[time-nuts] 5065A phase-noise

John Miles john at miles.io
Sun Jan 12 22:01:42 UTC 2020


> Now, while these measures is performed not on the oscillator itself but
> oscillator and output buffers, unless something is wrong with output
> buffers on either box, the output buffer noise seems to support good
> enough flicker and white noise to support the output of both
> oscillators. Then running with that assumption it seems like the
> 00105-6013 has excess flicker noise issue while the 05065-6097 has
> excess white noise.

HP's signal conditioning approach was OK for 5061As and 5065As with the original 00105-series oscillators, but it doesn't do the 10811-based versions any favors.  I've found it handy to add a 'native' front-panel 10 MHz output to all of my units, repurposing the 100 kHz or 1 MHz jack.  

It is possible to improve the white noise floor substantially at the same time.  There are numerous possible strategies, but they all come down to tapping into the 10811's output signal before it enters the divider used to emulate the old 5 MHz oscillator.  A reasonably-quiet amplifier with good isolation and moderate-to-high input impedance is called for.  E.g., http://www.ke5fx.com/norton.htm , which yields PN plots like the (somewhat hard to read) 5065A trace at the bottom of the page at http://www.ke5fx.com/rb.htm .  

These days, I'd probably just use an LMH6702.  It's somewhat noisier than the discrete 2N5109 amp, but still good enough to preserve the 10811's performance while providing isolation in the 100 dB vicinity at 10 MHz.   Isolation with the common-emitter amp is adequate at ~40 dB but not great.  

Either way, a highly recommended mod for all 5061As and 5065As with the newer OCXOs.  

Re: your trace with the conspicuous flat area between 10 Hz and 1 kHz, my guess is that you have some excess intermittent noise or jumpiness in one of your dual references.  Try hitting ctrl-r to see if the cross-spectrum average is higher than either of the two raw FFT output traces, maybe?  That's something that won't normally happen in a good run.

-- john, KE5FX






More information about the Time-nuts_lists.febo.com mailing list