[time-nuts] gpsdo 10MHz steering resolution

[Bob kb8tq]

Hi

The quick version:

Going straight from 10 MHz to 1.5 GHz with no spurs is a tall order if the 
10 MHz comes from some sort of system. The normal answer is to put in 
a cleanup oscillator (often a OCXO) in the 100 MHz range. That locks through
a narrow loop and takes care about as well as you can. 

A typical GPSDO has a very small LSB. The spurs in the output are far more 
likely to come from the various digital gizmos associated with the design. The 
same issue (spurs from the digital side) shows up in several Telecom Rb’s as
well. 

At some point in the 0.1 to 0.001 ppb range you get into the basic issues with 
GPS. It is very noisy close in and a lot of averaging is done to get a GPSDO 
to work. The base signal out of a number of GPS modules have jitter in the 1
ppb range, even after correction. 

The grubby details:

One mili-hertz at 10 MHz is 0.1 ppb. Sub-mili-hertz at 10 MHz might be
0.01 ppb. On a peak to peak / 1 sigma basis at a 1 second sample rate, that is doing 
well for a GPSDO that is OCXO based. Run the measurement for a day and you
will find 0.1 ppb peaks. 

Measured as a 1 second ADEV something around 0.01 ppb to 0.001 ppb is doing 
quite well. The often mentioned TBolt is one example of a GPSDO that does this 
sort of performance. 

Spur wise, you can easily find spurs in the 60 dbc range on some OCXO’s. Finding
a device that is clean to 100 dbc would be unusual. A 1 KHz spur at -70 dbc would
be unusual but by no means impossible. (20 log N from the -30 on your display). 
To get things down in the -60 db range, the GPSDO would need to be better than
-100. That is the “why” of the cleanup loop / 100 MHz OCXO.

One very “simple" solution:

Don’t try to use a GPSDO as the input to the device. Run it off of a “free running”
OCXO. That will give you the best spurs and reasonable stability. Then compare 
the LO to a GPSDO via a secondary chain. Feed the data from that chain into the 
SDR and let it correct the display. There are a *lot* of systems out there that do it 
this way …..

Bob



> On Nov 25, 2019, at 3:34 AM, Mike Ingle <finndmike62 at gmail.com> wrote:
> 
> Hi all, this is my first post to the the list.  I am working on a time
> stamped RF receiver system which can stamp at a fraction of a ps
> resolution, with a jitter (1 sigma) of about 10ps.  I have been using a
> truetime XL-AK Irig receiver as my 1pps and 10MHz references.
> Unfortunately, the truetime 10MHz has strong spurs which my 1.5GHz PLL is
> unhappy with, leading to spurs in my RF.  Can anyone recommend a good GPSDO
> with sub millihertz steering?  For example, I bought a GPSDO from a ham
> here in germany, and the 10MHz is pure, but has a step resolution of
> 0.2Hz.  I have attached the truetime 10MHz output.  best regard --mike
> 
> 
> 
> 
> [image: IMG_0269.JPG]
> <IMG_0269.JPG>_______________________________________________
> 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.