Hi,

ususally in a GHz vco design, the inductor is on the order of <500ph with very high Q, and the frequency operation is way below the self resonance of the inductor.  

In the case of an on chip low pass LC filter with f3dB ~300MHz, the inductor value is usually >10nH with the self resonance no higher than couple GHz.  Does that mean, I can't trust the simulated AC response highers than the inductor self resonant frequency?

I guess my question is
1.  What's reason for operating inductor below it's resonance for the high Q inductor (such as in the vco case)
2.  Is the reason for #1 applies to the case when you have a big inductor ( low Q) in the range of 10 - 20 nH?

Thanks.
Kelly

Kelly,

In addition to the variation of unwanted parasitic capacitance and hence the self-resonant frequency that vivkr mentioned, there is the case of what is commonly called loaded Q.  If you where to load an ideal (infinite Q) parallel tank with a parallel resistor you will find that the highest loaded Q occurs when the inductor is the smallest and the capacitor is the biggest.  In practice there is not an infinite Q inductor or capacitor and therefore it is not that straight forward.  Also on chip components further obscure this by making the Q of the inductor heavily dependent on the geometry of the metal used and area used to make the best Q comes at a cost.  You have a multi-variable optimization problem that is not easy to explain by saying one way is better than another.  With that said operating the VCO with an inductor at self resonance is risky because if the modelling is incorrect or all parasitics are not accounted for which you should expect to be the case, then there is no varactor voltage that will ever allow you to obtain the desired frequency.

http://rfcooltools.com