[Today’s run: 1.x mile with wife and dog]
I was reading an old article on varactor diodes in the QRP Classics book, written by Doug DeMaw of course.
I have a nice QRP rig, the Willamette by NT7S, I use it all the time. It has varactor diode tuning. It works great but the diodes change capacitance in a non-linear fashion. So my frequency dial starts out at 14.000, and the intervals between tick marks becomes smaller and smaller, all the way up to 14.090. I make the dial nicer looking by changing the way I mark it up. At the low end I mark every kHz and put a bigger mark on the 5’s. When that starts to scrunch I skip to putting in marks for every even kHz and the 5. At the highest end I just have 10’s and 5’s marked (between .060 and .090).
In the old days they used air-variable capacitors for tuning (a “bread slicer”), and they managed the tuning rate by making the fins of the capacitor a strange ‘D’ shape. The right shape would provide linear capacitance change over the range of motion of the capacitor elements.
So, I’ve been pondering how to linearize the varactor capacitance curve. Really haven’t found a solution yet. (I probably never will, seeing as how there have been a lot of people smarter than me cover this ground multiple times… as part of their jobs… over the years.)
But I thought it was a interesting problem. I’ll let you know if I think of anything.
One reply on “Varactor diode : needed improvement”
I’ve almost given up on the idea of combining capacative elements as a means of producing a more linear response.
Usually varactors are used with a fixed voltage source and a potentiometer. So one method would be to change the potentiometer to the curve necessary to give constant angle-change for given capacitance change. The guts of the potentiometer resistance mechanism could be changed. Or, by using some kind of strangely shaped wheels or gears, the angular movement could be given the right variety.
I remember seeing an article in a recent SPRAT newsletter where a guy divided the overall voltage curve into discrete values, then produce those values with a microprocessor. Problem with that solution is that you get a fixed selection of values.