LF: Measurements on 10MHz references

[Andy Talbot <andy.g4jnt@gmail.com>]

In the file    http://www.g4jnt.com/DownLoad/10MHz_ReferenceSpurii.pdf    are plots of the 10MHz outputs from several 10MHz references and oscillators.  

They include four GPSDOs of varying types, an HP5061 Caesium standard, an oldish high quality TCXO recovered from a 1980's vintage STL comms receiver, and my long in the tooth MSF frequency reference using a Toyocom OCXO. 
This latter, and the Cs standard, both generate 5MHz which was doubled for these measurements using a full-wave rectifier doubler.

 

All plots were made using an SDR-IQ on the same settings: 0.37Hz FFT bin at spans of 10kHz and 100kHz with 4 times averaging.  The final two measurements on the two Simple GPSDOs were made by G0API using his SDR-IQ with the same setup as mine.   For all plots, signal level / attenuation were adjusted to be as close to full scale as possible without overload.

 

Some artefacts at the -100dBc to -110dBc region seem to be in common across several plots, and I suspect these are locally generated or due to the SDR-IQ.   However, there are enough and sufficient differences to see that some sources are definltely better than others, and that the cheaper cost conditioners do allow spurious signals onto the output.

 

My own ZAZ GPSDO, the second set of plots, originally showed unstable sidebands at 650Hz which I realised were coming from an LCD backlight inverter.  With this removed, it is now shown to be one of the better sources - but with the high spec TCXO that was recovered from the old Decca Navigator chain, berhaps not too surprising.  

 

Teh second ZAZ plot, using the modern Morion OCXO is the cleanest in terms of spuriou, although the phase noise of that oscillator is higher than many of the others.   This is, I believe, characteristic of that OCXO which was used originally mainly as a a timing unit, specifically  for its fast warm up and low retrace errors rather than for its noise performance.

 

 

Andy

www.g4jnt.com