[time-nuts] 35601A as stand-in for 11848A ?

John Miles jmiles at pop.net
Sat Jan 5 19:54:23 EST 2008

> Thats why its phase noise floor is relatively high.
> Saturating both mixer ports and using a capacitive IF port termination
> should achieve a lower phase noise floor.
> The tradeoff is that careful calibration of the mixer phase sensitivity
> frequency response needs to be done.

That's a good point; how are you (Said) calibrating the system?

For simplicity I'd drive the mixer's LO port at its full spec and start with
the RF port unsaturated, at (presumably) 0 dBm.

Substitute a signal generator for your DUT at the RF port, with the
amplitude and frequency matching the DUT's.

Then, drop the signal generator's amplitude by an amount exactly equal to
the sum of the LNA gain and LPF insertion loss.

Then, detune the signal generator by 10 kHz in one direction or the other.

How strong is the sideband that you observe on the analyzer?  That's your
'carrier amplitude' for the measurement.  You then subtract the LNA +LPF
gain, 6 dB for the mixer's SSB phase-detector response, 10*log(RBW), and add
a couple dB for the analyzer's filter/detector noise response.  With a 40 dB
LNA, the total calibration equation comes out to about -44-10*log(RBW).
This is what you add to the observed dBm values on the analyzer to obtain
dBc relative to the carrier amplitude observed earlier.  Needless to say,
all this is a lot less error-prone with software to do it for you.  I
seriously doubt I typed it all in from memory correctly. :)

Even an unoptimized setup should have a floor well into the -140 to -150
dBc/Hz range.  The usable floor of the 11729B/C is about -160 dBc/Hz.  Gross
deviations from this mean you're doing something wrong.

-- john, KE5FX

More information about the time-nuts mailing list