[time-nuts] PM-to-AM noise conversion (was RE: New Question onHP3048A Phase Noise Test Set)

John Miles jmiles at pop.net
Thu Jan 17 22:11:37 UTC 2008


> The phase angle between the USB and LSB noise components is random when
> translated to baseband so when averaged over time the resultant
> amplitude is the same as if one just added the powers of the 2 components.
> When the 2 sidebands are coherent the phase shift between them is fixed
> so that their amplitudes add (vectorially).
> The phase angle between the LSB and USB components when translated to
> baseband depends on the modulation mechanism creating them.
> With both LSB and USB components translated to baseband one has no
> additional information so that some "reasonable" assumption has
> to be made.

Ah, I didn't realize that the two sidebands weren't mirror images.  That
could make sense.

It does seem to make a big differene where the spur came from.  My earlier
test used the second FM lobe from the 8640B's internal 1-kHz oscillator (the
first lobe was driven offscreen).  I just repeated the test with an external
30-kHz source, with the amplitude turned way down to allow me to look at the
first lobe.  In that test (green/red traces) the 11729C and 8566
measurements were only a couple dB apart:

http://www.thegleam.com/ke5fx/11_MHz.gif

I then used a crystal filter to lop off the lower sideband from the FM'ed
signal generator.  As expected this had no effect on the amplitude measured
by the spectrum analyzer directly, but the spur levels on the 11729C with
the SSB filter in place are now about 8 dB too low, instead of 6 dB too high
as the DSB measurement was before.

Really scratching my head now.  I understand why the SSB filter makes a
difference, because the mirror image of the FM spur is coherent.  I don't
understand why the 11729C's view of this DSB measurement (11.5 MHz, 30 kHz,
first lobe) is closer to 'correct', while the earlier one (20 MHz, 1 kHz,
second lobe) was ~6 dB too high...

-- john, KE5FX






More information about the Time-nuts_lists.febo.com mailing list