[time-nuts] Close in phase noise of microwave VCOs
john at miles.io
Thu Jun 18 18:29:26 EDT 2015
> When you go very close in on something like a VCO, you get much higher phase
> than we normally worry about. Some of the “assumptions” that underly the
> are no longer true. Small angle of modulation is one, but there are a few others.
That's a good point as well. L(f) isn't defined in situations where the noise is strong enough to appear in higher-order Bessel sidebands In these contexts it's better to think in terms of residual FM rather than phase noise.
See http://www.ke5fx.com/stellex.htm for one interesting case where a noisy X-band source is returning questionable results. I measured it three different ways -- with a spectrum analyzer, an HP 3048A quadrature PLL, and a TimePod with a downconverter. I'm still not sure why the 3048A gave different results in the 100 Hz - 1 kHz region relative to the other two analyzers. I'd expect a small-angle violation to result in an artificially high noise plot but the opposite seems to be occurring. I keep meaning to go back and re-run those tests with more careful attention paid to the reference signal level and a few other potential confounding factors. Also need to run a discriminator measurement on the 3048A to see how it compares to the PLL result.
-- john, KE5FX
Miles Design LLC
More information about the time-nuts