[time-nuts] low noise multiplication to 100 MHz

Bob Camp kb8tq at n1k.org
Tue Jan 26 12:14:38 UTC 2016


> On Jan 25, 2016, at 8:36 PM, Gerhard Hoffmann <dk4xp at arcor.de> wrote:
> Am 25.01.2016 um 18:20 schrieb Graham / KE9H:
>> There are clock distribution parts designed to do this low noise frequency
>> conversion and distribution.
>> Consider TI  LMK04100
>> ....
>> 150 fs class jitter.
> But only if you integrate the noise only from 12 kHz offset to 20 MHz.
> It is a telecom spec.

You take the phase noise over that range and calculate the jitter from the phase noise. 
There is no “phase” (as in audio phase) information in the phase noise information you collect. There is just 
amplitude. Since you do not have angle information, you have to make some assumptions
about how the noise in each region sums up. 

How much of a difference can that be? Take a look at how the Fourier components of an impulse
add up. The angle of the components can matter quite a bit. Keep the same amplitudes, but fiddle
the angles between the components — you get a very different waveform. 

The calculation has the nice property that you *can* do the math with normal data. Run on 
broadband noise it works pretty well. When you get into noise regions that *may* have 
correlated noise (say from a modulation process). 

So, yes, there are a number of questions you need to ask when you see a jitter spec. A
number done by the formula will be very different 12 KHz to 20 MHz compared to 1 Hz to 
100 KHz. The directly measured noise may be different from either number.


> regards, Gerhard
> _______________________________________________
> 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_lists.febo.com mailing list