It's wonderful to watch your ideas take shape, and weird to watch the shape change as the ideas mature into workable products. The latest radio is no exception. I went through a panic stage four days ago, and the direct-conversion receiver / cw rig turned into a Pixie II, and back again. Do a search for the Pixie II, it's a fascinating exercise in minimalism. The PA transistor doubles as the receive mixer; keying the PA emitter shorts the receive path and puts the transistor into full drive. On key-up, recovered audio is passed to an LM386, and into a pair of 32-ohm headphones. Usual power source is a 9V PP3, and the housing can be anything from an Altoids tin to al fresco.
I did say it turned back again, and it remains a variant of the Rev. George Dobbs G3RJV's famous 'Sudden' receiver. I'm using Micrometals iron toroids in the receive preselector, and a 3.58MHz ceramic resonator as the frequency controlling element. I need tighter control than George's VFO original, because it's also transmitting. I've settled (for the time being) on a 30kHz band, from 3.55 to 3.58. This includes the QRS 'sandpit' at 3.555 and the QRP centre of activity at 3.560. The radio has a 700Hz transmit offset, given by arranging for a pair of resistors to be switched in and out of the frequency control voltage 'totem-pole', which feeds the LED varactor. Not only am I using a LED as a varactor, but I find that forward bias gives more linearity. This method is not original, but has been used with success in the radio home-brewing world for some years. The other feature of the forward bias is that the LED can be brought out to the front panel, adding interest as the radio is tuned. The LED dims and brightens, glowing well at the lowest frequency.
I'm currently waiting for parts, and I have two evenings ahead of me where I'm committed elsewhere, so it'll be just ideas until the latter part of the next weekend. With a following wind, the prototype may be running inside a week, and taking reports from Southern England. We'll see.