Comparing two HP Z3805A GPSDO 10 MHz frequencies for lab use

[edpalmer42]

The 53230A has a resolution of 20 ps according to the picture.  If you're measuring a 10 MHz signal, 20 ps equals 0.072 degrees.  Is it making multiple measurements and averaging them to get all those digits?  If so, it's hiding a lot of the information that you want to see.

Ed

[awallin]

The 53230A has a resolution of 20 ps according to the picture.  If you're measuring a 10 MHz signal, 20 ps equals 0.072 degrees.  Is it making multiple measurements and averaging them to get all those digits?  If so, it's hiding a lot of the information that you want to see.
Ed

In normal continuous frequency counter mode "CONT" the 53230A uses some kind of internal averaging. This artificially lowers the numbers if you calculate e.g. ADEV statistics. You have to use the undocumented RCON mode to get a correct Allan deviation number.
http://www.anderswallin.net/2015/06/cont-vs-rcon-mode-on-the-53230a-frequency-counter/

This papper has some background, and measurements on the older 53132 which does lambda-counting by default, and produces something close to the Modified Allan Deviation when you post-process the data.
http://www.researchgate.net/publication/6308454_Considerations_on_the_measurement_of_the_stability_of_oscillators_with_frequency_counters


AW

[edpalmer42]

I have a Fluke/Philips PM6681 counter (50 ps resolution).  It has a 'Single' mode that disables all the averaging and gives you the results of one measurement.  I always use Time Interval & Single when I'm doing AlDev runs - except when I forget.  (Why are these results so good?......Oh, crap!  )

Ed

[VK5RC]

Be a bit careful with a HP/Ag/KS 53230, the input BNCs are mounted directly on the PCB and it the signal trace line can end up fractured, esp on Channel A; if out of warranty (as mine was!!!!!) HP has kindly left a small SMD pad so you can solder the bodge without having to dismount the PCB. Photo of my bodge!

[HighVoltage]

Be a bit careful with a HP/Ag/KS 53230, the input BNCs are mounted directly on the PCB and it the signal trace line can end up fractured, esp on Channel A; if out of warranty (as mine was!!!!!) HP has kindly left a small SMD pad so you can solder the bodge without having to dismount the PCB. Photo of my bodge!

Interesting and good to know.
My 53230A is pretty new and probably does not have this problem as of now.

I just found a good way of getting the Data from the 53230A for an Allan Variance analysis.
I will try to perform that tomorrow on both GPSDOs

[HighVoltage]

In normal continuous frequency counter mode "CONT" the 53230A uses some kind of internal averaging. This artificially lowers the numbers if you calculate e.g. ADEV statistics. You have to use the undocumented RCON mode to get a correct Allan deviation number.
http://www.anderswallin.net/2015/06/cont-vs-rcon-mode-on-the-53230a-frequency-counter/

Thanks for the great link.
It sounds like something I could do with the tools I have

[HighVoltage]

Ok, today I installed an Agilent free software package called "Stability Analyzer 53230A" that can be downloaded from Matlab Central here:
http://www.mathworks.com/matlabcentral/fileexchange/31319-stability-analyzer-53230a

It integrates in to Matlab and works perfectly with the Agilent 53230A over a network connection.
I just had to enter the IP address of the instrument to the software and hit the Analyzer button.

Although, I have no experience with Allan or Hadamard deviation, it is really cool to get this data so easily analyzed from the instrument.

Setup:
The 53230A counter is hooked up to one GPSDO on channel 1
The software takes as many samples as I select, in this case I did 1E3 and 1E5 samples
And here are the results:

I probably have to play a lot more and read up on Alan and  Hadamard deviation, to understand it all.
Next is probably to take a few measurements with both GPSDOs and compare them.
I am not sure, if I will see any differences, but probably not, since both GPSDOs are so similar.


Please let me know, what you think about this stability analysis.

[Zucca]

Interesting mod on a Z3805A!

https://www.eevblog.com/forum/chat/what-did-you-buy-today-post-your-latest-purchase!/msg3144112/#msg3144112

Thanks to gamalot!

[HighVoltage]

That indeed is a good modification.