Scratch-built 8-band HF/6m transceiver (EI9GQ) – Part 1 Receiver

Eamon EI9GQ’s book (‘Building a Transceiver’, RSGB 2018) started me down the path of another modular transceiver project. For this build I wanted to continue working with surface mount but without the compulsion to pack it all in tight. More space and the freedom to replace a module later. It would be a Shack Sloth rig (a base station), not a Summit Prowler, so the space, weight and power budget shackles fell off from the start.

Eamon’s design is a conventional multi-band single conversion superhet with AD9951 DDS VFO and crystal BFOs. I would use a Raduino (Nano and si5351), perhaps not as clean but easy for me to reproduce. Eamon’s design uses hand-made diode ring mixers operating at +7dBm LO injection (or higher) and a healthy distribution of loafing gain and buffer stages. The design uses a number of familiar approaches and techniques which would not stress my pragmatic home-brewing style. And I would use the set of three Telrad 9MHz KVG 8-pole crystal filters which have been burning a hole in the junk box ever since they landed from Israel six months ago.

Initial build

Construction was done on my usual hand-drawn and etched printed circuit boards, ‘muppet’ style, copper-up, with 0.1 inch headers for all connections. As the cold Melbourne winter weeks rolled by, modules tripped off the work bench — 4-stage band-pass filters for 160, 80 and 40 (initially), each with its own pair of switching relays and a 2N7000 relay controller on-board, followed by a BPF rack (8 sockets, one per band in the 160 to 6m range), with PCF8574 de-multiplexer for I2C filter selection.

Four section bandpass filter (160m) by EI9GQ, with switching relays at either end and a 2N7000 switch; trim caps mounted on PCB ‘shelf’ for accessibility.

Rack for 8 BPF cards, with a PCF8574 demux card for band selection via I2C.

Next, a hand made receiver diode balanced mixer and a post-mixer amplifier with push-pull 2N5109s (I used 2N2219As), which I trialed as a receiver pre-amp with some success.

Receiver mixer board, from left, space reserved for a preamp (relay switched), then hand made diode ring mixer, diplexer, 10dB post mixer gain stage with 2 x 2N2219As.

The receiver product detector (another hand made DBM) followed by a low impedance common base audio pre-amplifier and my favourite TDA2003 audio power amp stage followed. With headers for two audio filter daughter-boards.

Double balanced diode product detector, low impedance (common base) preamp, headers for audio filters, TDA2003 audio power amp.

Then an IF amplifier, three pairs of BF246s in cascode configuration, tuned drains, so much gain it was hard to tame.

9MHz IF amplifier by EI9GQ, three pairs of BF246A JFETs. IF derived AGC at right.

Finally, I mounted the three KVG 9MHz filters on a board, with an L-match at each end of each filter, and miniature relay switching with more 2N7000 drivers, in anticipation of direct or I2C crystal filter selection (USB, LSB, AM).

9MHz IF crystal filters (KVG, 8 pole, 2.7khz/3dB), with relay switching and impedance matching.

On this build I adopted a simple mechanism of placing a 3-pin 0.1″ header at each stage’s input and output. 0.1″ headers and pins are cheap and good enough for small signal RF paths at HF. The convention I adopted was to put signal on the middle pin and earth either side so that the 3-pin header is reversible. Headers at each end of a module allow the stage to be isolated for testing/debugging or temporary replacement with an alternate module/board. Carl K0MWC describes his implementation of this technique here.

VFO (PLL/Controller)

I used one of Farhan VU2ESE’s Raduino modules, comprised of a Nano, si5351 and socketed 2×16 LCD. As this was to be a base rig, I tried out a large format 20×4 LCD that I had on hand. This is pin compatible with the smaller HD7044 style displays. Raduino doesn’t bring analogue GPIOs A4 and A5 out to pins, so I had to hack a small 0.1″ header socket onto the back of the board for this. After some juggling of the backlight display voltage (it draws 0.8 amps at full brightness and could pass as a room light at night), I built the whole assembly onto an aluminium bracket with a PCB for connectors and power running along the base.

Because I had the display real-estate, I allowed for an Adafruit real-time clock module on the I2C bus, to bring time out to the big display — useful for contesting and logging. I also anticipate adding another Adafruit breakout, the INA219 voltage and high-side current sensor, for monitoring supply voltage and PA current. Yet another I2C accessory is a thermometer, which could be placed directly on the PA heatsink. Not essential, but nice to have. Sample data is shown in these pictures as a placeholder for when these cards go in later.

The big LCD is a real eye-ful, and can be read from the far side of the room. I modified the code to spread available frequency, VFO, mode, s-meter, supply voltage, current, and time information across the four lines (not operating yet).

From breadboard to chassis

An aluminium sheet was adopted temporarily as a base to tape down the receiver modules for initial testing. This approach allowed breadboarding of the receiver modules, before committing to the ultimate case.

Testing receiver modules on an aluminum baseboard.

After a month or so of using the receiver as shown en plein air, I wanted to show it off to the local homebrew group (the one hosted by Amateur Radio Victoria). So I got cracking on an aluminum case. Having had a good experience with angle aluminum worked with hand power tools, I followed the same approach. I used 100 x 25 x 3mm angle stock for the front, rear and sides, with a 3mm sheet aluminum floor, and a drop in 2mm aluminum lid sitting on side angle rails. Sides are fastened to the base with countersink M3 bolts to allow disassembly. A few permanent fixtures such as the rails are fastened with pop rivets.

Case from 3mm aluminum angle and sheet taking shape. 29cm wide, 32cm deep.

With LCD and some controls mounted. Note the striking similarities with Summit Prowler IV, behind.

Socketry, switchery

I used my preferred sockets and switching mechanisms, push-buttons for VFO and band control, pushbuttons for keyer messages, a conventional microphone socket, an in-built speaker with a switched external speaker socket. A DB-9 socket on the back panel will expose some of the control lines, including T/R for an external linear or transverter.

Performance

How well does it work? Pretty well. The video above will give you an idea, bearing in mind that the antennas are all simple dipoles and we are around sunspot cycle minima. I couldn’t demonstrate it above 30m as I don’t have antennas for higher bands, and they are all but dead at this stage anyway.

There is no sophisticated test equipment in the VK3HN shack to measure sensitivity or other essential receiver characteristics, but in A/B tests with my Icom IC746Pro and Summit Prowler IV (another homebrew project, a superhet 6 band SSB/CW transceiver using SA612 mixers) it is equally sensitive, and quieter than both these other receivers. It is my preferred receiver in the shack for 160, 80 and 40 and 30m. With the antenna disconnected and audio and IF gains up it is difficult to tell it is switched on. I am happy with all of EI9GQ’s modules and at this stage would not change any of them. That proves what I thought about Eamon’s designs — the circuits are well conceived, accurately reproduced on the pages, and work well. And he has managed to get the receiver gain and stability just about spot on.

I am particularly impressed with the post-mixer amplifier stage, which is very similar to David Norton’s noiseless feedback amplifier (see Clifton Labs Preamplifier, the product is here, the circuit may be online elsewhere), a broadband 10dB RF preamp with push-pull 2N5109s (I used 2N2219As as mine is a dedicated 9MHz IF amplifier). I used it in front of Summit Prowler IV and it noticeably improved that receiver’s sensitivity (liveliness) across 160 to 30 meters, so much so that I intend making another one up to retrofit into that rig.

The receiver draws 1 amp, but over 700mA is the large LCD backlight. The Nano/si5351 assembly draws almost 100mA, the Norton post-mixer amp 80mA (it’s hot in more ways than one), and the pairs of relays in the bandpass and IF filter (4 energised at any time) about 25mA each.

Showing off the rig at the October 2018 meeting of the Homebrew Group at Amateur Radio Victoria. OM in the middle is Glenn VK3PE who has also built the EI9GQ transceiver.

Disclaimer

I’m yet to get the EI9GQ AGC circuit working. Glenn VK3PE built this circuit and reports that the AGC line swings high (not low) to decrease gain, so if you build it, you may need to invert its output, depending on what IF amplifier you use.

The frequency on the display in the video is 1.2 to 1.6 kHz low.  This is because I have not calibrated the si5351, and additionally, the exact frequency of the crystal BFO for the LSB filter has not been allowed for in the code.  The same will need to be done for the USB filter for 30m and up.

Line 3 on the display is test data.  This will become live when the voltage, current and temperature breakouts are added.

The realtime clock starts working half way through the video.  This is because the video segments were shot over a week during which I was still working on the receiver .

Next

Part 2 of this series will continue with the transmitter build and on-air testing.