How could I sim. VCO + Mixer + LNA in spectre ?

Dear all,

I have a question about the spectre simulation method of the VCO + Mixer
+ LNA. I hope anyone could help me.

Actually, I have successfully simulated the VCO and Mixer+LNA
separatedly using the spectre. Which is using the special function, such as
pss, pxf or pnoise etc, to simulated the phase noise, oscillating frequency
in the VCO and 1dB compression point, iip3 or power gain in the Mixer+LNA.

However, how could I simulate the overall circuit, VCO+Mixer+LNA, to get
the simulation result such as 1dB compression point, iip3, noise figure
etc.?

Best Regards,

Charles

 

You're going to have difficulty simulating the VCO+Mixer+LNA with a large signal
input to the LNA. That's because whilst PSS can run in both driven and
non-driven (i.e. oscillator) modes, in both cases it needs to simulate over 1 period
of the beat frequency of all the frequencies. With a driven input, and a free-running
oscillator, the beat frequency is chaotic - a small change in the oscillator frequency
can mean a huge change in the beat frequency - so you'd never get it to lock
onto the PSS fundamental.

You can simulate a small signal input to the LNA (because you then only have
the oscillator frequency to contend with). However, that won't enable you to measure
1dB compression point or IIP3 (but you could measure the noise figure).

So, the best bet is to characterize the VCO, and then replace it with a
behavioural model of an oscillator (for example, osc from rfLib). Then the
circuit becomes driven, and you can then simulate it with either PSS or QPSS
(depending on how the frequencies turn out).

Regards,

Andrew.