Where is the phase shift on this oscillator?

Question

I didn't think the oscillator below could work, but I built it, and it does:

I used 10 uH for the inductor, 2.7 nF for the capacitor, and got oscillations around 8 MHz.

The feedback loop is a series resonant circuit, which should not have any phase shift at resonance. And, indeed, the resonant frequency is around 1 MHz, and that's not where it oscillates.

But it does oscillate at a much higher frequency! Where does the phase shift come in?

Also very odd: The oscillation frequency changed when I changed L and C, but in very unpredictable ways. Dividing or multiplying C by 2 might not change the frequency at all (even after power cycle). Changing L had a more immediate affect.

And: The wave was not a sine wave, but a sine with some weird "humps" at the top (and much smoother at the valleys).

And: Simulation on CircuitLab shows no oscillation whatsoever!

I thought that perhaps the series LC is phase shifting at a second, higher frequency, not its resonant frequency. But, from my analysis, a series LC will never shift phase by more than 90 degrees.

How does this oscillator oscillate? At what frequency? Where is the phase shift to compensate for the 180 degrees of the common emitter?

Answer 1

There's parasitic capacitance in various places of this circuit that can easily make it oscillate, see orange C's in altered drawing:

Obviously we get oscillation if we have the \$180^\circ\$ phase shift in the transistor and in addition another \$180^\circ\$ delay. And we can get delay because:

1. The parasitic C in the output choke can make the collector load capacitive (if we are above the self-resonance of this \$470\mu\$H choke, which is not specified in the description). This gives delay between the output current coming from the collector and the resulting output voltage.
2. The feedback network becomes inductive, as you explain, above \$1\$ MHz approximately. So it gives delay between the output voltage and the resulting feedback current.
3. The transistor's parasitic input C will give delay between the feedback current and the resulting input voltage.

Together these three mechanisms can at high enough frequency give up to \$270^\circ\$ delay, more than the \$180\$ we need. So oscillation will take place at some frequency where not all this potential phase delay is realized. Of course this makes it a bit ill-defined at which frequency a circuit like this will oscillate...

Answer 2

Here is a simulation microcap v12 that shows that this circuit can "oscillate" ... and at what frequency ...
Note that I added some "parasitic" components.
Serial resistor with inductors, and "parasitic" capacitor with the choke inductor (not well designed, load not added ...).

I made a "stability" test and concluded.
Oscillation at a higher frequency (15 MHz verified with stability diagrams).

And the waveform of the oscillation (voltage pulse) ...