[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 16:09:39 EST 2008


I see what you mean, Bruce, but why are the two sidebands considered
incoherent, just because they're noise?  Since the IF is 0 Hz, the other
sideband appearing at the analyzer input jack is a folded image of the same
noise spectrum, right?  Given that, shouldn't the SSB correction be -6 dB
rather than -3 dB?

Here's what I get when I deliberately run a PN trace of an FM'ed signal:
http://www.thegleam.com/ke5fx/spurs_fm.gif (signal being measured)
http://www.thegleam.com/ke5fx/spurs_pn.gif (result from PN software)

These were taken at 100 Hz RBW with SA noise correction applied, so the
spurs in the PN plot are really 18 dB higher than shown.  Given that, the PN
plot shows the correct spur amplitude, -60 dBc, when measured on the
analyzer directly (magenta trace).

The same spur measured via the 11729C appears at -54 dBc, or 6 dB too
*high*.  So Martyn's 3048A is reading spurs 6 dB too low, and I'm off by the
same amount in the other direction, for some reason...

-- john, KE5FX


> The explanation is very simple.
> The 3048 calculates L(f) (a single sideband measure) from the mixer
> output where contributions from both sidebands are combined.
> For incoherent (noise) components the powers from the 2 sidebands add.
> For coherent components the amplitudes add.
>
> Thus to correct the DSB measurements to SSB 3 dB is subtracted from all
> incoherent (noise components) and 6dB is subtracted from all coherent
> components.
> (By definition L(f) is symmetric about the carrier - although this isnt
> necessarily true for large offsets).
>
> Since the input to the 3048 includes a single frequency SSB spur,
> subtracting 6dB produces a calculate SSB spur amplitude 6dB lower than
> that measured by the spectrum analyser.
> If equal amplitude (and frequency) spurs symmetric about the carrier
> were injected their amplitude would be calculated correctly by the 3048.
>
> Bruce
>
>





More information about the time-nuts mailing list