Maker Pro
Maker Pro

average fractional frequency instability

Hi,

I need a high quality RF signal source, for example from OCXO or RF
signal generator.
But I need to know, what is a fractional frequency instability
measured during 1s period. One can find information about SSB phase
noise spectra in datasheets, but how to convert this numbers to
frequency instability?
In RF signal generator I can set requested frequency on a display. But
how stable generated signal is? Can I expect that generated frequency
will change about 1e-8 or maybe 1e-12 during 1s?
Of course, this depends on the internal reference and phase loops
inside the box, but such a data should be published in datasheets.
From SSB phase noise spectra one can obtain time jitter, but it is
possible to convert this value to frequency deviation?

thanks,

E.C.
 
T

Tim Shoppa

Jan 1, 1970
0
Hi,

I need a high quality RF signal source, for example from OCXO or RF
signal generator.
But I need to know, what is a fractional frequency instability
measured during 1s period. One can find information about SSB phase
noise spectra in datasheets, but how to convert this numbers to
frequency instability?
In RF signal generator I can set requested frequency on a display. But
how stable generated signal is? Can I expect that generated frequency
will change about 1e-8 or maybe 1e-12 during 1s?
 Of course, this depends on the internal reference and phase loops
inside the box, but such a data should be published in datasheets.
From SSB phase noise spectra one can obtain time jitter, but it is
possible to convert this value to frequency deviation?

The information you want would be included in the "Allan Deviation"
plot. Google that term and "Allan Variance". 1 second is getting a
little long-in-the-tooth but the phase noise is very definitely
related.

A really good stable crystal oscillator, e.g. HP10811A, will often
have an Allan Deviation of 1E-12 to 1E-11 at a tau of 1 second. Lesser
crystal oscillators might be 1E-9 to 1E-6 in the same tau.

Tim.
 
The information you want would be included in the "Allan Deviation"
plot. Google that term and "Allan Variance". 1 second is getting a
little long-in-the-tooth but the phase noise is very definitely
related.

But I can't find AVAR plots for frequency synthetizers.
I've made a simple calculation. From Agilent E8663B datasheet we know
L(f) [dBc/Hz] for a given frequency. We can calculate
S_phi(f) [rad^2/Hz] using relation

S_phi(f) = 2*10^( L(f)/10 )

Allan variance is related with S_phi(f) by equation (12-27) from
http://tf.nist.gov/timefreq/general/tn1337/Tn061.pdf (page 204). For
data taken from
Agilent's specification and for carrier frequency 1GHz, I have
AVAR(t=1s)=2e-13s, but this value doesn't agree (it's too low) with
AVAR for example HP10811A (http://www.leapsecond.com/pages/z3801a-
osc/). Maybe I made mistake during calculations, but is this way of
obtaining AVAR is correct?

thanks,
E.C.
 
Top