How Low Can We Go, With a VXO?

The announcement of a building contest for Pacificon, 2002, set off a chain of experiments in my shack that, so far, has produced a very nice 10 Meter transceiver, using a 20 Meter VXO as its signal source, two entries to the VXO contest that tuned from above the basic crystal frequencies to frequencies well below the bottom of their respective bands, and several ideas for future QRP projects.

Since I had somewhat of a base of information to build on, having done some VXO experimentation with the SMK-1 and Rock-Mite circuits, I was eager to see just how far a VXO could be pulled from the actual crystal frequency and still function as a VXO.

In addition, I wanted to learn more about how a VXO with a wide tuning range could be used in either Direct Conversion or Superhet transceiver schemes.

In the contest, the rules required that 7.040 and 14.060 MHz crystals be used. Since we would logically want to retain the ability to operate on those frequencies, as they are the recognized QRP calling frequencies for their respective bands, I wanted to end up with VXOs that tuned from a few KHz above the actual crystal frequencies to well below them, at least down past the low ends of the respective bands.


The ubiquitous 2N2222 NPN transistor was employed in a basic Colpitts oscillator configuration. Selected values of capacitance were used to pull the VXO frequency up, and a series of surface mount inductors were used to pull the VXO frequency down.

The key ingredients for achieving a wide tuning range were: The use of paralleled crystals of like frequency; The use of a switching scheme to select the different values of capacitance and inductance used to “pull” the crystal oscillator through several tuning ranges; And, the selection of appropriate feedback capacitance values not only for the band of choice, but to allow for providing a frequency offset scheme.


What was learned from the experiments included the following:

A VXO using two crystals of like frequency in parallel could be pulled a very long way from the actual crystal frequency. As examples of how far, in the Pacificon VXO Contest the 20 Meter VXO described below was judged as having a tuning range of 178 KHz, from 14066 down to 13888 KHz, tuning well below the bottom of the band. The 40 Meter VXO did even better, tuning over a 309 KHz range, from 7044 KHz down to 6735 KHz, tuning even further below the bottom of the band before losing its stability.

The VXOs could be pulled further with a set of relatively small value inductors connected in series than it could with one inductor of the total value.

If one wanted to be able to tune the VXO from above the crystal frequency to well below it, the simplest and most practical way to accomplish this was to provide a switching system that selected a series of capacitances and inductances, in turn, in order to provide several tuning ranges.

Because of the non-linear effect a tuning diode and added capacitance or inductance has on the tuning range of a VXO, a simple switched capacitive offset circuit like those familiar to us in the Pixie and other simple transceivers could not be used.

One set value of capacitance switched into and out of the circuit would provide a useful frequency offset over part of the whole tuning range, but the amount of offset would change as the VXO was tuned, both within a particular range and over the entire tuning range, getting larger as the voltage on the main tuning diode decreased, and getting larger as the oscillator was pulled further from the actual crystal frequency.

This effect holds true whether a simple switched capacitance offset scheme or an RIT circuit is employed. As a practical matter, this means that offset capacitor and RIT tuning diode capacitance values are based more on how far the VXO tuning range is from the actual crystal frequency than the frequency of operation itself.

If one didn’t mind not being able to operate on the exact crystal frequency, one could achieve a very useful and relatively wide tuning range without having to utilize a switching scheme. Such as being able to tune from 14.000 to 14.044 MHz, using a pair of 14.060 MHz crystals. Or, achieving a somewhat lesser tuning range in the low end of 40 Meters with a pair of 7.040 MHz crystals.

RIT could be incorporated into a Colpitts oscillator circuit used as a VXO, by applying it to the emitter of the Colpitts oscillator and by selecting both an offset capacitor and tuning diode capacitance value which would provide at least a minimally acceptable amount of offset at the highest frequency the VXO tunes, where it is hardest to pull.

As stated above, the amount of offset provided by the RIT increases as the crystals are pulled further away from their actual frequency, and as the oscillator is pulled lower in each tuning range.

As an example, the RIT in the 14 MHz VXO used as the basis for a 10 Meter DC Transceiver provides an offset of 700 Hz at the high end of the tuning range, and about 2.7 KHz at the low end. The same reception tone can be heard throughout the entire tuning range by first zero beating the received station with the use of the “Spot” switch, and then adjusting the RIT knob for the tone desired.

If either RIT or capacitive offset is applied to the emitter of the Colpitts oscillator, the value of the feedback capacitor from the emitter to ground should be reduced in value, so that the series capacitance of the offset capacitor and tuning diode used in the RIT circuit, in parallel with this capacitor, will return the total capacitance to near the original value.

Even though the continuous output of a VXO may remain stable in frequency when the oscillator is pulled a long way from the actual crystal frequency, it is much more susceptible to pulling from external sources, such as changes in the load, or from a signal being radiated back into it, the further away from the crystal frequency it is tuned. Therefore, shielding should be used if the rig is to be operated on the fundamental frequency of the VXO.

In the 20 Meter VXO entry, there were gaps in coverage of the tuning range, even when the surface mount inductors used were mounted right on the back of the DIP switch used to select them. In the 40 Meter VXO, there were no gaps at all, and there was a small overlap on each tuning range. This difference is possibly caused by the greater effect stray impedances would have on the higher frequency circuit. This fact made it necessary to use a capacitive offset scheme in the 20 Meter VXO entry, so that the entire range would be covered, with no gaps.

The Circuits

Here is the circuit for the 20 Meter VXO Contest entry. In this circuit, no RIT was employed, but a capacitive offset was used so as to provide coverage of all included frequencies.

The frequency ranges shown were those obtained in the shack at NB6M with a pair of 14.060 crystals. Performance with one of the parallel crystals replaced by the “control” crystal in the judging at Pacificon was very close to the same.

Here is the circuit of a practical 20 Meter VXO, with one tuning range, 14000.5 to 14044.0 KHz, and with RIT. This very stable VXO was used as the heart of a 10 Meter Direct Conversion transceiver.

There is no pulling of the VXO at all on transmit, no chirp, and no drift. However, if this VXO is to be used in a 20 Meter transceiver, the VXO, with its RIT and buffer amplifier circuits, would need to be totally shielded.

The same buffer amp that was used in the 20 Meter VXO Contest entry could be used to follow this VXO. It is designed to put out a little over 2 Volts RMS into a 50 Ohm load.

Here is the circuit for the 40 Meter VXO entry:

The RIT in this circuit works very nicely, and provides a very useful amount of offset throughout the tuning range. The judges in the contest found this VXO’s tuning range to be 309 KHz, from 7044 down to 6735 KHz. Although it pulled further than that, it lost its stability at that point.

This VXO has been used to drive an older, tube type transmitter with good results, by installing a half-watt, 51 Ohm resistor from the output to ground, inside the VXO case, so that the buffer amplifier was provided with the proper load.

Here is the circuit for an experimental 80 Meter VXO that uses a pair of 3.579 MHz crystals. It tunes from 3545 to 3565 KHz, which provides nice coverage of the 3560 KHz QRP calling frequency.

Quite a bit of “leverage” is being used here, in order to pull these color-burst crystals down enough to cover a nice, 20 KHz, more or less, frequency range around the 80 Meter QRP frequency of 3560 KHz.

The RIT circuit, with the values shown, does provide a very useful amount of offset throughout the tuning range. However, more experimentation is needed, both with the offset capacitor value and the resistor values around the RIT pot. More info on this VXO will follow as it is developed into a repeatable, practical circuit.

Parting Thoughts

The results of these experiments, and the experiments of the several others who have been and are working on VXOs, show that, even in circuits that we have taken for granted for years, new and different techniques can and do produce remarkable results. Most of us, myself included, had accepted and were happy with the idea that a VXO was good for only a few KHz of tuning range. Who would have thought of being able to build a VXO with a wide enough tuning range to cover a large portion of the CW band, or of applying RIT to it?

Although most of the very wide tuning ranges of the two VXO Contest entries described here was wasted, outside our allowed frequencies, and, in fact the further the VXO is pulled from the crystal frequency, the more non-linear the tuning is, they do show that very good results can be had if we start with a pair of crystals on a frequency a few KHz above the upper limit of the desired tuning range, and then apply similar techniques to those described above in order to pull the oscillator a reasonable distance.

The good news is that the limits of “a reasonable distance” have changed. They have increased by a large margin.

The very practical 14 MHz VXO circuit above, which tunes from 14000.5 to 14044 KHz, is a good example of what is an easily repeatable and very useful VXO circuit.

The thing is, it can easily tune several more KHz below the bottom of the band, which tells us that if we had a pair of crystals on about 14085 KHz or so, we could build a stable VXO, with RIT, that would cover the lower 65 KHz of the band in one tuning range.

Likewise, I believe that a stable VXO, with RIT, could be built that would tune from 7000 to 7050 KHz, or higher, in one tuning range, if we had a pair of crystals on about 7070 KHz or so. These ideas are worth pursuing, because of the frequency stability these VXOs provide and their easy duplication.

Currently, experiments are being made with a wide range VXO using a pair of 7.122 MHz crystals. Initial results have shown that a VXO using a pair of these crystals can be made to tune from 7094 down to 7000 KHz. Like the 80 Meter VXO shown above, this one needs further refinement, both in the basic oscillator circuit and in the RIT addition. There will be more on that, later.

The results described above for each of the VXOs were had when two crystals of like frequency were paralleled. What can we do with more than two?

Everything I had read about the so-called “Super VXO”, using paralleled crystals, said that two crystals in parallel gave a pretty good gain in tuning range, but the addition of a third crystal gained nothing more. That was not what was found when three crystals of like frequency were paralleled here. More tuning range was available than with two.

The non-linearity of the tuning of the VXO, getting worse the further away from the actual crystal frequency the oscillator is pulled, would seem to limit the practicality of a very wide range VXO. But, what happens if we apply the tuning voltage to the varactor diode in some non-linear fashion in order to compensate for that?

The point is that many things are left to try and experiment with, and that is how we learn and how we make our hobby even more enjoyable.

Leave a comment

Your email address will not be published. Required fields are marked *