Every now and again, I come across web-sites featuring designs for discrete component radio circuits. My designs have always tended toward the specialized IC method. If for example, you want an analog multiplier, use an analog multiplier IC. As a result, I’m always impressed by the ingenious circuits designers have come up with to get the job done.
Some of these circuits can be found on the Radio Constructors web-site. Radio Constructor was a British magazine aimed at the radio/electronics hobbyist which ran from the 1930’s to the late 1970’s. Among the designers featured were Douglas Hall, G.W. Short and R. A. Penfold. The surprising thing was that a lot of these writers weren’t professional engineers or technicians (Hall for example was a career diplomat), so even if some of the designs were suspect to say the least, you have to admire their inventiveness.
R. A. Penfold has been writing radio and electronics articles since the mid 1970’s. Along the way he wrote “Short Wave Superhet Receiver Construction” in 1991. You’d think that in the connected age of the 21st century a basic short wave receiver wouldn’t hold that much appeal, but there’s still something about picking up a distant radio station, or eavesdropping on amateur radio operators with nary an internet connection in sight that piques the interest. In addition, building circuits from older schematics and, by now, obsolete parts forces you to think about the concepts behind the design.
That said, I decided to have a crack at building a Short Wave Superhet receiver using more accessible components, and whatever I had lying around.
2.0) The Basic Circuit
I’ve posted the original schematic from the book below:
The original VFO is shown here:
My version is shown here:
There’s not much difference in the circuit configuration between my version and the original. The original used a 40673 for TR1, and BC547s for TR2 and TR3. I’ve used a surface mount BF998 on an adapter from Digikey for the mixer, and good old 2N3904’s for Q2 and Q3. I used the indicated IF transformers and ceramic filter which are available at Mouser and Electronix Express. I had a reduced drive variable capacitor on hand for VC1, but they are available at Midnight Science. In the pre-selector circuit, I used a varactor diode tuner D2, controlled by VR1. The NTE618 is available from Mouser, and it may lend itself to some sort of automatic pre-selector control later on. For now, the potentiometer control seems to give adequate results.
The main stumbling block in older circuits is the inductors. The original schematic called for some Toko inductors which by now seem to be obsolete. I’ve listed the values and part numbers below for reference:
Freq. Range Preselector (T1) VFO (T4)
1.7– 5.0 MHz KANK3333R KANK3426R
5.5– 14 MHz KANK3334R KANK3337R
12.0 – 30 MHz KANK3335R KANK3428R
Descriptions of the coils can be found here: toko.pdf
By way of an alternative, I had some Amidon type toroids on hand and decided to make my own transformers. The original uses plug-in inductors, and separate tuning for the VFO and pre-selector, which eases the requirement for variable core transformers. The values I came up with are listed below:
Freq. Range Preselector (T1) VFO (T4)
1.7– 4.9 MHz T68-2 T68-2
70 turns #28 AWG sec. 66T # 28 sec
6 turns #22 pri. 6T #22 pri
Inductance 28uH Inductance 25uH
4 – 11 MHz T68-6 T68-6
34T # 22 sec 33T #22 sec
6T #22 pri. 4T #22 pri
Inductance 5.3uH Inductance 5.0uH
9 – 29.5MHz T50-6 T50-6
17T #22 sec 14T #22 sec
4T #22 pri. 3T # 22 pri
Inductance 1.5uH Inductance 1uH
The transformers were each mounted on a small piece of perf board with a four pin single in-line connector which is plugged into a socket on the front end board.
2.2) I.F./AGC Amplifier
The IF/AGC circuit consists of transistors Q2, and Q3, ceramic filter CF1, transformer T3 and their associated components. The transistors are in the familiar common emitter configuration. Resistor R6 sets the input impedance of the ceramic filter, and the output impedance is set by the input resistance of Q3. The original receiver used BC547 transistors which have a lower hfe (typically 90) than the 2N3904. As a result, in this version, the input impedance is too high which led to oscillations at 455KHz. The solution was to lower R8 from the original 2.2M to 470K. This sets the DC input resistance to approximately 1.5K Which is closer to the specified input impedance.
Higher gain transistors also caused problems with the AGC operation. Audio levels are detected at the anode of germanium diode D1. The IF is filtered by C8 and R10. R9 and C6 further filter the audio to derive the AGC control voltage. Since the audio is derived from the anode of the diode, strong audio levels will correspond to greater negative voltages at the junction of C6 and R4. This in turn will decrease the current in Q2 decreasing its gain. Since Q2 and Q3 have higher gains, the AGC action is insufficient and low input signal levels are amplified. This generates an IF signal which becomes noticeable especially in SSB mode. The solution was to lower the value of R4 to 3K from the original 8.2K, the low level IF still exists, but is acceptable.
3.0) Product Detector/ Audio Power Amplifier
The BFO, product detector and audio power amplifier were placed on a separate board to make future enhancements a bit easier. The schematic is shown below:
The schematic for the original BFO is shown below:
Its the usual Hartley configuration but when examined on a scope, the output, while a recognizable sine wave, looked very distorted. It didn’t seem to affect performance, but I decided to switch to the current version for esthetics. In this version diode D2 clamps the FET’s gate signal to ground to prevent conduction of the gate source diode. The diode is reported to increase phase noise in oscillators, but this doesn’t seem to be a problem at 455KHz SSB or CW. The BFO frequency can be set by adjusting the core of transformer T1. D1 is used as a varactor diode which can vary the BFO frequency via VR1. TP1 can be used to monitor the BFO frequency while having minimal effect.
3.2) Product Detector
At first glance, the product detector might seem a bit unusual. It seems like it it should be some sort of differential pair mixer. On further inspection the circuit turns out to be a high input impedance version of the familiar single transistor mixer. The higher impedance results from the differential (or long tailed pair) connection of Q2 and Q3. The BFO output is isolated by R3, and is input to the base of Q3. At this point, the BFO sees a higher impedance than it would if it were connected directly to the emitter of Q2. The mixing action occurs due to the exponential voltage/current relationship in the base emitter junction of Q2 (or Q3).
The actual input impedance of the product detector can be approximated by the equation:
Rin = 2 * hfe/gm
where hfe is the current gain or beta of the transistor and gm is the transconductance given by:
gm = Icq/25mV where Icq is the DC collector current
Q2 and Q3 are biased at 0.5 mA which sets gm = 0.02 mho.
This in turn sets the input impedance to Rin = 2 * 150/ 0.02 = 15 K ohms, which is high enough not to load the BFO. Its also high enough to swamp the 1K resistor R3. Indeed, shorting out R3 has no apparent effect. Possibly R3 was added to accommodate lower gain transistors.
The output of the bipolar mixer includes the BFO and IF signals as well as various harmonics and inter-modulation signals. These are filtered out by C4, R5, and C5, R6 filters in the collectors of Q2 and Q3, and have 3dB cut-offs of about 4KHz. The IF signals are further filtered by the low pass filter at R9 and C6 with a 3dB cut-off of 27KHz.
3.3) Audio Power Amplifier
The audio power amplifier is the only departure from my new found fascination for discrete component circuits. Its a standard LM386 application, however, and its in the book, so I’ll allow it.
The input to the volume control VR2 is switched between the product detector output and the A.M. Output from the front end board by switch SW1A. SW1B switches power to the BFO and product detector, otherwise the audio amp can pick up the product detector output while in AM mode.
Power amplifier U1 is a standard LM386 application. C10 boosts the audio gain. C13 isolates the output DC from the speaker. R10, C12, and C11 ensure stability. In some versions of the LM386, R10 and C12 are not required but are included here for flexibility. Besides, its in the book.
As vintage general coverage receivers go the Short Wave Superhet seems to work fairly well. Its difficult to know how well it should work given my sub-optimum antenna, time of year, and location, but I’ve been able to pick up distant foreign A.M. stations as well as amateur operators with reasonable clarity.
The band switching with plug-in coils is a bit awkward, but that’s to be expected. The main issue seems to be with tuning. Even with the reduced drive variable capacitor, tuning is very sensitive. Possibly a bandspread capacitor would alleviate the problem, although a synthesizer with a better oscillator and frequency display would be even better.
All in all, its a pretty good receiver, and will provide a good basis for comparison when (or if) I start to make improvements.