[time-nuts] Is this DOCXO an SC or AT cut rock?

Magnus Danielson magnus at rubidium.dyndns.org
Sun Aug 10 03:51:20 EDT 2014


On 08/10/2014 08:54 AM, Scott Newell wrote:
> I'm curious if an OCXO has an SC or AT cut crystal in it. There's no
> EFC, so I can't measure sensitivity. I did measure the frequency change
> from cold startup, thinking that an SC cut would stabilize quicker than
> an AT. I don't have a good AT OCXO to compare against, but I do have a
> Morion MV89A (should be an SC, right?).
> Graphs:
> http://n5tnl.com/time/mv89a/mv89a.png
> http://n5tnl.com/time/mv89a/unknown.png
> Thoughts?
> The two units are the same size and shape. The Morion ran without the
> outer can.

Sure looks like a SC cut oscillator. We did this exercise many years 
ago, and if you dig the archives you can find the numbers for AT, BT and 
SC cut oscillators. AT cut was in the -145 ppm range as I remember while 
SC is -20-25 ppm off.

Good that you made the cold-start measurement!
Extra points for measuring currents. Interesting to see the shut-off 

War-story: I was testing one oven oscillator which looked nice on paper. 
Measuring it I saw a bump in the ADEV at 7 s, and looking at the 
frequency & phase plots I saw the wobble. Hooking up a DMM I can barely 
see the wobble. As I power-cycled the oven, I see the same wobble 
already from power-up. Turns out that the oven controller has a very 
high Q value and is ringing and that the ringing never really dies out.
The vendor had transfered a design to a new board, so the oven was 
sitting on a ceramic board rather than a FR-4 board inside the can. FR-4 
has much more thermal mass so it kept the loop nice and stable, but not 
the re-tuned oscillator.

> Measurement note: my 5370A is sick, and my 5345A doesn't do GPIB. So I
> shot video with the counter (1 s gate time, using a tbolt as the 10 MHz
> ref) and meter (measuring current) in the frame. From there, I used vlc
> to spit out still images at about 1 Hz, then imagemagick to cut out the
> sections with the seven segment displays. ssocr
> (<http://www.unix-ag.uni-kl.de/~auerswal/ssocr/>) did the real magic of
> converting the seven segment displays in the images to text. Had to do
> some hand cleanup on a few data points, but it was a lot less
> frustrating than typing in thousands of data points by hand. I'm sure
> it'd work better with a more careful layout and lighting.

Messy, but it works. Hope you find a smoother measuring path soon.

Sure would like if TimeLab could log current and temperature too.


More information about the time-nuts mailing list