After completing a 6-band SSB/CW QRP transceiver (Summit Prowler IV) I found myself thinking about a more compact QRP SSB/CW rig for SOTA, with two of the main day-time SOTA HF bands (40/20m). The design driver this time was to try a different ‘form factor’ — I wanted a rig with a narrow and long case, such that it would easily slide inside a backpack, and on a summit sit vertically against a rock or be hand held. All my SOTA rigs so far use both front and back panels for connections and controls, so they need to sit level on a horizontal surface. As most rocks or tree stumps are low, you can’t easily read the display. Some designers get around this by putting the display on the top of the box, a sensible adaptation but one that makes the rig look like a flounder. Because I spend a lot of time building and using these radios at home on the shack bench as well as on a summit, I wanted a design for use in both situations.
Backpack convenience was not the only motivation. Another came from being with Glenn VK3YY and others on Mt Selwyn (VK3/VE-049) in February 2018 during the annual VK3 SOTA weekend. Glenn pulled out his Elecraft KX2 with a short center loaded whip and counterpoise lying across the grass. In the first few minutes, while others were still looking for squid pole supports, Glenn was on 14.310MHz SSB working ZL3CC, ZL1BYZ and ZL1WA. The audio was clear and crisp, the band almost silent. Reports 59 given / 55 received. Standing there on a pristine 1,450m summit, holding what was essentially a walkie talkie and talking to New Zealand, made for a memorable SOTA moment.
Glenn VK3YE with KX2 on Mt Selwyn (VK3/VE-049), February 2018.
So I started considering the classic ‘walkie talkie’ form, such that the main controls and display are on the (small) front panel with the antenna socket (because the rear panel would need to be sealed, able to be planted on a rock or bare patch of dirt). Other controls, such that there are, would have to be built-in (small speaker and microphone), offloaded to the side of the case (Push To Talk, keyer push-buttons), or dropped altogether (keyer speed control, microphone gain). For inspiration I looked to the Elecraft KX2 and the Youkits TF2B.
A selection of walkie-talkies from Ian VK3JQ’s collection. Photo courtesy of Ian and David VK3KR.
The small front panel (top) of the walkie-talkie case forms an important constraint — the LCD panel would need to be small and the SO239 socket would have to give way to SMA. It seemed like an interesting design and construction challenge. I started googling and sketching.
Space inside the case would be at a premium so I decided on a bilateral design where the IF and mixers at either end are shared between receive and transmit. I looked at the classic BiTX40 but settled on Andy Hunter G6LBQ’s variant, which uses small (T37-X) LPF toroids, an MC1350 for adequate receiver IF gain, and ADE-1 double balanced mixers, saving the trouble and space of hand-making these,. As well, it’s a well-proven multi-band design.
I would build another Arduino Nano/si5351 VFO/BFO/Controller, wired up as per VU2ESE’s Raduino module, and running my Summit Prowler script so that all my recent homebrew rigs share similar controls and features. In a multiband rig, every additional band adds complexity and takes up space, so I decided on 2 bands, 40m and 20m, primarily because only one double throw relay is needed at each switching point, and with the future (solar cycle 25) in mind, it seemed a good choice.
On the brash assumption that I would make the rest of the transceiver fit come what may, I started the project by making the walkie-talkie styled case. I made the simplifying assumption that battery would be outside the case — a LiPO pack could easily be strapped to the back, a pragmatic trade-off that is taken by many portable operators who find the internal battery option in most commercial rigs an unacceptable compromise.
Some of the small parts I found for this build are best described as ‘cute’. I found a cute 8×2 LCD (Rockby) and a small mechanical encoder for tuning (Altronics), a very cute 8 ohm 1 watt full range speaker that I’ve used previously in my Wilderness SST rig (Jaycar). I made up the chassis with a thick lower plate and front, rear and sides in 40x15mm angle aluminium, with 10×10 brackets set into the side for a drop-in 1.5mm aluminium lid. This all took a while, working just with hand tools, as I wanted it to look reasonable and fit snugly. I used M2.5 bolts, nuts and washers (here) so that any part can be removed for access or modification.
A small rig needs some small parts. I used this 8 ohm full range 2 watt loudspeaker, a product of the consumer electronics age, from Jaycar. It has a heavy cone that needs a decent amount of drive, an LM386 would be woefully inadequate. I used my favorite car radio audio IC, the TDA2003.
I built an Arduino Nano, Adafruit si5351 breakout board, some voltage regulators and clock buffers into a board cut to fit tightly into the top third of the chassis. The circuit and pin assignments are as per VU2ESE’s Raduino. All my recent Nano/si5351 VFOs follow this pattern, which means I can download my universal controller script and it runs without modification. This time I added an alternate code path to handle the smaller 8×2 LCD, selected by Arduino pre-processor primitives #define, #ifdef and #ifndef.
This script supports multiple VFOs, automatic band switching, PTT and break-in TR switching, S-meter, SWR meter and a memory keyer. All of these features ‘just work’ if you arrange the appropriate signals or connections to the Arduino pins. The script is here if you want to use or repurpose it, as at least 3 homebrewers have done.
A +7dBm mixer like the ADE-1s needs a buffer/amplifier after the si5351 clock output for isolation and a bit of gain. I used this circuit on each of the three clocks, delivering about 1 volt P-P (5 dBm) at the mixer LO port.
Choice of IF
In a departure from 12MHz IFs I’ve been using of late I chose a 9MHz, mainly because I had a Minikits order ready to go and it was too easy to click ‘Add to Basket’ on a matched set of eight 9MHz crystals, six of which could go straight onto the receiver/exciter board with no further screening or matching. The IF in the Arduino script is selected by #defines so changing the IF generates no extra work.
I built the receiver/exciter, a bi-lateral BiTx arrangement as per the G6LBQ design, with the exception of a TDA2003 in place of the LM386. Andy has made a number of improvements to the basic BitX, including a MPSH10 VHF amp transistor replacing the 2N3904 at the receiver first RF stage, a MC1350 IF amp, a six crystal filter, LM324 audio derived AGC and ADE-1 double balanced mixers, and some T/R switching/muting transistors.
I’d resolved to not use IRF510s because of reduced power output on 14MHz. The G6LBQ PA uses Motorola RD16HHF1 FETs which do not suffer from this problem, don’t need more than 13v for full output, and don’t require electrical isolation of the tab. The G6LBQ PA uses three of these devices, one as an over-spec driver and two in push-pull for a conservative 10 watts — way too much power in a small chassis and an impractical current draw (bias alone draws 0.75 to 1 amp). So I decided on a single RD16HHF1 for about 5 watts on both bands using a circuit from Glenn VK3PE’s website.
Receiver debugging and testing
The receiver worked first time, once I diagnosed that I’d put the MPSH10 in the wrong way — it has an unconventional pinout arrangement. The bandpass filters resonated as expected. The MC1350 was distorting until I got the tuned circuit on its output resonant, but now the receiver sounds clean and stable, and as the video illustrates, the receiver is more than sensitive enough. The small internal speaker is adequate but insensitive, even with the TDA2003 driving it. On the bench, an external communications speaker makes for armchair listening.
Pressing PTT resulted in a sharp chirp of audio down the transmitter, kicking the power/SWR meter up to full power. I posted on Groups.io/G6LBQ and Andy confirmed that the squeal didn’t occur in the original design. So it must be coming from the big electrolytics in the TDA2003 audio amp stage, which is powered from the 12 volt (Rx) line. I tried leaving 12v on the TDA2003 so the electros stay powered up between Rx and Tx; leaving switched 12v (Rx) on the 2N3904 preamp, and leaving the 2N3904 ‘PTT click silencer’ on +12v (Tx). This cured it.
Transmitter debugging and testing
The transmitter took a bit more effort, and is still a work in progress. Instability crept in as the PA bias was wound up. First step was to debug the PA and driver stages. Removing the exciter drive from the PA input and linking the keyed (buffered) si5351 clock 1 straight onto the base of the 2N5109 driver (via a coupling capacitor) gave a clean few watts of CW output. With the rig on the bench, I worked Gerard VK2IO in a National Park a few hours out of Sydney, for a 579 report. First QSO!
Reconnecting the exciter output to the PA, the signal was choked with RF feedback. With reduced gain I again worked Gerard from his second park of the day, this time on SSB — he heard me, and gave me 4×7 with RF feedback. QSO #2! Premature perhaps, but a little encouragement along the way.
The oscilloscope showed a dirty signal. I tried various supply decoupling methods, including ferites and more bypass caps, without success. I then took the power off all exciter stages and reapplied it one stage at a time working back from the PA. The last exciter gain stage (2N3904 mixer post-amplifier) was stable. But the next stage, between the crystal filter and the mixer, howled.
In all, the transmitter required lots of screening and decoupling. After re-making several of the transmitter IF gain stages to achieve better layout, I discovered the speaker-mic curly chord cable was not screened. Replacing this with a 4 conductor shielded audio cable (not curly, separate conductors for mic, speaker, PTT switch and earth), I managed to tame the transmitter sufficiently to get some usable RF power out on both SSB and CW. I got much better with the brass modelers stock and tin snips. The biggest noise reduction in the receiver was achieved by shielding the 6 pole crystal filter, a must in future receiver projects.
On a summit
I determined the point of PA biasing where the transmitter went unstable on both modes. At this point the transmitter delivered about 25V p-p (32dBm or 1.6 watts). Around 6dB or one S-unit down from 5 watts. Two watts if I talk loud. This little rig had burned quite a few hours on the bench and I was eager to give it a try from a summit.
On Saturday 6th October I slipped away to SOTA summit Mt Donna Buang (VK3/VC-002), 1259m (a popular drive-up summit), arriving in the late afternoon. Unbeknown to me (because I do not always check SOTAWatch Alerts as I should) it had been visited on the same UTC day by VK3s AFW and PF, both no doubt keen to pick up late winter bonuses. I got there close to sunset, strung a 40/20m trap vertical, and encountered the Oceania DX contest in full swing. My flea power was not enough, and despite spotting and calling, no QSOs. I did hear a VK calling me on CW but I was not quick enough on the push-button keyer to engage. VK and ZL signals were not strong and, very soon, the light dropped. The best part of this activation was the sounds of birds calling on the heavily wooded summit at dusk.
A week later, I tackled Sugarloaf Peak near Buxton (VK3/VN-011) and then Mt Strickland (VK3/VN-030). This time in the middle of a perfect sunny Spring day, aided by a resonant full sized link-dipole, I managed to work VK2, VK4 and VK5 for a qualification. The action is caught on the video above. At Mt Strickland, a few hours later, the highlight was working ZL1TM, mid-afternoon, with 2 watts CW, and a 539 report (1,626 miles, 2,618km). It’s amazing how the grey-line can open up propagation directionally. Two activations. Summit Prowler V christened.
This little package is much smaller than anything else I’ve built and has the potential to be the first choice for future activations, particularly when 20m improves. It provides for the ‘walkie talkie’ operating experience, and the built-in keying (via push-buttons), keyer and microphone make it simple to get on air quickly. The receiver is sensitive and pleasant to listen to. The transmitter needs more work to further reduce RF feedback in the IF chain, so that the full 5 watts can be delivered to the antenna on both bands.
Post Build Reflection
This was my first all-surface mount rig, using 1208 resistors, capacitors, MMBT3904 s and SoiC ICs, and I won’t be reverting to wired components any time soon. Once you make the adjustment, SM components are easier, faster and (unexpectedly) a more satisfying way to build RF circuits. And considerably cheaper. A good illuminated magnifier and workspace, tweezers, jeweler’s screwdriver set, a fine iron tip and solder, and a bit of patience, are essential.
Building a two-band transceiver into a tightly packed chassis presents challenges, which necessitated quite a few hours of debugging and diagnosing in the transmitter. Chasing down RF feedback is a chore and it was tempting to toss it aside. But persistence got me to a stable 2 watts of SSB and CW RF and I again learned how important basic RF techniques are, even at 7MHz.
Higher density amplifies the need to prototype. If I were to do a second prototype I would build RF-tight compartments into the chassis for maximum stability. This is where the physical and electronic design need to come together. It’s more difficult but not impossible to do this now. Some space could also be saved by abandoning the consumer Nano and designing a custom front panel board with the MCU, LCD, controls, si5351, buffers, power regulators and headers. But that’s a project in its own right.
Summary of changes to the G6LBQ design
My build employs the following changes to Andy’s original Bitx design:
- Arduino Nano/LCD/si5351 with a single broadbanded 2N2222 buffer/amp for VFO (CLK#0), transmit Carrier Oscillator for CW (CLK#1) and BFO (CLK#2)
- TDA2003 replaces the LM386
- IF is 9MHz, VFO high-side injection (fVFO = fsig + IF)
- Crystal filter capacitors are 68p and 82p, couplings are the original 100p, the filter is a little bassy, I don’t mind the sound but smaller values would widen the passband
- PA uses one RD16HHF1 in a standard QRP 5W PA circuit
- For T/R switching, one digital input on the Nano senses the PTT switch going low, which then raises another digital output to pull in the two T/R relays on the PA board; one relay has a set of contacts that switch 12v between the low level stages of the Rx and Tx
- The script provides 4 VFOs, automatically pulling in the band switching relays for 20m
- CW is done by enabling/disabling the si5351 CLK#1 to form the dots and dashes, which also handles semi break-in and shifting the transmit frequency by 700 Hz; this keyed clock is injected into the PA (pre-driver).
Otherwise it is exactly as per the G6LBQ MkII circuit diagram.
Thanks to Andy G6LBQ for designing and supporting the G6LBQ Bitx. Andy is currently releasing selected modules for his follow-up analog/digital all-band HF transceiver, the Irwell. Details available in the G6LBQ Groups.io community.
I make my own, ask the community if you want PCB patterns.
Can be found here, free to be used, repurposed, cannibalised, as several builders have done. Please let me know if you use it in a project.
There were no unusual components used in this build. All the usual sources, see Favourites for my preferred suppliers, or if still stuck, ask via a comment.
Peter DK7IH is a master homebrewer (35 years) who works with available components to create compact QRP transceivers. Here’s his 20m walkie talkie, a real exemplar of good looking radio craftsmanship.
Check out his ultra compact shirt pocket rigs. Remarkable. And leave a comment on his blog!
From Melbourne, Google will navigate you to all three summits.
Mt Donna Buang is a popular mountain spot for Melburnians, a sealed road all the way to the top, a high lookout tower for excellent views, and a constant stream of day trippers. Has been popular with VK3s for decades, due to the easy access, there being no commercial RF on the summit, the public access tower, and line of sight back across the city.
Sugarloaf Peak is 1.5 hours drive north east of Melbourne, at the southern end of Cathedral Ranges National Park. It is well known and I always encounter other hikers at the top. These bloggers describe it as ‘one of the toughest day walks close to Melbourne’. The climb to the SOTA summit is about 45 minutes and steep. There are two tracks, Canyon Track (hard) and Wells Cave Track (very hard). I took the former (the latter would be very difficult carrying a SOTA pack and pole). The difficulty is due to how much scrambling over granite rock face you are happy to do. Even the Canyon Track has you scrambling on all fours, pushing the squid pole ahead of you. The secret is to keep your center of gravity low, and concentrate on foot and hand holds. I wouldn’t attempt it in even the lightest rain or impending showers. All that said, it’s not that hard for an active person with two legs and two arms on a fine day, and lots of fun! Recommended!
Mt Strickland is a dirt road drive-up off Acheron Way, access details here. Tony VK3CAT shows that about 400m down MMBW Track you cross into Yarra Ranges National Park, so you could probably set up there and offer both Parks and SOTA points to chasers.