Back to the beginning
My first test oscillator was a Colpitts. Right away I ran into a common problem: hysterisis. Once started, the oscillator would continue to run at less voltage. Starting and stopping were both abrupt. There followed the usual messing about with biasing, before I could begin to test it as a regenerative detector.
Experimental Colpitts crystal oscillator
I spent more time on this experiment than I wanted to. Thinking to use the upper feedback capacitor as a regeneration throttle control, I removed the fixed cap. The oscillator still ran! Removing the lower capacitor had the same effect. So much for that idea. I was able to see an input signal, applied to the source through a 10:1 step-up transformer, modulating the oscillator. Again, though, I didn't see the large increase in gain I was expecting.
About this time, AA1TJ sent me a link to a VXO-regen he built: the AA1TJ "Xtaflex". Based on the sound recording Mike provides, his receiver has plenty of gain and exhibits all the benefits of a crystal-controlled regen. Its tuning range is limited, however, and Mike used an obsolete transistor, so I was unable to duplicate it.
Consulting the elders
At the end of a long day at the bench, I like to settle down with the QST archives. A search for "VXO" turned up several articles, and they referred me to a few more. In the early days of VXOs the Pierce oscillator was king, and dual-section variable caps were the order of the day. Here's a simplified version of a design by Doug DeMaw.5
An early FET VXO design by Doug DeMaw
The first application of this modified-Pierce scheme I found was by W3BWK.6 (This article is a good theoretical introduction to the VXO and is recommended.) C1A and C1B are the dual variable capacitor (100-pF per section was common). Using more series L should provide good results with only C1B. The Pierce attracted me because it has a smaller parts count than the Colpitts, and the feedback arrangement should make it easier to inject the signal diretly into the FET gate. So I built a modified-Pierce VXO.
Again, I ran into the hysterisis hurdle. After burning a couple of hours trying to tweak the Pierce into starting and stopping gently, I recalled how smoothly my Colpitts worked. What I needed was a better way to test what I had.
Being a CW operator, I naturally think of regen receivers that oscillate. But the true measure of a regen is how it behaves just at the verge of oscillation, where stage gain and selectivity are at their best. W3LW realized this as far back as 1933, and built a regen with a separate tunable heterodyne oscillator for CW reception.7
Just for fun, here's a Colpitts regenerative-detector circuit from 1923. It was published in "The Radio Experimenters' Guide," put out by the Newark (NJ) Sunday Call.
Colpitts Regenerative Detector, 1923
Back to the Colpitts
I went through a process of analyzing the selectivity and relative gain of the stage below the point of oscillation. A Hewlett Packard 312B selective voltmeter lightly coupled to the detector output let me inspect the relative gain and selectivity at different supply voltages (I adjust supply voltage to control regeneration). While I don't have hard data, I was able to empirically determine that the stage does operate like a regenerative detector, and man, does using crystals give you a sharp peak in selectivity!
The last remaining problem was that the added reactance of the T50-2 transformer was causing the oscillator level to vary greatly over its tuning range, and drop out at the high end. This is where we came in! The solution was easy: a transformer with fewer turns on the secondary. Here's the final circuit:
KR1S VXO Regen, final version
T1 ended up on a T37-2 core, 2 turns on the primary, 10 turns on the secondary. Because the FET's source resistor is so heavily bypassed, I got better results using the drain as the output port. If you build this circuit, experiment with different values of source resistance. In general, the lowest resistance that gives reliable oscillator starting will probably produce the smoothest operation.
What about the rest?
What's different about this receiver is the VXO detector, even though there isn't much new there, either. The regenerative stage is powered from an LM317LZ adjustable regulator, which is configured to provide 2 to 3 V, for smooth regeneration control. The front-end is a standard aperiodic common-gate FET. Instead of direct-to-audio conversion, I'm using an infinite-impedance detector by Felix, VK4FUQ, from the Elliott Sound Products site, "AM Radio - Approaches to high fidelity AM broadcast band reception." The detector drives an off-board preamp designed by Terry Ritter, which drives an LM380.
How's it work?
Just fine, thanks. It covers the important part of 40 meters for CW QRP purposes, 7019-7038 kHz. There is no "pulling in" on strong signals, and stability is excellent. The receiver should be scaleable to any HF band for which fundamental-mode crystals are available. I plan to use a similar design, without the VXO, as a selective IF amp in a simple superhet receiver.
Increasing the supply voltage from 2.5 to 6 V causes very slight frequency shift, so it should be possible to run it a little harder, to drive a transmitter. Unfortunately, output level does vary across the range, though the change is most noticeable at the high end. It should be possible to level the output in a Class C buffer. Then, using steering diodes and the usual T/R tricks with MOSFETs, a transceiver isn't out of the question. The regen stage will need careful shielding and bypssing, and transmitted RF will have to be kept away from the preamp port.
This was a fun little project. The dead ends and false starts taught me a lot about FETs, regens and VXOs. Thanks to Mike, AA1TJ, the guys whose circuits I copied, the folks at Yahoo regenrx, and all the regen pioneers who inspired me.
1 Wildenheim, "Regenerative Detectors And A Wideband Amplifier For Experimenters," Ham Radio, March 1970, p. 61.
2 Kitchin and Murphy, "An Ultra Simple W1AW Receiver," QST, May 1997, p. 34.
3 Grammer, "Rationalizing the Autodyne," QST, January 1933, p. 11.
4 Kitchin, "High Performance Regenerative Receiver Design," QEX, November/December 1998, p. 24.
5 DeMaw, "Some Practical Aspects of VXO Design," QST, May 1972, p. 11.
6 Shall, "VXO — A Practical Variable Crystal Oscillator," QST, January 1958, p. 11.
7 Robinson, "Regenerative Detectors," QST, February 1933, p. 30.