[time-nuts] Is this a correct way to measure at phase noise?

Erik Kaashoek erik at kaashoek.com
Fri Mar 24 18:24:56 UTC 2023


Hi All,
I'm trying to understand if the tinyPFA, with a yet to be release FW 
update, can be used to measure close in (0.1 to 390 Hz) phase noise.

For testing I'm using a 16 bit DDS signal generator with both outputs 
set to 10MHz and one of the outputs is phase modulated with white noise 
at 6 degrees width which I'm assuming ,gives -110dBc/Hz. This was 
confirmed by using a calibrated SA to measure the noise level
The other output uses a 0.6 degrees phase modulation using a sine wave 
at 10 Hz to get an absolute power level reference
Both inputs are connected to the tinyPFA so it can measure the phase 
noise between the two generator output signals.
With the tinySA using 0.01 s samples per bucket and an the default FFT 
length of 1024 I get the noise from the generator at -110dBc so this 
seems to be the correct sample length for /Hz display of phase noise
Setting the sample length to 0.1 s the sine modulated signal stay's at 
the same level and the noise moves to -120dBc so a factor 10 more 
samples drops the noise 10dB
Setting the sample length to 0.001 s the noise moves, as expected, to 
-100dBc while the sine modulated signal again stay's at the same level.

When I'm not making an error this seems to imply that I can measure 
phase noise with different sample lengths to look at different 
frequencies as long as I add the correct correction from the sample length.

Here is an instrument noise floor measurement using two identical inputs 
with 0.1 s sample length, so you have to add 10dB to get the measured 
noise level, giving a -125dBc/Hz noise floor at 1 Hz offset, not very 
good but usable for (very) close in phase noise or when measuring PLL's


and this is a measurement of an OCXO using a low phase noise DCXO as 
reference with the same settings so the 0.1Hz offset phase noise appears 
to be -73 dBc/Hz and the 1 Hz offset phase noise around 95 dBc/Hz



I hope someone can either confirm my measurements or explain what I am 
doing wrong.




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