38 Special QRP Tranceiver Kit

A 30m QRP Transceiver

The NorCal QRP Club sponsored the first annual Dayton Building Contest in 1996. We selected 2 projects, the 49er 40 Meter Transceiver designed by Wayne Burdick, N6KR, and the Pipsqueak Regenerative Receiver designed by Paul Harden, NA5N. To add a little spice to the contest and generate some technical articles, we decided to have a design contest to go along with the building contest, with the winning designs published in QQ and QRPp. The winning designers will be awarded a plaque from NorCal and the winning design kitted. This year, we had two great winners, the “38 Special”, a 30M transceiver designed by Ori Mizrahi-Shalom, AC6AN, and the “Rainbow Tuner”, designed by Joe Everhart, N2CX.

NorCal will kit the “38 Special” and the New Jersey QRP Club will kit the “Rainbow Tuner”. We found that NorCal did not have the time to do both kits, and the New Jersey QRP Club graciously stepped forward to help us out. QRPers are the winners in the design contest, because there are now two great $25 kits out there to build.

The point of the design contest was to provide kits for the Dayton 97 Building contest, which is again sponsored by NorCal QRP Club. There will be 3 prizes in each category, with the judges decision final. The rules are simple. The project must work, and mods are allowed. You may build from the kits available, or you may home-brew your own. Chuck Adams, K5FO is the head judge, and he will select two additional judges the night of the contest. Judging will be held at the ARCI Hospitality Room at 8:00 on Saturday night of Dayton.

I would like to thank all of the designers for their efforts and time that they have put in to providing the rest of us with great projects to build. Good luck builders, and we will all get to see the winners in Dayton. 72, Doug, KI6DS

What radio …

  • is a superhet with only two NE602s?
  • has no discrete transistors?
  • sold for $25?

Can’t guess? The answer is the “38 Special”. The “38 Special” is an exercise in minimization. We feel that it is “special” because of the following:

  • It works on the under-utilized 30M band
  • Runs of f a 8V supply (due to limitations of some components)
  • Special in its architecture
  • Excellent price/performance ratio
  • Last, but certainly not least, the fact that it offers a great platform for another year of hacking…

The “38 Special” is the result of an effort which started with hacking the 40-9er transceiver. A novelty radio, the direct-conversion 40-9er was quite limited, especially for weak signal work. The “38 Special” rectifies most of these problems and even adds a few nice features. It came to life due to the design contest challenge and the never ending enthusiasm and encouragement by Doug, KI6DS, who also came up with the great name.

General Description

The “38 Special” is a superhet transceiver for the 30M band. The first thing you’ll notice looking at the schematics is the lack of discrete transistors. I have been playing with this concept for some time and the 38 Special was the right vehicle. The transmit section relies on a TTL buffer. I extended this concept to utilize the same TTL chip for other tasks. Although it’s an NE602-based superhet transceiver, the “38 Special” incorporates only two NE602s. I reuse the product-detector as the transmit mixer by channeling different signals to that chip on receive and transmit. Also, the traditional LM380/386 is gone in favor of a more versatile dual op-amp circuit for the audio section.

Receiver Front-End

The receiver front-end starts with a back-to-back diode switch. The “38 Special” utilizes 1N4007 diodes for the switch. This diode has a PIN structure and provides low insertion loss, although it suffers from poor zero bias isolation and does not offer a strong IMD performance like an RF-rated PIN diode. (1) It is superior to the 1N914 or similar diodes in this type of design. Next is a toroidal impedance transformer with a 10.1 MHz tuned circuit at its output, providing additional front-end selectivity to that offered by the transmit output network. Provisions were made to include a 10 KOhm pot for RF-GAIN control. The 10.1 MHz RF signal is fed to the input of the NE602 receive mixer, where it is mixed with the 22.1 MHz VXO to generate the 12.0 MHz IF frequency.

The VXO

The superhet circuit enabled me to use standard crystals and avoid the high price of custom ham-band crystals. Many crystal combinations work for most HF bands. I chose a high frequency first crystal to achieve a high frequency swing. (2) This required a relatively high IF in the simple receiver. The NE602 Collpits oscillator required a high DC bias for a large swing, provided by a 3.9 KOhm resistor at pin 7. Although well below the value recommended, this resistor provides for stable operation of the NE602. “Rubbering” the crystal with a varicap allows relocating the tuning pot away from the oscillator, if desired. A 1N4000-family diode works here nicely as a varicap. (3)

A little assist from a molded inductor yields a tuning range of 25 KHz. A little hint here for the experimenters, do not replace this inductor with a toroid. The low Q helps to increase the pulling range. The VXO signal is mixed inside the NE602 with the received signal to produce an IF output of 12.0 MHz, which is the difference between the VXO and the RF frequency. During transmit, a 22.1 MHz signal is taken from the Collpits oscillator and injected into the input of the transmit mixer.

If Filter

The IF filter is implemented with a single crystal. It is a few KHz wide, due to budget constraints. Provisions for a better IF filter are included in the board layout. The main selectivity is achieved at the audio stage and the wide IF filter greatly simplifies the alignment of the receiver. The wide IF filter provides very little “wrong” sideband attenuation. Due to this, the “38 Special” in the stock form cannot be classified as a single-signal receiver. It will take a much sharper IF filter to achieve that. Although you will hear the same signal twice, the sharp audio filter totally eliminates the “off” signal.

Product Detector

In the cost cutting tradition I left out the “traditional” third NE602 for the transmit mixing. Instead, the 38 SPECIAL reuses the product-detector for the same function. On receive, an oscillator (IF frequency) is mixed with the IF signal and it results in a low-level audio signal. On transmit, a signal from the VXO is mixed with the IF frequency oscillator in the second NE602. The selection of the input signal to the second NE602 is done by means of a 4066 analog multiplexer. Other than the switching of signals with the 4066, the receiver is similar in concept to most NE602-based superhet rigs.

Receive Offset

Sharing the product-detector and transmit mixer required a “trick” to achieve a receive offset. The 38 SPECIAL “pulls” the IF frequency oscillator about 500 Hz up on receive with a 100 pF capacitor in series with the 12.0 MHz oscillator crystal. During transmit, this capacitor is shunted to ground with a parallel forward-conducting diode, so the crystal oscillates right on its fundamental frequency, resulting in a zero-beat transmit signal. The down-conversion at the first mixer and this oscillator pulling up on receive combine to yield the “right” receive sideband at a lower frequency. So, although the IF filter allows either sideband through, it is easy to identify the “right” one. This is not an issue when you call a CQ. The answering station is on the right frequency, if it zero-beats with your transmit signal.

Audio AMP/Filter

The audio is filtered and amplified by an NE5532A dual op-amp, instead of the “traditional” LM380 or LM386 chips. (4) The 5532 requires more external components, but it gives a higher gain, and more important, the circuit also forms a sharp band-pass filter. From that point of view, the “38 Special” is superior to most NE602-based radios. This amp delivers about 60 dB of gain while driving a walkman-style headphones. The filter offers a 50 Hz -6 dB bandwidth and about 400 Hz at -30 dB. This circuit uses a dozen more components than an LM386. But they are probably the most cost-effective components in the whole radio!

Transmit Chain

As mentioned before, the product-detector doubles as a transmit mixer. The signal on the output is filtered by a tuned circuit, and the 10.1 MHz output of 100 mV is used to drive the two-stage transmit amplifier. A few examples in the literature describe the use of TTL chips (and other logic devices) for a low-power transmitter. (5)

The most interesting article on the subject was written by Len Smith and appeared in QST. Len used an octal inverting buffer with eight individual active devices, but really utilized only five of them. This circuit gave me the idea of using the leftover devices for sidetone generation and other tasks.

One of the inverters is biased as a high gain linear amplifier. The 100 mV at its input comes out as a few volts on swing at the output. The single inverter is strong enough to drive the final circuit, which is made of four parallel inverters. Depending on the output matching, this circuit can deliver well over half a watt of output power. I chose to leave it at 400 mW for the sake of cool and safe operation of the final. The board supports additional circuitry (not supplied with the kit) for a higher output of up to 5W.

The Output Network

The use of only 8V supply required very low impedance for the final to give any appreciable output power. The matching is easily done by an L-C-L-C type network.

Sidetone

Two inverters combine in a simple oscillator circuit, as described in the ARRL handbook. The oscillator is clamped to ground during receive but is free-running during key-down. The output is attenuated by means of a large series resistor. There was no need to filter the square waveform, as the audio bandpass circuit does that anyway.

Additional Logic

One inverter of the octal buffer chip forms the receive/transmit logic. It inverts the logic state of the “key” line, so when key is down, this signal is at full Vcc and when key is open, this signal is at 0V. The availability of the T and R signals simplifies the implementation of the T/R circuitry.

Alignment

The “38 Special” was designed with a novice builder in mind. There are only two alignment steps, although more parameters could be tweaked by the experienced builder. The receiver alignment consists of peaking the front-end trimcap for the highest receive audio.

The transmitter alignment is a bit more tricky. The radio transmits 10.1 MHz signal, when it has a 12.0 MHz IF signal in the output of the transmit mixer. This requires care when tuning the transmit filter trimcap. One way of doing that is by listening to your signal on another receiver. Tune close to your center frequency and look for the adjustment that results in an output with the least close-in spurs. The two alignment steps do not require any test equipment, although having such equipment can improve the alignment.

The Next Step

The final design reflects cost-cutting and other changes to simplify the alignment and kitting of the design. It is a superhet with offset and sidetone at a 40-9er price! As such, it has limitations, of course. I look at it as a product and a development platform. There will be many that will assemble the basic unit and have lots of fun with it in the stock form. But there will be those who want to do things their way. This radio was designed for both.

I will not continue its development. I leave that to the hackers and tinkerers out there. The “38 Special” was designed to continue the tradition that started with the 40-9er – mods by the dozen. There are many possibilities. In fact, some mods are being developed as we speak. I certainly encourage people to do just that. All I ask is that you share with the QRP community and let us all know of your adventures…

For those that build it stock or custom, plain or modified, hot or low key – have fun. I can’t wait to see the entries for Dayton and Pacificon next year. I also hope to get more activity on the 30M band. Maybe we will finally know what the propagation properties really are on 30…

Acknowledgements

Again, many thanks to Doug Hendricks for the encouragement. Special thanks to Dave Fifield, KQ6FR, who built the second prototype and beat me to the first QSO – 549 on 200 mW from San Jose to Spokane! Dave is a master builder, both of circuits and enclosures. He came up with many additions and suggestions that made this radio nicer and more robust. Special thanks to the XYL, who became a “radio-widow” during the last month…

Notes

  • QST, Dec 1994 pp. 25-27.
  • A similar concept is used in the Mizuho MX7S radio.
  • See Jim Pepper’s Deluxe QRP station, although not for rubbering a crystal.
  • One designed by NN1G uses the 5532A for an audio amp but the filter seems to be not as sharp (that radio has a sharp IF filter)
  • QRPp Mar. 1994 p. 58 and W1FB’s Design Notebook p. 156.
  • Solid State Design p. 53.
Published
Categorized as Kits

Leave a comment

Your email address will not be published. Required fields are marked *