The 6088 submini tube has a 1.25V 20mA filament (25mW) and pentode mode gm=650uS while triode mode gm=860uS. One reader very kindly sent me two of these nice tubes, CK6088 by Raytheon, to experiment with!

The 6088 seems from the datasheet to be more suitable for an HF oscillator than the CK512 tried previously (higher gain). I tried Andy G4OEP's XFY43 circuit which at first did not work. With a higher supply voltage of 54V the circuit sprung to life, producing 4V peak-peak output. So at long last, success building an oscillator with a submini-tube! The oscillator stopped oscillating below 34V supply voltage.

I did some further optimisation empirically, to improve the performance of this oscillator. I was able to get it to oscillate down to 27V supply but no lower. The process of optimisation was largely a matter of trial and error - using two variable resistors and three variable capacitors, and observing the effects of changing everything. This photo (CLICK for bigger version) shows how things look during optimisation!

The final circuit diagram is:

Next to measure the performance - I put the oscillator in the box (from previous VFO experiments). Note the close-up photos below showing how a surface-mount 8-pin chip version of the TMP36 temperature sensor was soldered to a tiny piece of board, and sandwiched under the 6088 tube. Another was soldered to the board near the inductor. The third sensor was soldered to the outside of the box, supposedly to measure "shack temperature" but in fact, obviously (stupid me) actually measuring the temperature on the outside wall of the box, feeling the slight heat coming through the wall.

The following charts show the output amplitude variation with supply voltage, and the output amplitude vs oscillator frequency (using the tuning capacitor and inductor shown). The BLUE line was the original circuit using the component values Andy G4OEP used on his XFY43 tube. The RED line is my circuit above, with slight modifications which I found improved the performance.

Now connecting the oscillator to the Arduino frequency-counting data logger, I got the following data over 1 hour duration from switch-on. Note that no attempt at all has been made to balance temperature coefficient of the components.

The frequency drift is about 3kHz but the interesting part of this, is that there is apparently no initial down drift that was observed on all the previous oscillators. The reasoning was that the initial downwards drift is due to the valve itself stabilising as it reaches thermal equilibrium, followed by an upwards drift with temperature. Whilst the temperature drift can be attacked using the right temperature coefficient components and lots of patience, the initial valve-drift cannot. So it is extremely promising that here, there seems to be no initial drift. Unless, of course, the valve drift is in the same direction as the subsequent temperature drift.