VCO classes of operation

Hi,

I am starting to design a VCO operating at 4.4GHz with a tunning range of 10% and I'm looking into the state-of-art to choose the topology.

I'm seeing a lot of papers presenting VCO's class-a, class-b, class-c and so on, but I couldn't find any literature explaining those classes other than the papers.

Can anyone indicate me a bibliography explaining these topologies, comparing and showing the math description of each one?

Thanks!

The VCOs are genrally A-Class devices.Other classes are used for particular purposes.
Because B-Class and C-Class systems produce unwanted harmonics levels.

MOST VCOs grow the oscillation amplitude until it is clipping at the power supply rails. So in essence, they are ALL Class C.

There are some VCOs (mostly low frequency or VHF) that have a variable attenuator element that keeps the active device from clipping. Famously, Hewlett & Packard started their company making an audio oscillator like that that used a filament lightbulb as a variable resistor to keep the active device in the linear region--thereby reducing harmonic products. You could try that...and keep the active device in a more Class A condition.

There are some papers out there reporting an improved phase noise if you keep the active device more linear. I believe the theory is, if the device is linear, the 1/f noise (audio) will not upconvert to microwave frequencies as readily

biff44 said:

MOST VCOs grow the oscillation amplitude until it is clipping at the power supply rails. So in essence, they are ALL Class C.

Click to expand...

What you're describing is a class A amplifier oscillating until it is overdriven into class AB. A class C amplifier would never ring up in the first place (unless perturbed by a large stimulus), since its small signal gain is near zero.

that is correct. You need enough gain at start up for oscillations to build up. That means class A biasing is the needed topology. But a few milliseconds later, it will be operating as class C (or maybe D or E, depending on your point of view).

biff44 said:

that is correct. You need enough gain at start up for oscillations to build up. That means class A biasing is the needed topology. But a few milliseconds later, it will be operating as class C (or maybe D or E, depending on your point of view).

Click to expand...

You're deciding by watching collector/drain collector but the reality is not as you think.For instance, cross-coupled oscillator's current is purely sinusoidal and they work as Class-A.
In fact, mentioning on the class of the oscillators does not make sense.They are autonomous circuits and everything occurs phenomenal in them so their structures are chaotic, not deterministic.
Therefore talking on their class -in my opinion- has no meaning at all.

biff44 said:

that is correct. You need enough gain at start up for oscillations to build up. That means class A biasing is the needed topology. But a few milliseconds later, it will be operating as class C (or maybe D or E, depending on your point of view).

Click to expand...

Even in the case where the VCO starts clipping and distorting, a class A can't turn into a class C, since it is still biased in saturation (Vgs>Vth), not cutoff. The only way it could operate as class C is if its gate bias is changed by some controller after the oscillation has become self sustaining. My impression of literature on "class C" VCOs suggests that this is how they work.