[time-nuts] 5065A phase noise

[Bob kb8tq]

Hi

There are a *lot* of ways 60 Hz (and other power related) signals can get into an
RF chain. Pretty much anything that modulates the ground can modulate the (net)
supply to various stages. Modulate the supply and you get AM and PM noise…..

Taking care of ground loops is a significant part of any precision measurement.  The
issue is most obvious on a phase noise plot. The same sort of noise also impacts
ADEV measurements. 

=====

For the kind of noise levels in your plots, a simple single mixer driving a sound card
will do a fine job of measuring phase noise. Something like a Mini Circuits RPD-1 running
into a low noise op-amp “preamp”. If you terminate the mixer in about 5K ohms the
op-amp does not need to be super duper. An (very old) OP-37 would do fine. 

Ground loop wise, a sound card presents it’s own set of issues. A QA401 isn’t exactly cheap,
but it does a pretty good job.

https://quantasylum.com/collections/frontpage/products/qa401-audio-analyzer <https://quantasylum.com/collections/frontpage/products/qa401-audio-analyzer>

Bob




> On Oct 15, 2020, at 1:11 AM, AC0XU (Jim) <James.Schatzman at ac0xu.com> wrote:
> 
> Timenuts-
> 
> I was able to borrow a Holzworth HA7062C phase noise analyzer recently and used it to measure phase noise for several oscillators.
> 
> I have a few questions:
> 
> 1) I noticed that ground loops can be a significant problem - generating noise peaks at multiples of 60Hz and/or multiples of 120Hz. If this were merely additive noise, then no noise peaks near the 5MHz or 10MHz carrier should result in the analyzer, so non-linearities must be involved. Where are those non-linearities arising and how can they be eliminated?
> 
> 2) Using several 50 ohm baluns and disconneting power supplies from ground eliminated some of the 60Hz-related noise in my tests but not all. The attached is a phase noise plot of a 5065A, run into a Wenzel low-phase-noise doubler run off a battery, run through a balun into the analyzer. Some 60Hz-related peaks are still evident - should I assume they are produced internally by the 5065A?  In this case I could imagine that some additive 60Hz noise on the output of the 5065A gets nonlinearly transformed by the doubler. Unfortunately, the HA7062C lower frequency limit is 8MHz, so I had to use the doubler (or something similar).
> 
> 3) See the 105B plot, which is drawn directly from the 10MHz OCXO (I modified the 105B, adding a TAPR buffer of the 10MHz OCXO to feed an external port),which has bad 60xN Hz peaks despite my using a balun to eliminate ground loops on the output.  So no doubler in this case - where are these 60Hz-related peaks coming from?
> 
> 4) Both 5065A and 105B oscillators show peaks at 10KHz (multiples), 100KHz, 1MHz, etc. Is this nonlinear feedback from the dividers in the OCXO devices?  I could remove the dividers to see if that has an effect...
> 
> Another question - I can't afford to buy the Holzworth analyzer for my personal use, but I would like to continue my phase noise experiments. Any advice on acquiring or building an adequately sensitive instrument that doesn't break the bank?
> 
> BTW, I recently visited the Holzworth factory. Their designs and manufacturing processes are impressive.
> 
> Thanks in advance-
> 
> Jim<5065A_2.png><105B_2.png>_______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.