[time-nuts] "Best" GPSDO

[Jim Lux]

On 9/28/12 8:31 PM, Chris Albertson wrote:
> On Fri, Sep 28, 2012 at 5:29 PM, Bob Camp <lists at rtty.us> wrote:
>> HI
>>
>> Sort of an open ended question, but there is a fairly simple couple answers:
>>
>> SInce it's close in phase noise and not far removed, things like PLL's are going to transfer it directly from the reference to the output. It will of course scale by 20 log N where N is the amount you multiplied or divided the reference frequency by. Double the frequency and the phase noise goes up by 6 db.
>
> So in my example case of scaling the 10Mhz t-bolt to 14.5Mhz  Assuming
> a perfect DDS chip the T-Bolt's phase noise would be scaled up by 20
> Log(1.45)     I'm assuming this works, that I can go from 10MHz to
> 120Mhz and then to 14.5MHZ and the total effect is the same as going
> directly from 10 to 14.5, except for the noise the equipment
> introduces as added.
>
> You can guess the real question here: "how good does the 10MHz
> reference need to be to test real-world receivers?
>

It has to be quieter than the oscillator in the real world receiver.  If 
your real-wold receiver is a cryogenic ruby maser with a downconveter 
driven by a hydrogen maser reference, then the answer is "really, really 
good"..

If the real world receiver uses a run of the mill TCXO, then not nearly 
as good.

The 20log10(N) thing does work pretty well.  In a PLL synthesizer, 
you'll pick up a little extra noise from the phase detector and other 
circuitry, but for back of the envelope to see if your idea is going to 
work, the 20log10(N) is just fine.

This gets into a whole interesting area of microwave source design, 
because "inside the loop" the phase noise is the reference oscillator 
multiplied up (20log10(N) noise), and outside the loop, it's the 
microwave oscillator.   So you have an interesting optimization problem, 
particularly if you want tuning over a wide range. Wide range VCOs 
implies that the MHz/volt gain is quite high, so noise on the tuning 
signal shows up on the output.  The resonator is often lower Q, so that 
it can be moved around by the control signal (usually some sort of 
varactor scheme), and that means the "medium distance away" phase noise 
suffers.     High performance DROs for instance, have a tough time 
tuning the entire 50 MHz deep space comm bands at 7 or 8 GHz