Low Frequency Block Converter

This converter shifts the VLF/LF bands, from below 20 kHz to over 350 kHz to the shortwave frequencies, determined by the selected crystal. In this prototype, the crystal frequency is 5 MHz, so the LF band comes out at 5.02 MHZ to 5.35 MHz, suitable for use with a shortwave receiver. Notice the lack of an expensive mixer or special IC!

 

Circuit and Component Notes:

bullet

A capacitor is connected across the input connector to achieve a total capacitance to ground of about 2000 pF, including the antenna cable capacitance. My prototype has a 470 pF because my antenna cable is quite long with about 1500 pF capacitance.

bullet

The 1 meg resistor and NE2 help fight charge buildup and voltage transients, and the transistor junctions are asked to handle the surges when the bulb fires. If you want a little more protection, connect a couple of silicon diodes from the base of the first transistor to plus and ground (diode cathode to plus, and the other diode anode to ground). Those diodes will conduct if the neon fires. I suspect the transistors can handle it, with one base-emitter junction conducting via zener breakdown. If lightning does hurt it, the transistors are cheap!

bullet

Use good capacitors for the two 1000 pFs on the input; either COG ceramic or mica would be fine.

bullet

The transistors aren't particularly critical and other types should work fine.

bullet

The output transformer isn't very critical and just needs to have a reasonable impedance at the output frequency. I used a random ferrite balun core with 18 turns, center-tapped for the primary and 3 turns for the output. You could get similar results simply winding turns on a short piece of ferrite. If you are familiar with RF transformers, a low-Q resonated transformer would be fine, too. If the radio has a 50 ohm input, you might want the primary inductance to be over about 6 uH to have adequate bandwidth. More is better. Less will work due to all the conversion gain the circuit provides.

Improvements:

bullet

The experimenter might want to increase the emitter resistor from 15 ohms to soften the oscillation. Too big, and it will stop. Just right, and there will be less harmonic bleed-through. Not shown in the schematic is a 220 pF capacitor that I added to mine across the output connector to kill some of the higher frequency harmonics from the crystal oscillator. They weren't really causing a problem, but the capacitor reduces them on the 'scope without hurting the gain.

bullet

The balance of the transformer determines how much of the oscillator signal reaches the output. One could remove or add a turn to one end of the transformer to seek a better balance. A high carrier can kill the sensitivity of the shortwave receiver. The balance was fine on mine, reducing the oscillator signal to about 100 mV, p-p. Lower is better, but my receiver could reject that signal as close in as 10 kHz.

Construction isn't critical as my point-to-point bird's nest illustrates:

I added a few drops of silicon rubber to keep things from shifting! But the outside case looks good, at least! You can see one of the diodes mentioned above - I'd just leave them out.

I modeled the input filter with LTSpice: