates the 120V DC HT down to 0 to 100V DC, continuously variable, also performing current limiting and a high SWR cut-out control. The PWM heart of the module is a TL598C PWM controller, with a variable duty cycle pulse train at 120kHz. This drives an IR2110 low side gate driver to a switching FET, that swings the HT across a 120uH inductor and 470uF low-ESR capacitor. A low-value series shunt resistor is monitored by a transistor that turns on at a threshold voltage drop, backing off the PWM controller’s duty cycle. This regulator is identical to that used in my 200W AM transmitter project. For a schematic and PCB (designed later) see Module #2 on this page.

VFO/Controller

I opted for my Arduino Nano/si5351 VFO/Controller. Happily, the original rack chassis had sported a 16×2 LCD and so a perfectly cut and beveled slot and mounting posts were there for the taking. I built a Nano/si5351 and 16×2 LCD to the Raduino circuit on a custom board to fit the front panel cut out and posts. Being that this rig was not a superhet transmitter, I adjusted my script to output a VFO at the signal frequency (7MHz) in transmit (not with the usual IF offset). My script is here. #define SS_VK3SJ_40AM_TX at line 51 to pull in the right code for this project.

The Arduino Nano controls:

• LCD control and data lines
• PTT sensing
• T/R relay control
• Transmitter enable line, which enables the modulator to place DC HT onto the H-bridge PA
• Receiver muting.

I decided to omit any software and hardware for reading RF power (as the PA voltage and current are displayed on front panel meters) and SWR, given the base station antennas always have these inline, and I did not want to over-complicate this build when it had streatched out so long.

Mic amp

The microphone preamp, a 2N5484 FET and TL071, was made up on a small etched board and mounted in its own screened box, including the microphone gain potentiometer, all fitting snugly onto the front panel. This one-off assembly avoided the need for long screened audio cables between the board and front panel. There are no tone controls, this module may be replaced with a more sophisticated mic amp paired with a preferred microphone type.

Pulse Width Modulator

This module takes an HT supply in the range 0..100 volts DC, and line level audio, generates a modulated pulse stream at the chosen Pulse Duration Modulation frequency, and performs power switching into a low pass filter. The result is a modulated DC voltage, suitable for powering an H-Bridge module to generate high quality Amplitude Modulation.

The clock is a 4060 clock/divider that divides an 8MHz crystal down to a 125kHz clock. This clock pulse is transformed into a ramp wave by a Miller Integrator, and fed to one input of an LM311 linear comparator, with line level audio on the other input. The result is an audio modulated pulse stream at 125kHz. This drives both high and low sides of an IR2110 gate driver, then a pair of IRFP260s in push pull, followed by a four pole Low Pass Filter with hand-wound RM10 inductors to effectively convert the PWM into a varying DC voltage (the modulated DC supply) to the RF module.

After assembly and initial testing of the 12v circuitry, the LPF output was connected to a 10 to 16 ohm dummy load. The modulator’s low pass filter has been designed for a load impedance that matches that of the H-Bridge module. As well as following the original designer’s values, I modeled the PWM LPF using SVC Filter Designer from Tonne Software to check the cutoff frequency (28kHz), and input and output impedances.

This module is sized to power and fully modulate up to two of the H-Bridge modules (module #5). This modulator is identical to that used in my 200W AM transmitter project.

RF Power

This module is an H-Bridge class D switch, not a power amplifier in the pure sense as it is non-linear, rather a switching module capable of delivering over 100 watts of power into a 50 ohm load, or other loads with a different output transformer turns/impedance ratios.

For the four FETs you could try Infineon IRFP4019, aimed at class D audio amplifiers, available and priced a few dollars each. I used a better device, the IPP530N15 (was out of stock globally for many months, check supply!). The IPP530N15s have a lower gate capacitance and also a lower drain source on-resistance (Rds) which improves efficiency. The module includes a gate driver (IXDD614, still available), which can be driven with a 5V TTL square wave from a crystal oscilator, synthesiser or PLL (followed by a 74HC-series TTL buffer or equivalent). I have had excellent results wit this driver and FET pairing, with efficiencies of around 90% from 1.8 to 7MHz.

This H-Bridge is identical to that used in my 200W AM transmitter project (which used a pair driving a W8JI RF power combiner). For a schematic and PCB (designed later) see Module #5 on this page.

7MHz Low Pass Filter

The LPF is a conventional W3NQN design. I used T106-6 toroids and 1.2mm enameled copper wire, probably capable of a kilowatt. The capacitors are beautiful 1kV glass mica pieces, quite pricey but essential. I tried what I thought were decent quality 1kV ceramics at one point, and they got worryingly hot! My local supplier for these glass mica pieces is PKLoops, you will need to email them for a current stock list, check out their other products as well.

Comments

Testing and final alignment was done one module at a time. The safest approach with the power switching modules (Regulator, PWM, H-Bridge) is to bring up the 12V section and validate correct operation, then apply HT at around 10V with an appropriate load attached, apply drive, and carefully monitor the gate and drain waveforms.

In general, these circuits have mostly been easy to get going, and stable. A few FETs were blown up in the H-Bridge when operating at around 100 watts or more, mostly due to ragged looking drive waveforms across the gates. It is fairly much essential that you use a current limited DC HT supply — to test and put these H-Bridge modules on air withour current limiting is tempting fate. The Regulator module includes this feature.

With switching circuitry, most of my problems seem to have traced back to improper gate drive waveforms. When the drive looks good, you can turn up the HT to the switching FETs and there should be an almost linear increase in the output waveform amplitude. This is particularly impressive on the H-Bridge, where you see the board deliver 5 watts on a 8 to 10V HT, then up to 120 watts as the DC rail approaches 80 to 90 volts (depending on the load impedance presented by the output transformer’s primary at the frequency of interest).

The H-Bridge’s IXDD614 low side gate driver can draw 600 to 800mA on a 12V DC supply. It has been worth reducing the supply down to 10V, 9V and even 8V whilst monitoring the H-bridge’s output waveform. Most times, the output square-ish wave maintains its shape when the driver’s supply is reduced, which allows the IXDD device to run a lot cooler.

Another protection mechanism for modules permanently installed in an AM transmitter is SWR protection. An SWR bridge and detection unit can be used to detect high SWR and kill the PWM board (thereby dropping the HT to the H-Bridge) in the presence of high SWR. The one I used in another transmitter is module 3 here.

I have not included a complete schematic for this project as is my usual practice, as all of modules are described in other posts and pages. As noted, the H-Bridge, Pulse Width Modulator and 100VDC Regulator are each described on the AM Modules page, with schematic diagrams and some build notes.

Leave a comment below if you want to scratch build any of these. I can share prototype PCB Gerbers if you wish, but these are my own first version prototypes, and I cannot guarantee these are not without minor issues. If you try any of these modules yourself, let me know how it goes.

Acknowledgements

Thanks to Laurie VK3SJ and Wayne VK3ALK for guiding me in understanding and reproducing these modules over several years.

Build notes

Secondary winding 2: 68 turns gave 24VAC (0.353 volts AC/turn); therefore 127 turns should give 45VAC.
Secondary winding 3: 15VAC == 43 turns
Secondary winding 4: 8VAC == 23 turns.

I’ve long been interested in compact and fairly minimal SSB and CW rigs with good performance. I’m not into bells, whistles or menus. Menus are for restaurants! When hiking, walking or bouncing around summits I want to minimise things that are not absolutely necessary, things that can go wrong. Less is more when it comes to a transceiver for portable work.

The first place to reduce unnecessary complexity is your mode. In Australia, a number in the SOTA crowd have slowly adopted CW as the mode of choice . This makes sense for operating QRP with sometimes compromised antennas. The CW trend has been assis ted by increasing and enthusiastic bunch of ZL activators who appear to use CW almost exclusively.

In recent activations it has been common to spot on 20m CW and be rewarded with 3 to 5 ZL chasers, all reliable reports between s3 and s5. Then, a spot for 40m CW should bring forth equal numbers of ZL and VKs. CW exchanges are formulaic, businress-like transactions with 73 GL and dit dits to conclude. No long social obligations concerning handle, rig, wx. A CW activation is efficient and fast. You can bag 7 or 8 chasers in minutes. Reducing your qualifying time let’s you keep moving, or, gives you more time to enjoy the mountain top experience.

There’s another noteworthy feature of CW activations. They nearly always use the same frequency. 7032, 14062 kHz. And on a SOTA activation, the standard procedure is that you spot with one of the apps, call CQ SOTA, and the chasers line up to work you. You hardly ever touch the dial. In fact, you hardly even need a tuning dial !

That got me thinking. How minimal could a CW multiband rig get? In a dedicated SOTA CW rig, do you really need to be able to tune around the band, or could you get by with fixed ‘channels’?

Concept

The concept for this project is that of a CW ‘appliance ‘, a device that you pull out of your pocket, plug in the antenna and paddle, choose your channel (aka band) and hit the keyer button to send CQ and get the activation started. The appliance would need to cover at least 40 and 20m, the two VK/ZK SOTA CW watering holes, and one or two additional higher HF bands, where short antennas offer interesting variety as Cycle 25 rises.

Five watts should be plenty. An inbuilt top-facing speaker with a headphone jack will suit all listening situations. Small and light goes without saying, as does the option to operate on an external 3S or 4S LiPo pack, possibly even strapping the battery to the rig.

It will need to be physically sturdy without being too heavy — 3 to 400 grams seems like a good target weight.

Choices

A simple, dedicated CW rig shouldn’t require a complicated receiver. A single conversion superhet is in order. I studied various designs by Steve Weber KD1JV, particularly his MTR5B and SodaPop. The Mountain Topper range are very well regarded, even romanticized by some owners. The MTR5B is a dual SA612 receiver with 4.915kHz IF. The more recent SodaPop uses a pair of JFETs in each mixer, but is otherwise similar. I also looked at the receiver in the Elecraft K1, also an SA612 design.

I’m a fan of the SA612, with a decent bandpass filter and a resonant antenna ahead, proper impedance matching and a bit of extra IF gain downstream. I have not had any problems with these receivers with basic but decent antennas on mountains or at home. What some northern hemisphere hams do not realise is that the bands in VK and ZL are more or less empty when compared to what we see on USA and Euro SDRs. Pull up a session on 80 or 40 anytime on my local receiver and see what I mean. Also, VK hams are capped at 400 watts which eliminates the ‘kilowatt around the corner’ problem we hear talked about. And our lower population density limits the Broadcast breakthrough suffered by some who live in densely populated areas. So we are lucky here, living in a region with a low density of hams, although it has its drawbacks as well.

I also looked at receivers using diode ring mixers such as the Bitx, but these receivers require higher oscillator injection levels that necessitate non trivial buffering and level setting over the rig’s intended frequency range. Gilbert Cell mixers have useful conversion gain and avoid this complexity to some degree.

I also looked at the QCX, which uses a higher performance quadrature detector. It’s an option in a compact and portable analogue receiver if you use Hans’ polyphase kit to do the audio phase shifting for a single signal audio output. Also the mixer requires a 4x VFO as input to the usual 74AC74 divider, not really a problem with an si5351 but I’ve not tried it before.

The best path to realising one of these would be to buy Hans’ High Performance Receiver and Polyphase plugin kits. The resulting assembly is only 80mm x 50mm, so with a VFO (no BFO necessary because it’s base-band) there are some good options for a partially scratch built multiband version of the QCX. Interesting. I’ll leave that concept for another time.

Schematics

Page 1 is the transceiver core:

Page 2 is the Arduino Nano, si5351 and controls:

Construction

Construction methods followed my established combination of stacked (hand-drawn and etched) PCBs housed in an aluminium sheet and angle case. The transceiver was designed as two self contained modules, the VFO/BFO and Controller (Arduino Nano and si5351), and a second housing receiver, BPFs, transmitter and LPFs.

VFO/BFO/Controller

This module was designed and built first. It followed the common pattern of an Arduino Nano, si5351 breakout board, 78-series voltage regulators, a discrete clock buffer on the CW clock (CLK0), sidetone filtering and some switching components. The module consists of two PCBs — a single sided hand-made base board is bolted flat against the aluminium base plate with side controls mounted directly on the board. Front panel controls are mounted against a double-sided hidden front panel PCB with perpendicular bracing pieces. Two 8-pin 0.1 inch DIL header sockets at either end support the daughterboard on top which houses the Nano and some logic.

• 

• 

• 

• 

• 

A vertical line of three miniature pushbuttons at the left hand end of the front panel implements the transceiver’s frequency control. The middle button is the channel button — push it, and you move to the next channel. A channel is a semi-fixed frequency in one of the four supported bands — 40, 30, 20 and 17m. Each of the six channels has its own LED on the front panel. The mapping of a channel to a band and frequency is fixed in the firmware (but is easy to change).

The upper and lower buttons ‘bump’ the channel (VFO) frequency up or down by 100Hz. So to move 1kHz from the default channel frequency, you need to pump one of these buttons ten times, counting as you go. After a few seconds, the current frequency is written to EEPROM and will persist over a power-down.

So, if you have ‘tuned’ the rig away from a channel (such as 7032kHz, the 40m SOTA CW calling frequency) how do you get it back? Easy! You hold down the channel button for a second and it reverts to the hard-coded frequency. If you wish to change any of the channel frequencies, you edit the Arduino script and upload it to the Nano, whose USB is accessible through a slot cut into the transceiver’s left side panel.

Receiver and Transmitter module

This module uses an upper and lower PCB pair, with transmitter on the bottom and receiver on top. In a departure from my usual T/R relay to switch antenna and DC power, both are done electronically. In fact the receiver is permanently on, so there is no need for a separate +12 volts (receive) line. The RF switching arrangement is copied straight from Steve Weber’s MTR5b, and is almost the same as is used in the QRPLabs QCX.

Receiver

The receiver is a standard superhet with SA612 Gilbert Cell receive mixer and product detector and a 5 pole homebrew crystal filter. The design is almost identical to VK2DOB’s MST3, and KD1JV’s MTR5B (which doesn’t have the additional IF amp stage). I built my crystal filter at 4MHz but only because I didnt have any 4.915MHz low profile crystals in the junk box. My filter exhibits steep skirts and a bandwidth of about 300 Hz. Just about right for CW.

I added an additional gain stage after the mixer which makes a difference to receiver liveliness, remebering that the 5 pole narrow crystal filter is a point of significant attenuation.

Band Pass Filters

In previous projects I have strictly adhered to tight bandpass filters, one per band, and always using hand wound inductors on T37 or T50 toroids. Favourite filter designs have been those of Eamon EI9GQ from RSGB RadCom, and Diz W8DIZ of kitsandparts.com, both easily reproduced filters. This time I tried something different — a different filter design using electronic switching and surface mount inductors.

The filters are taken from the hardware portion of the RS-HFIQ project, a modern baseband SDR. They are much broader in bandwidth than I’ve used in the past, as the sweeps show. This means that the Gilbert Cell SA612 receiver mixer will be exposed to more out of band RF energy coming down the antenna, which could result in overload. Let’s see.

The filters are electronically switched using a 2N7002 FET between the filter earthy end and real ground. Pin diodes (from Minikits) do the switching. This saves a relay and relay driver.

• 

• 

• 

• 

CW transmitter

The transmitter portion reproduces those of Steve Weber’s MTR-5B and SodaPop as well as Hans Summers’ QCX, and uses three BS170 JFETs in parallel driven by a high speed logic gate to deliver up to 5 watts of RF to the Low Pass Filters. Once the drive level was padded to ensure at least 4 volts was hitting the BS170 gates, it worked as expected.

This is a Class E switching configuration, so unlike a more conventional Class A or AB RF power stage there is no bias, meaning it draws no current at all between dots and dashes, and is around 90% efficient.

On the bench the transmitter was drawing 300mA at 14V for 3 watts of RF (remember the Digital VFO and Controller draw 80mA). Observant readers may notice that the driver logic gate is a 74HC00 NAND, not the usual 74HC02 NOR, only because the NAND gates were on hand. No drive problems have been observed as a result of this substitution.

Low Pass Filters

Continuing the spirit of simplicity and to save space, two LPFs are used to cover the four bands (40 and 30m, 20 and 17m), a common technique in QRP rigs. These are 7 element W3NQN filters. Remember that a resonant antenna plays a vital part in the transmitting system’s overall spectral purity.

Solid state TR switching

In another break from my past practice of using miniature Telecom relays for transmit/receive switching, the series JFET used in KD1JV’s designs was tried. An almost identical arrangement is used in the QCX. No appreciable received signal loss was experienced, and the JFET appears to be an effective blocker for RF power from the transmitter at the 5 watt level.

Receiver muting

Despite using a solid state analogue switch (TS5A3157) in series with the audio signal path, getting a silent CW break-in switch (from receive to transmit then back again) proved to be a major headache. On my PCB the TS5A3157 switch was inserted between the two op amp audio stages. This resulted in an annoying click going both into and out of transmit. No amount of bypassing or fiddling with signal levels made much difference.

I checked for DC levels around the input of the TDA2003 IC and found a DC offset of about 1.4V on pin 1 (input), which is always blocked with a series 2uF capacitor. Nothing unusual there. I wondered if this series 2uF electrolytic was charging or discharging, bur reducing it to 0.1uF made no difference.

Next, I build a small vertical board with a second 3157 switch, right next to the TDA2003, with just a series 100n capacitor from its output to the volume control, which itself was isolated from DC with 100n capacitors. That made no difference.

It is strange how you can get fixated on things like this. The rig was useable as it was, with what some might call an acceptable click on change-over. But I wanted a noiseless changeover, and the quest turned into a series of experimentation and debugging sessions that stretched far beyond what I’d expected.

I now regard noiseless T/R switching in a CW rig with an audio power stage capable of driving a loudspeaker to be a non-trivial problem. As I was studying the KD1JV (MTR, SodaPop) and G0UPL (QRPLabs/QCX) designs I realised that they both support headphones only, not loudspeakers. Could it be that lower volumes made this problem less pronounced?

The problem is as follows. You want a noiseless transition from CW receive to CW transmit and back again. It has to happen quickly to make even ‘semi-break-in’ work. But in transmit mode, you want the sidetone to come through in your speaker. So you cannot disable or mute the audio power amplifier stage, otherwise you lose the sidetone. As well, you want to have the sidetone come via the volume control, so that turning the volume up or down affects both receiver audio and sidetone.

I reluctantly decided to ditch the solid state audio switch (which was making an annoying click on both transitions) and replace it with a relay at the input of the volume control and audio power amplifier, switching the audio source between receiver noise and sidetone. Mercifully, this resulted in a silent Rx to Tx transition, but, when the transmitter dropped out, a nasty click! This was particularly annoying as I’ve successfully made noiseless TR switching with TDA2003s and a relay in two other rigs.

Finally I added a second relay to mute the audio power amp for a short period (after the last character had been sent and just as the rig reverted from transmit to receive). A second digital control line coming from the Arduino, and some orchestrated timing in software was needed.

Eventually, I achieved silent T/R switching, and it is a pleasure to use. How to mute the audio amplifier’s transmit to receive click more elegantly? If the audio IC I’d chosen had a mute pin, that would suffice. But the TDA2003 is an old car radio audio amplifier and has no mute. So I took the brute force action. Normally closed, this relay opens for a few hundred milliseconds and silences the click from the power amp. This arrangement is shown above for all to see.

Case and finishing

The case measures 70mm wide, 132mm deep and 32mm high, and is made from hand worked aluminum angle and 1.2mm sheet for the base. A top cover is from 1mm sheet.

The front panel is finished with all purpose metal primer, three enamel coats (colour is called ‘aluminum ‘ and is an appealing silver-grey). Lettering is rub-on DecaDry. Two coats of clear satin enamel spray seal the panel. The side panel is labels applied direct to the aluminium angle piece, with a satin clear top coats.

• 

• 

On the Summits

After a few weeks of bench testing it was time to try the little rig in the field. Two nearby SOTA peaks, Mt Vinegar VK3/VC-005 and Mt Gordon VK3/VN-027 in the Yarra Ranges acted as a proving ground and offered 10 activator points in total. Both are miles from residential areas and offer the chance to play radio in a noise-free environment.

After a 90 minute drive followed by a 90 minute (at times strenuous) walk from Acheron Way up four wheel drive tracks to the summit, we were on-air on Mt Vinegar at around 1.25PM local time. Antenna was a linked dipole for 20 and 40m on a 9m pole. Starting on 20m, two of the regular New Zealand chasers called in, ZL1BYZ and ZL1TM, weak but workable, 539 reports coming back. VK2IO provided a third 20m QSO. Moving to 40m, four chasers (VK2IO again, VK2WP, VK5IS, and VK5HAA) called in with reports ranging from 419 to 559.

The rig performed well as expected, although the audio output level (or receiver gain) on 20m seemed a touch low.

From here we moved on to Mt Gordon VK3/VN-027 on the outskirts of Marysville, a drive-up four pointer with a comms and fire watch tower, and a great view of the Cathedral Ranges to the north. 20m yielded just the one QSO with ZL1BYZ (thanks John, you are amazingly reliable!). A QSY down to 40m caught VK2GAZ, VK5HAA, VK2LI, ZL3MR, and VK2IO again, with all R5 reports ranging in strength from 2 to 5. Now, later in the afternoon (we finished around 5PM), both 20 and 40m were more lively and the receiver correspondingly louder.

• 

• 

Improvements

Back on the bench a few fixes and improvements were made. The hole on the side panel was widened to stop the CW keyer message button sticking. The single 2N3904 IF amplifier stage, originally using a resistive collector load and a series coupling capacitor into the cyrstal filter, got a 10 turn FT37-43 bifilar transformer on its output which improved its overall gain by some dB. A number of minor firmware changes were made. The top plate was cut and this greatly improved the speaker volume. Never judge an un-baffled loud-speaker.

Size and weight

Comparisons with the tiny and much loved Mountain Toppers are enlightening. The MTR-5b (the inspiration for SP-X) apparently weighs 6.4Oz or 181 grams. That’s light! I believe this is sans batteries. SP-X weight 332 grams, a lot more. About 27g is attributable to the speaker which the MTF-5b doesn’t have.

The MTR-5b is 4.27″L x 3.2″W x 1.34″T. I make that 10.8 x 8.2 x 3.4 cm or 301 cm3. SP-X is 14 x 7 x 3.2 or 313cm3 — about the same volume.

To get the weight (and size) down further, you’d need to ditch the homebrewer/maker-parts (the Arduino Nano and si5351 breakout) and use exlusively surface mount components on a purpose-designed and fabricated PCB. This represents a big step from a prototype like SP-X to a product that can be produced and sold in a run. There are examples all over the crowd funding sites. It’s the logical next step but it requires different skills and it’s not really my game. Kudos to Steve Webber for his achievement!

Closing comments

SP-X, like all my projects, are prototypes without complete revisions and iterations to follow. I’ll never go back and build a second version of SP-X with the workarounds and mistakes corrected. As a consequence I’ll live with a few re-worked stages (such as the receiver muting fix). A more considered solution to the muting problem might involve a comprehensive end to end design of the audio stages from detector to loudspeaker. Perhaps you’d have two digitally controlled potentiometers on the I2C bus to act as faders between the two audio sources and an audio power IC with muting that you knew could be trusted to switch silently. Maybe there is a simpler way of doing this in a rig with a 5 watt audio stage. Feel free to let me know in a comment!

I’m very happy with how this little rig turned out. It is compact, light, useable, simple, and as versatile a portable QRP CW station as I’ll ever need. I’ll be happy to trust it to get me the four QSOs on any VK3 activation in the future. It simplifies and lightens the rest of my load, particularly the battery which is half the weight of its predecessor. If I built it again I’d fix the receiver muting and probably try to accommodate a LPF for each band. Other than that, I’d build it as it is.

And channelised SOTA CW is a breeze — who needs a tuning knob and display anyway?!