[time-nuts] Phase noise measurement (was - no subject)

Adrian rfnuts at arcor.de
Fri Aug 20 14:40:57 UTC 2010


Mark,

you have the following options:

- HP (Agilent) E5052A/B or R&S FSUP Signal Source Analyzer (works for a 
single DUT, though limited to 1 Hz offset, normally useful for 10 Hz up 
to 40 MHz).
- Compare two identical DUT's with a HP 3048A or similar PN test system 
and subtract 3 dB, assuming that the PN characteristic of both DUT's is 
identical.
- Compare 3 similar DUT's  with a HP 3048A and calculate the individual 
PN using the three cornerd hat method.
- set up a cross correlation PN measurement system similar to the E5052A 
and have fun. You will however need two - as good as possible, but 
preferaby not more than 10 dB worse than what you want to measure - 
VCXO's like HP 10811A's...
- you may build your own HP 3048A alike system, but be prepared to 
invest serious money and time, and much more time than you thought in 
the beginning... (if that is what you're after, you'll have the most fun 
you can).
- find someone who has one of the above and talk him into measuring yours.
- search the web for published PN data of the model you have and take 
these as a reference (give or take a few dB).
Btw. do not assume that the phase noise of a disciplined VCXO is the 
same as the VCXO alone.
Also keep the power supply contribution into account that can be 
surprisingly high.
And, the PN of most frequency standards is significantly lower than what 
you can measure with any spectrum analyzer with PN measurement software 
(except for the R&S FSUP of course).

Adrian


Grant Hodgson schrieb:
> Mark
>
> You've come to the right place - well, that is if you want to devote a 
> significant amount of your life in the pursuit of ever-more accurate 
> time and frequency measurements....
>
> If you've only got one source then you need to use the frequency 
> discriminator method (aka delay line method) of phase noise 
> measurement.  Basically you take the output of the source, split it in 
> two, delay one of the signals, re-combine the two and then measure the 
> resultant signal on a base-band spectrum analyser.
>
> There are loads of references to this on the web, which describe the 
> method in more detail, including :-
>
> The Art of phase noise measurement - Dieter Scherer
>
> and
>
> HP Application Note AN270-2
>
> both available from John Miles web site
>
> www.thegleam.com/ke5fx/gpib/pn.htm
>
> The references at the end of these articles, especially the HP ones, 
> are particularly useful.  The operating manual for the HP 11729B or 
> 11729C Carrier Noise Test Set is also highly recommended.
>
> Yes, there's some maths, you need to understand the relationship 
> between phase and frequency measurements, but you don't necessarily 
> need ALL the theory that most of the papers give - don't give up just 
> because of a few differential equations :)
>
> The limitation of the frequency discriminator method is that the noise 
> floor of the measurement system is often worse than the DUT, 
> especially if your DUT is very good, and it's even worse if you're 
> trying to measure close-in noise.  The Sherer article gives a good 
> graph illustrating this. If you're trying to measure the phase noise 
> of the oscillator inside a Tbolt then I don't think that a frequency 
> discriminator will be sensitive enough, although I might be wrong.
>
> Despite what you said, you might want to consider buying an HP 10811 
> oscillator or similar which you could use in a phase detector 
> measurement system which is likely to give superior results.
>
> Hope that helps
>
> regards
>
> Grant
>
> Mark wrote :-
>
> My new GPSDO leaves me with the question of "how do I measure the 
> phase noise of what is by far the best oscillator I own... without 
> buying a better one to compare it to". That question is what brought 
> me to time-nuts. I'm starting to read some papers on oscillator 
> characterization that are collected together in a technical note from 
> NIST that a co-worker pointed me towards, but some of them are giving 
> me a math-induced headache.
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to 
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.
>




More information about the time-nuts mailing list