The final boxed version of my dual-6088 Franklin VFO uses the BC221 tuning capacitor and includes a DF96 valve as buffer. This VFO is built on a piece of un-etched copper PCB board, and bolted to the back wall of the aluminium/steel box. The tuning capacitor occupies most of the space in the box. There's a BNC socket for the RF output and 7-pin socket for power supply voltages. The power supply requirements are 50V HT, and 1.25V, 1.25V and 1.4V for the valves.

This is the first drift run, using nothing particularly special as capacitors - just ceramic capacitors. Still not bad, 250Hz down in an hour! The oscillator frequency this time is 5.116MHz.

Now to try to improve the drift, and get the oscillator on the wanted frequency range.

1. Replaced the inductor with a T50-6 with 22 turns of copper wire, as thick as I could fit on the core.

2. Put a 68pF silver mica capacitor in series with the BC221 tuning capacitor, and the series assembly across the inductor.

3. Put fixed 100pF and 150pF polystyrene capacitors (negative temperature coefficient) in parallel with the inductor.

Using this combination: the oscillator tunes from approximately 6.0-6.5MHz with 50 turns of the tuning capacitor, with approximately 7kHz/turn at the bottom end and 12kHz/turn at the top end. Pretty ideal :-)

The drift however, shows too much upward drift. The polystyrene capacitors have a negative temperature coefficient, to cancel the +35ppm temperature coefficient of the T50-6 inductor. But clearly they are over-cancelling, resulting in a drift upwards:

Finally, I replaced the 150pF polystyrene capacitor with a 120pF silver-mica capacitor (slightly positive temperature coefficient). Now the drift is nearly perfect (see below); Down just over 150Hz from switch on, and very little thereafter. In one measurement run I sat and watched the oscillator for 30 minutes, while the frequency stayed within +/-2Hz!