What I had to do to stabilize a VFO...

by John Seboldt, K0JD, Minneapolis, MN

New document, 12/26/96

From the basic homebrew literature (QRP Classics, DeMaw's notebooks, various ARRL Handbook projects, etc), I could never get VFO stability that really satisfied me.

What's my standard? As a musician with absolute pitch, I listen at g4 (c3 being middle "c"), 784 Hz in the equal-tempered reckoning. A half-step higher is 830.6 Hz, a half step lower is 740 Hz. Since I can easily hear these approximately 40 Hz increments over the course of a half hour QSO, I would hope for something less than that -- say, 40 Hz or less per HOUR, then the typical half-hour QSO amounts to about a quarter tone, which if I don't listen too hard doesn't bother me.

With a little deeper reading in the amateur literature, you will eventually find more precise material on why VFO's drift, and details on how to apply temperature compensation. An excellent recent example is in Dec. 1993 QST, "Measuring and Compensating Oscillator Frequency Drift" by Wes Hayward, W7ZOI. Of special interest is the statement on page 41: "There is no fundamental thermal stability difference between a Hartley or a Colpitts oscillator, or any other topology I have tried." So much for the building of many oscillator circuits in my stability pursuit! Apparently LC component stability is the major key.

I got less than that with the circuit below, after a bit of temperature compensation. I'll detail my circuit, knowing full well your mileage will vary, then describe some drift sources.

[My VFO schematic]

Buffers are not shown. Just go to the homebrew books. I have one FET source follower, followed by a MOSFET amp, both dead-bugged onto the side of the VFO shield box.

Granted, there are ambiguities in the values of the components. I never measured the trimmer or tuning capacitor values, for example. And there is no certainty of the precise temperature characteristic of the N330 cap -- it came from a Radio Shack assortment! The tuning diode, providing about +- 3 kHz range, is ancient, from the days when Radio Shack sold such things! On the other hand, most capacitors are are Panasonic NP0 chip caps, taken from one of those prototyping kits available from DigiKey.

The coil is about 25 turns #28 wire on a T-44-6 core, tapped at about 5 turns. I'm too lazy to open my VFO and count it for you -- besides, you'll probably have to tweak it yourself with your available components. Once the turns are squeezed to the right place to adjust the tuning range, a bit of Q dope holds them in place.

This is built on a one-sided etched board, which fits neatly into the PC board shield box and is soldered right to the stator pins of the particular tuning capacitor, with just a little bead of solder to tack it to ground.

A few perspectives on drift

When I started with the Lewallen "Optimized QRP Transceiver" VFO, it never really settled down for me in various iterations. It always kept drifting down, down, down about 100 Hz/hour, never stopping, even when using brand new NP0 Panasonic chip capacitors. Well, at least this is as it should be -- with a T-44-6 (yellow) inductor core, which has a positive temperature coefficient of inductance (that is, greater inductance, therefore lower frequency, as temperature increases).

If you have an RIT tuning diode, you have another source of drift. The tuning diode itself, if limited to a few kHz of tuning range, will not do much, but it is still another positive temperature coefficient device -- positive temperature coefficient of capacitance, so again more capacitance, therefore lower frequency, as temperature increases.

And what about the circuits used to drive the RIT diode? Especially if there are more semiconductor junctions? Lots of variables here. Interestingly, a series diode in the control voltage line will tend to compensate for the tuning diode's temperature characteristic, but not precisely. (With increasing temperature, the diode's DC voltage drop decreases, therefore increasing the output voltage. This reduces the tuning diode's capacitance, therefore increasing the frequency.)

What to do? Temperature compensation. And with all those variables, you have to do it after the rig is pretty much finished -- open it up, change a capacitor, close it up, let it cool (overnight or at least an hour), test it, and repeat!

If you're really scientific, you build some kind of box with an internal thermometer to track drift with temperature, as described in that Wes Hayward QST article.

The lazy way out? I just aimed for reducing the warmup and long-term creep, with a quick check of temperature drift by putting my shack lamp a few inches from the unit for several minutes!

The result...

Here is a rough measurement of this 7 MHz VFO when multiplied to 14 MHz. Divide by 2 for actual drift at the fundamental. For the first 5 minutes, the drift was downward about 100 Hz. It then crept back upward, reaching the starting point after another 30 minutes. In another hour, it have moved up about 40 Hz, in the next hour up about 20 Hz, in the next hour up another 12 Hz. Finally I had a VFO that at least started settling down!

After stabilizing for 3 hours, I left it sit overnight. In 7 hours, it crept up 32 Hz -- an average of 4 Hz/hour. Not bad, eh?

Back to main doc