[time-nuts] First few measurements of my Arduino Due GPSDO

SAIDJACK at aol.com SAIDJACK at aol.com
Thu Sep 25 13:42:09 EDT 2014

that is a very good suggestion.
Sometimes it is sufficient to do that for a minute or two, some products  
will already "freak out" and bounce around significantly trying to recapture 
the  phase after such a short period.
Some points:
* If you disconnect the antenna, you may change the temperature and  
power-supply voltage of the board every so marginally as the antenna current  
drops-off immediately, and that effect can be significant. This can cause  
disturbances in some lower-cost, lower-power products that are measurable.  
Depending on where the antenna is disconnected, this could also cause a  
mechanical jerk to the PCB and thus affect the measurement. Lastly an antenna  cable 
will act as a heatsink to the PCB, so removing it changes the thermal  
behavior of the board. Since we are trying to measure parts per trillion,  
everything counts. This is most important when trying to measure the holdover  
drift of a unit, since we want to disturb the unit as little as possible when  
going into holdover.
* Some products allow you to create a Dirac step on the EFC voltage by  
sending a command to change the DAC setting. The command to do so in the 
Jackson  Labs products is: SERV:COARSEDAC X where X is [0..255]. This will give a 
good  indication of loop damping, and it creates a known and repeatable EFC  
voltage step
* Removing the antenna will cause the GPS to go into hyper-drive, fire up  
all its internal correlators, and generally consume the most amount of 
power.  This can have a measurable effect on the boards' temperature. Some 
products  allow going into manual holdover while still keeping the GPS chugging 
along as  if nothing happened. For the JLT products that command is: 
Lastly, instead of removing the antenna, it is also possible to introduce a 
 step into the frequency simply by turning the unit 90 degrees or 180 
degrees...  On most products that will cause a frequency shift of 1ppb to 4ppb. 
It will also  have side effects as convective cooling is now changed, and the 
unit was  "jerked", but on smaller units that is a viable option.
In a message dated 9/25/2014 10:29:19 Pacific Daylight Time,  
tshoppa at gmail.com writes:

Lacking  a GPS simulator (do any of us have one?) for testing loop response,
it is  very interesting to do a "Holdover test". Unhook or block the GPS
antenna  for an hour or 24 hours, hook it back up, and make a graph of time
offset  and tuning word as it recovers.

If you see the tuning word (this is  what Z3801 owners call the "EFC"
number) doing something very far from  optimal damping, then you can start
tweaking PLL parameters.

Tim  N3QE

More information about the time-nuts mailing list