Amplitude noise in LC oscillators

[Davor]

You may try it, and I'd be glad to hear about your results.
However, if you don't apply some sort of delay compensation, your circuit will remain unstable. Regarding the noise generation, this anti-Colpitts topology seem to be the closest to the ideal as described in literature, because the current-intensive angle is very small, and hence the least prone to phase noise generation. Provided the delay is compensated, of course.

[Davor]

BTW, the ESR of "normal" capacitors is quite small, so no problem omitting it in simulation. ESR values of 10nF capacitors are typically under 1 ohm.

[Davor]

A little bump...
My oscillator works, and amplitude is rock solid.

[Teleno]

A little bump...
My oscillator works, and amplitude is rock solid.

It may seem so, but if you look at my circuit you'll notice a current sink subtracting all the excess ampltude and passing only the top 3V of the peak to the amplifier. The flicker noise I was looking at is there. Set the vertical scale of your osciloscope to 2 or 3V per division, focus it on the peaks of the wave and tell me how much flicker you see there.

[Davor]

None
Mine is avoiding 1/f sources completely - it is digital, yet resonant. Wonders of PLL at work.

[Teleno]

None
Mine is avoiding 1/f sources completely - it is digital, yet resonant. Wonders of PLL at work.

I'm talking 100's of millivolts on a sine that's peaking at 40V. Double checked?

[Davor]

Yup.

[Teleno]

I believe I've got the answer to this kind of flicker noise.

There are so called "pule bias oscillators" where the energy lost in the LC tank is restored by means of a short constant current pulse applied once per period.

I have found that the flicker in the usual oscillatpor is caused by differences in the phase of the energy feedback from one cycle to another. A precise and repeatable zero-crossing detector used to determine the moment of switching greatly reduces amplitude flicker.

I'll try to post an example simulation soon.

[Davor]

Yup.

I know of a paper that says the very same thing, but their solution is completely opposite.
The very problem of phase noise, as opposed to amplitude noise, is that ... it is the same noise. The source of both phase and amplitude noise is the active component flicker noise, which is being mixed to the carrier frequency. The source of flicker noise is thus a consequence of an active component running in a linear regime. So either make it a switch, so it does not run in a linear regime which sources the flicker noise to be mixed, or make it super linear by a local feedback so that the flicker noise becomes weaker, as in application of a degeneration resistor which was a go-to solution for low phase noise oscillators until recently.

My solution works with a switch, and an independent zero crossing detector.

Knowing that both phase and amplitude noise stem from the same source is helpful for us, as there is a vast body of research in minimising the phase noise. What works for reduction of phase noise, also works with amplitude.

[Teleno]

I've found a solution as a pulse-biased oscillator. https://www.geotech1.com/forums/show...598#post264598