VK4BXI wrote:Tests that I've done imply that the 10MHz signals from the GPS satellites are changing frequency randomly by small amounts.
I can assure you this is not the case. I have been deeply involved with the GPS and, in particular, GPS time and frequency standards, since before GPS reached it's Initial Operating Capability (IOC) in 1993.
Using the reference oscillator as an external trigger I then connected the two GPSDO's to the two inputs of my 100MHz scope.
I allowed everything to settle for a day or so and then looked at how the two traces were moving around. At times both traces were almost "rock" stable and moved together against the 10MHz crystal oscillator at a very low rate.....taking some 50 ~60 seconds to move a complete sine wave across the screen, indicating that my crystal oscillator was "spot on".
Those interpretations are exactly backwards. The two GPSDO outputs moved together, which means they are doing what they are supposed to, which is to generate highly accurate and identical 10MHz signals. It's the crystal oscillator that was not as close to 10MHz as the two GPSDOs. Trigger off of the GPSDO and watch the crystal oscillator output "move". The fact that both GPSDOs were locked together is a clear indications things were working as they should be.
The surprise came when one or other of the GPSDO's abruptly changed frequency by about 0.05 to 0.1 Hz and stayed that way for a few minutes. At times both were "off" ....and in different directions. The changes appear to be "random".
This is caused by a faulty or poorly designed GPSDO, and not any fault or design issue with GPS. Which GPSDO demonstrated instability, the Datachron or the VK4AMG unit?
I am coming to believe that either this is deliberate or a function of how the GPSDO's operate handing over from one satellite to another. I wonder if anyone else has noticed this or has an idea of what is going on ?
There is no "hand over" in GPS, it does not work like a cellular radio system. As the constellation in view changes the solution for frequency will not change appreciably and if there are short time periods (seconds, not minutes) where a solution is not available (unlikely unless you are in deep urban or natural canyons) the short term stability of the oscillator that is being disciplined is sufficient to maintain a very high level of accuracy. And, in a properly designed GPSDO, that oscillator will start to drift in the absence of a good GPS solution, but it should NEVER jump.
Again, I can 100% assure you GPS time and frequency transfer are far superior to your crystal oscillator and there are no "random frequency changes". The same is true for all extant GNSS systems including BeiDou, GLONASS and Galileo. Your experiments need to use a GPSDO as the master reference, and it appears you have at least one bad or poorly designed GPSDO.