In a standard LC VCO, the #1 noise contributors are the cross-coupled pair transistors.  After that, it is the tail-curent transistor, and multiple papers have talked about adding a tail capacitor, or a combination of L and C to filter out the higher harmonics at that node.  Along the same line, a VCO in a large IC system usually runs off an LDO instead of an "ideal" supply, and I believe that the effect of noise up-conversion from the LDO should be similar to the noise up-conversion of the tail current transistor noise, i.e. nonlinearity -> AM-to-FM conversion.  I am seeing in simulation that the LDO is a significant noise contributor, but since I have not seen papers talking about this, I assume there must be some flaw in my thinking or my LDO design.  Has any one seen this problem before? Note: I'm interested in very close-in phase nosie (e.g. ~1kHz offset).  Thanks.

LDO coud be a bigger contributer. Ther are some tricks:

1. The AM-PM coefficient could be lowered by inpecting which nonlinear device or diode caps are sensistive to the VCO supply. Sometimes is useful to add a stronger nonlinear cap over bias diode to compensate.

2. The typical bandgap used as reference multiplies the noise voltage of the kT-generator by 15-20. Because the low frequency noise is rejected sometimes a lowpass filter is inserted between the bandgap and the LDO regulator.

3. The next approach is to use a less noisy reference which has still good supply rejection but will change with temperature. A diode voltage multiplied reference is negative within temperature but if your center frequency margin are not violated or you use continious recalibration it makes sense.

If the papers on the topic are coming from MTT or SSC IEEE journals, I expect the problem will not be mentioned. Why? A lot of journal publications are academic research and many ignore non-ideal factors outside the immediate circuit analysis being published.

That's just the reality of the matter, sad to say, but IEEE journals tend to suppress the practical aspect of the problem. (and yes I DO review for these journals, so been there, and seen this before.)

Jerry