I did this design for the Part 15 Longwave (1750M: 160KHz -- 190KHz) band. There is not a whole lot of commercial rigs that can cover this band, and the general coverage receivers that do don't cover it very well. The twin problems are a restriction of 1.0W of DC power into the final, and an antenna height restriction of 15M. The latter makes for a very
short vertical. Such antennae are EEEEENORMOUSLY
reactive with miniscule radiation resistances. Even with a highly efficient Class E power final, getting out 5.0mW is quite an accomplishment indeed.
General coverage receivers weren't designed to dig the typical "LOWFER" signal out of the muck. More powerful NDB stations, but not Part 15 signals. That leaves few alternatives but to DiY. To that end, this is a dedicated design that tunes across the 1750M band. This makes optimization possible. This is a necessity due to the nature of the signals you're trying to receive.
To that end, the passive doubly balanced mixer was chosen. This design isn't easily overdriven, is significantly quieter than either singly balanced, or unbalanced mixers. The only drawbacks are gain loss (make that up with subsequent amplification) and a rather high power requirement. Since Vf= 0.7V for a Si diode, the minimum power required is 4.9mW (6.9dbm). It works better with higher LO power: improved noise figure and higher gain (less loss actually). About 20mW represents the point of diminishing returns. Since the passive DBM works much better if all its ports are properly matched, a -3db resistive pad is included. The LO strip puts out 40mW (20mW + 20mW across the resistive pad).
The front end includes a Butterworth derived BPF to keep AM BCB out of the mixer. The audio output includes a third order Butterworth LPF. This sets an upper frequency of 3.18KHz. A CW filter (fc= 830Hz; Q= 10) is included for noise reduction and improved clarity. This filter can be switched out while searching for a signal, and switched in once one has been found. This requires a stable LO that won't drift out of the filter passband.
The LO is based on the series tuned Colpitts that was popular back in the 1950s. This topology allows for a high-Q tuner with a large L and small C. At these low frequencies, that's important since variable capacitors aren't all that large. By running it into very light loading, it is load isolated from varying loads that cause frequency pulling. Spectral purity is improved by the use of a tuned buffer, and a Pi-L matching network to match impedances between the 625R output impedance of the final and the 50R impedance of the DBM IF port. The oscillator shows some start-up drift due to the warming of the ferrite core the tuner inductor is wound. This takes place slowly enough, and soon settles down so that it is not an inconvenience.
The first audio pre was designed to match the DBM IF port. This being a grounded base design biased for an input impedance of 50R. It still manages a respectable Av= 30V/V. This stage is isolated from the DC rail by means of an active decoupler. This is important, given the high gain of the audio strip. More care than usual is required if audio frequency oscillations are to be avoided. The audio final is a conventional design, and puts out 1.5W -- sufficient for driving the speaker, a car radio replacement designed for 3.0W.Front EndLO StripAudio Strip