T1: 1.2 to 2.1uH slug-tuned with 50 Ohm coupling link
T2: 3.5 to 6uH slug-tuned; 16t primary, 4t secondary
T3: AF step-down transformer, 11H unloaded primary inductance, 22-1 turns ratio
CV1: 15 to 30pF air-variable
C3, C11: 100pF
C6, C9, C15, C16: 100pF @ 100V
C4, C13: 22nF @ 50V
C5, C14: 100nF @ 100V
C7, C10: 1000pF @ 50V
C17: 4.7uF @ 50V electrolytic
C18, C19: 39pF @ 50V
C20: 3.9nF @ 100V
C21: 47uF @ 100V
R1, R4, R8: 1Meg
R2, R6: 47k
R3, R7: 1.5k
X1, X2: 11.0592MHz quartz crystal
L1, L3, L4: 10.7MHz slug-tuned IF "can" with internal capacitor removed, ~2.3uH
L2, L7: 3.5 to 6uH slug-tuned
L5: 2.2mH RFC
L6: ~10uH slug-tuned
HP: 600 Ohm headphones
V1, V2, V3: Raytheon subminiature pentodes; type QF721 (February 1953 date-code)
The first converter stage generates a VFO signal in resonator CV1/C1/T2 via link-coupled feedback from V1's anode. The VFO tunes from 2.9465 to 3.0165MHz in order to cover a receive frequency range of 14.0 to 14.07MHz. The 20M input signal from the antenna is injected via resonator T1/C2. Converters such as these can be plagued by interaction between the input and VFO resonators, however the problem may be alleviated to a large extent by operating the VFO at a frequency well-removed from the input signal resonator. In the above receiver, adjustments made to the 14MHz input tank circuit, or changes in input loading, pull the receive signal beat-note only very slightly. The low VFO operating frequency also aids the receive frequency stability.
The quartz crystal bandpass filter configuration dates from the 1930's; I believe it was first known as the "Telefunken Filter." DJ2KY appears to have further popularized this filter amongst European hams with a pair of articles titled, "Ein Amateur-Kleinsuper mit Quarzfilter." These appeared in the July 1956 and May 1957 editions of Funktechnik magazine. In the above circuit, L2 forms a parallel resonant circuit at the IF frequency in conjunction with the quartz crystal's parallel capacitance (holder capacitance). However, given this small capacitance (~5pF or less) would require a rather large resonating inductance, I've added an external capacitor, C8, in parallel. Using an 11.0592MHz "microprocessor" crystal taken from my junkbox I measured the bandpass filter center frequency at 11.0535MHz. The opposite sideband signal suppression approaches 30dB...a modest performance by modern standards, but it feels like luxury having exclusively used simple regenerative and direct-conversion receivers these past four years!
I chose to use this filter primarily out of historical interest. No doubt the more common, one-crystal, half-wave bridge configuration would have worked as well; or better yet, a three or four pole ladder crystal bandpass filter.
As explained in my April 22, 2012 blog post, I had originally hoped to build a 20m superheterodyne receiver using only two sub-miniature pentodes. However, the quartz crystal filter had too much interaction with the crystal-controlled oscillator frequency in the second converter stage. I found it necessary to include an IF amplifer/buffer stage between the first and second frequency converters.
The second converter stage is somewhat unusual inasmuch as the screen grid (G2) is used as the control grid of a quartz crystal-controlled Miller oscillator that serves as my BFO. I chose the (parallel resonant) Miller configuration given its suitability for use as a VXO. Using an adjustable inductor in series with the 11.0592MHz crystal I had no trouble pulling the BFO frequency down to my 11.0535MHz bandpass filter frequency plus or minus ~800Hz.
I found that adjusting L7 to produce the maximum BFO signal amplitude at the V3 screen (G2) overdrove the second mixer and reduced its conversion efficiency. The situation was greatly improved by de-tuning L7 such that the BFO signal amplitude measured ~500mVpp on the V3 screen (using a high-impedance probe). This seems reasonable given that my early 1970's edition of the ARRL's Amateur Radio Handbook recommends that the sum of the signal and oscillator voltages impressed upon a pentode mixer grid should not exceed the grid bias voltage.
The receiver audio output signal level well suits me, although I admit that I generally prefer receivers with a considerably lower than normal audio output level. I think most folks would add at least one audio amplifier stage to this design.
Speaking of which, I did go on to build a two-tube version of this receiver. Only, I used a sub-miniature 1V6 pentode/triode in place of the 2nd converter stage shown above. The improved BFO isolation provided by the 1V6 allowed for the elimination of the IF amplifier/buffer stage. However, having become accustomed to listening to signals with the above receiver for several weeks, the minimized QF721/1V6 design sounded a bit too quiet for my liking! I ended up adding one-stage of audio amplification using a black, top-hat style, germanium transistor; type 4JD2A6, dating from 1958. The collector supply potential was provided by my 1.2Vdc filament supply voltage.
Here are two close-up photos of the QF721/1V6 receiver. The first converter and crystal filter are shown in the first photo (notice that I'm not canceling the quartz crystal parallel capacitance in this photo). The 1V6-based 2nd converter and transistor audio frequency amplifier appear in the second photo.
Both receivers appear to work equally well aside from the fact that the 1V6 is considerably more microphonic than the QF721.