A Homebrew 40-meter CW Transceiver

by John Wiseman, KE3QG

This article describes a homebrew 40-meter CW transceiver that I designed and constructed myself. It is not meant to be a cookbook-style article, so no schematics or detailed circuit descriptions are given. I have tried to give a high-level description of some of the interesting portions of the design, such as the tuner interface. I initially designed this radio to be a receiver only. My original design was all analog, including the VFO, where I used the tuning capacitor and shielded tank circuit from an old Heathkit HW-99 that I had laying around. The receiver was constructed as a collection of modules, with the idea of being able to easily modify certain pieces of it later. This strategy paid off well when I upgraded the design to a full transceiver later on.

The receiver modules were the VFO, BFO, audio amplifier, and RF boards. The RF board consisted of an input bandpass filter, RF amplifier, RF-to-IF mixer, 3-crystal IF filter, IF amplifier, IF-to-audio mixer, and audio interface circuitry. After I got this design to work successfully, I decided to upgrade the VFO to use a direct digital synthesizer (DDS). I purchased a board from Novatech to provide the basic functionality, but I needed a way of actually varying the DDS frequency so that I could use it as a tuner. I designed and built an adapter to the board that would receive signals sent from the parallel port of a laptop PC. I wrote a C program that ran under DOS that would allow me to input a frequency via the keyboard, then I could scan the tuner either upwards or downwards by using the ‘up’ and ‘down’ arrows on the keyboard respectively. I also added the ability to input a step-size resolution from the keyboard, so that I could fine-tune in resolutions of say 10 Hz., or coarse-tune in greater steps, for example 1000 Hz.

I was pleasantly surprised by the performance of the receiver, and I started wishing that I had the ability to transmit back to some of the stations that I was listening to. As such, I decided to go all the way and turn the unit into a full transceiver. To so this, I needed to add a secondary bank of parallel port registers to the DDS interface to store the necessary transmitting frequency. I also needed to design a sequencer to control the switching of the DDS registers, the DDS itself, the power keying of the transmitter’s RF amplifier, and the substitution of an oscillator side-tone for the normal audio output during transmitting. Operation of the sequencer is initiated by pressing down on the key. This switches the DDS to operate at the transmitter frequency for as long as the key is held down. When it is released, the register bank is switched back and the DDS returns to operate at the frequency necessary for the receiver.

The transmitter is driven directly by the DDS output. In other words, the desired transmitting frequency is generated directly by the DDS, then sent to an RF amplifier and out to the antenna via a lowpass filter. In receive mode, the DDS generates a frequency that when mixed with the incoming RF signal, will provide the correct IF frequency for the IF stages. The PC software is programmed to calculate the correct values for a desired operating frequency, and for every change in operating frequencies, two separate frequency values (1 for receiving, one for transmitting) must be sent to the DDS via the parallel port interface.

The transmitter RF amplifier is a 3-transistor design, with the final stage operating as a class-C amplifier for high efficiency. The output of this amplifier has been measured at 15 watts RMS into a 50-ohm load. Harmonics are reduced by using a 4-capacitor, 3-inductor lowpass filter on the output of the amplifier. All of the RF filter inductors and the amplifier bifilar transformers (22 in total!) were designed with information found in the ARRL Handbook, and handwound on the appropriate toroids.

Mechanically, the radio is constructed as a collection of small circuit boards. The main receiver board is a copper-clad board with components soldered directly to the copper and to each other. Some etching of the board was done by hand with a Dremel tool when better stability was required for the mounting of the components. Most of the other boards are built using general-purpose pre-drilled prototyping boards commonly available at Radio Shack. Where power and ground planes are required, I used runs of thick copper tape. After testing the radio, all of the boards were mounted in the chassis of an old HW-99, with the receiver section on the top and the transmitter and sequencer located on the bottom. All interconnections between modules are made with short runs of thin coaxial cable. Power for the radio is provided by a dual +15 volt and –15 volt bench power supply. Where required for noise considerations, various modules have their own local 3-terminal voltage regulators powered off of the main input supplies. See Photos 1 & 2.

Photo 1

The top side of the chassis showing the receiver modules. Clockwise from the top right - DDS with parallel port interface; RF board; audio; BFO. The volume control is on the front panel, while the antenna connection, audio output, and keyer input are located on the rear panel.

Photo 2

The bottom side of the chassis showing the transmitter modules. The transmit/receive sequencer is at the top left; the 1st 2 transistor stages of the output amplifier are at the bottom left; and the final stage of the RF amplifier is in the middle, mounted on the copper-clad board. Because the amplifier is mounted on the bottom of the chassis, a small DC fan is mounted on the enclosure, just over the output transistor's heatsink. This fan can be seen in the lower left-hand corner of the picture. The board in the middle contains two voltage regulators (+ and - 5 volts for the sequencer and DDS boards).

With the radio in this configuration, many QSO’s were performed with excellent signal reports being received. I then decided to upgrade the user interface for the tuner by doing 2 different things. The first approach was a software solution and the second was a hardware solution. The idea for the software upgrade came from seeing advertisements of radios in QST that used visual representations of radio front panels drawn on the PC screen. Since my wife was working as a Visual Basic programmer and she was doing similar things at work, I asked her to create a VB program that would ‘draw’ a radio front panel on the screen, and she could call the C function that I had written previously to interface to the PC’s parallel port. With this program, operating frequencies may be directly entered via the keyboard, or changed with only a click of the mouse on the appropriate display button. See Photo 3.

Photo 3

A close-up view of the PC screen showing the Visual Basic user interface. Frequencies may be entered in 3 different ways. First, directly by clicking on the numeric keypad (right center) with the mouse, then clicking on 'Enter'. Second, by clicking the mouse on the 'up' or 'down' arrow keys located next to the frequency readouts. The resolution of the tuning steps is chosen with the step size selector at the far right. In this case, the marker is set for a resolution of 100 Hz. The third way to change frequencies is to directly go to a stored frequency. These frequencies are stored in the box below the 2nd frequency readout indicator. The memory is written by clicking on 'Save' and cleared by clicking on 'Delete'. Typically, only the first frequency readout (top one) is used. When the receiver incremental tuning (RIT) feature is turned on, the bottom readout becomes active to display the different receiving frequency currently being used. In the example shown, the radio is transmitting on 7.1250 MHz. and receiving on 7.1262 MHz. The Visual Basic program calls a C routine that drives the parallel port with the appropriate data and control signals.

The second approach to providing an upgraded tuner interface was to replace the PC with dedicated hardware. I was working for a company that had just purchased design software to utilize Altera FPGA chips, so I decided to try implementing this design as a quick test of the software. I designed the hardware entirely in VHDL, simulated the design with a VHDL simulator, synthesized the logic with a Synopsys synthesizer, converted the gate equivalent into Altera format with the Altera logic compiler, and finally programmed an EPROM for use on an FPGA evaluation board. This single chip not only provides the driving signals necessary to emulate a parallel port interface, but it also drives an electroluminescent display panel showing the user what the operating frequency is. Input is provided by an ‘up’ and a ‘down’ button, with selectable ‘fine’ or ‘coarse’ tuning capability of 10 or 1000 Hz. respectively. See Photo 4. A complete description of this design may be found in an article that I wrote for the December 1997 issue of QEX entitled "Modern Digital Design Tools for the Radio Amateur".

Photo 4

The Altera 81500 FPGA DDS and display circuitry in a prototype board. The toggle switch below the display module is the FINE TUNING control (10 or 1000 Hz tuning steps), and the two push buttons select UP or DOWN tuning.

Although not directly a part of this radio’s design, I have experimented with the use of digital filters, and I have tried them with this radio. I have designed several different lowpass and bandpass FIR digital filters, as well as an LMS adaptive filter, and have implemented them with a Texas Instruments TMS320C30 DSP evaluation module (EVM) in my PC. When using this board, I place the EVM between the audio output of the radio and the speaker so that the DSP is processing baseband audio. For more information on digital filters and how they may be implemented with this DSP chip, see the July 1996 issue of QEX for my article entitled "A Complete DSP Design Example Using FIR Filters".

The performance of the radio is surprisingly good. I am limited to a non-optimal 40-meter dipole strung between my house and a tree, as I am a victim of a restrictive covenant in my neighborhood, yet even with this antenna I have managed to make several contacts to the west coast and the Caribbean, and have copied stations as far away as Hawaii. I have performed some brief qualitative receiver comparisons between this radio and an Icom 735 by using a switch to quickly change my antenna to either radio, and I have seen that only in a few cases have I been able to dig a signal out of the noise with the Icom that I could not copy with the homebrew rig. For most signals that I would try to have a QSO with on either radio, the receiver performance is certainly adequate. I have not done any laboratory testing on the radio, so no official performance numbers are available for comparison.

It is certainly a thrill to operate a radio that I personally designed and constructed from the ground up while utilizing several diverse areas of electronics, but it is made all the more exciting when I receive high-quality signal reports from other amateurs. I always tell the operator on the other end that I am running a custom designed and built homebrew, and they are usually very complimentary on the overall quality of the signal. I have received excellent reports on the tone of the signal, with no apparent key clicks or power supply modulations to degrade the signal quality. All in all, several operators have expressed their surprise that it is not a store-bought rig, and that is taken here as the ultimate compliment.

Photo 5

The complete radio on my desk, ready to operate. From the bottom left - power supply, homebrew radio, Heathkit QRP wattmeter and antenna tuner. Although I typically use headphones, I also have an external Heathkit speaker that is not shown in the picture. On the right - laptop PC with Visual Basic tuner interface program running.