[time-nuts] Re: DIY Low offset Phase Noise Analyzer (Erik, Kaashoek)

[Bob kb8tq]

Hi

One of the “interesting” gotchas with some sound cards:

The frequency response goes to (say) 48 KHz. The noise
performance goes to (maybe) 20 KHz. Past that point there
is a nasty rise in the noise floor. How common this is … who
knows. 

For a lot of applications, the system floors out well before 
10 KHz. There really isn’t a lot of need for phase noise past
that ( other than for spurs ). 

It’s debatable just what the “best” approach for wide band 
spurs is. A notch filter in front of a spectrum analyzer is a 
often a good way to go.

Bob

> On Jul 13, 2022, at 9:30 AM, Erik Kaashoek <erik at kaashoek.com> wrote:
> 
> The oscillator I wanted to measure with the PNA (a cheap Chines TCXO) can now be measured with 30dB margin and the internal DOCXO is better than any other oscillator I have so I could say I'm done, apart from becoming a phase noise nut :-)
> The whole build is still Manhattan style with long component leads so making a more compact build on a solid copper PCB may further improve performance.
> @Ed:  The 7805 was temporarily replaced with a linear bench supply which made no visible difference
> @Bob: The audio input I'm using claims to be 192kHz but in reality it stops at 96kHz so the upper limit is 48kHz for now, which is good enough. The calibration method you advised  (offset DUT) works very well now the gain is switchable.
> Erik.
> 
> On 13-7-2022 18:54, Bob kb8tq wrote:
>> Hi
>> 
>> Well, obviously it’s now time to go get some better OCXO’s :) :) :)
>> 
>> One further note: It is tempting to go to multi section lowpass filtering
>> if you are trying to cover a range of frequencies. You also might try to push
>> the range of the noise measurement up to higher and higher offsets. There
>> is a point that all of this gets nutty. Peaking in the audio range is often the
>> net result. Spice is one way to watch out for it. Sweeping the circuit is
>> obviously a better way.
>> 
>> Bob
>> 
>