[time-nuts] xtal oscillator phase noise

John Miles jmiles at pop.net
Tue Jan 1 14:39:57 EST 2008


There isnt any; that's not a correct assertion for crystal oscillators in
the general case.

The math associated with multiplying or dividing a frequency is pretty
straightforward.  If an input edge is N picoseconds late due to jitter, the
corresponding output edge is also going to be N picoseconds late within the
bandwidth of the device.

So it's easy to see how phase noise in dBc/Hz is related to the percentage
of one cycle that a given edge is early or late.  Double the duration of a
cycle, and the effect of a constant jitter deviation is halved, or reduced
by 6 dB by the usual 20*log(N) amplitude rule.  Halve the duration of a
cycle, and the effect of the jitter is doubled.

Crystal oscillators do not behave anything like dividers or multipliers,
unless they have dividers or multipliers built in.  Only the >160 MHz Wenzel
oscillators use built-in mulipliers, I believe, and none of them use
dividers unless you order one explicitly as a separate part.

Higher-frequency crystal oscillators are generally cleaner at offsets beyond
a few kHz.  Only at close-in offsets do the 5 MHz ULNs have an advantage
over the 10 MHz ones, and there are no hard-and-fast 6-dB relationships as a
rule.

-- john, KE5FX (got a lot more stuff to post, but it's going to take awhile
to get it together...)

> -----Original Message-----
> From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com]On
> Behalf Of Tom Van Baak
> Sent: Tuesday, January 01, 2008 11:22 AM
> To: 'Discussion of precise time and frequency measurement'
> Subject: Re: [time-nuts] xtal oscillator phase noise
>
>
> > Tom -
> >
> > Nice data. It really confirms what I initially posted. Your measurements
> > were at 5 MHz, so, the expected number at 10 MHz would be 6 dB
> worse. The
> > -155 dBc/Hz number quoted by memory from me then was not that
> far of at all.
> > - Mike
>
> Hi Mike,
>
> What's the math behind why an equivalently good 10 MHz
> reference is always(?) 6 dB above a 5 MHz reference?
>
> /tvb
>





More information about the time-nuts mailing list