[time-nuts] LPRO-101 with Brooks Shera's GPS locking circuit
SAIDJACK at aol.com
SAIDJACK at aol.com
Sat Dec 16 18:11:42 EST 2006
In a message dated 12/16/2006 12:31:31 Pacific Standard Time,
phk at phk.freebsd.dk writes:
In message <20061216202309.BA059BE00 at ip-64-139-1-69.sjc.megapath.net>, Hal
>> You still want to produce some form of quality measure for the ADEV
>> shape in order to form some form of control loop.
>Suppose I build I good GPSDO. How do I determine how good it is? (or even
>if it is any good)
Since you pretty much know that the ADEV of the GPS receiver is a straight
line, you can get quite a long way by doing an ADEV calculation over
your measured errors.
one poor man's way to check longer term performance without the help of ADEV
is to show your GPSDO's output (the 10MHz, or preferrably the stabilized
1PPS output) against the GPS receivers raw 1PPS output on a scope that can do
color grading with infinite trace persistance. Color grading is nice to have,
but not even needed for this test. All you need is a scope with infinite
If you trigger on the the raw 1PPS rising pulse, the 10MHz trace should move
back and forth within the 1PPS typical sawtooth range (+-13 to +-20ns for an
M12+) and never wander outside of this range on your scope display if the
OCXO is perfectly locked. The left and right short-term wander (sawtooth) is
really well defined, and you can see over time periods of 10 minutes to hours
how your OCXO output wanders beyond this sawtooth. If it doesn't, you've got a
really stable unit :)
If your OCXO is locked really tightly, the displayed range will fit neatly
into a 100ns pulse of the 10MHz OCXO output, and you won't get aliasing. If it
wanders more than +-50ns, the measurement is inconclusive due to aliasing
(the screen is just a white bar in this case)
Say now that the width of the wander is never more than +-35ns over 12 hour
periods, this means your unit is probably locked to the GPS receiver with
better than 70ns/(12*3600s) = 1.6E-012 to the GPS receiver for measurement
periods of 12 hours. This figure actually includes the GPS sawtooth error not
likely to be present on your OCXO output!
That's pretty good already if you can achieve such a small wander window for
such a long time!
Of course you can wait exactly until the wander reaches the +-50ns window,
then do the math for that particular time-period. If you have a raw GPS 1PPS
and a stable OCXO 1PPS you are not limited by aliasing, and can measure much
less stable OCXO's.
This method won't tell you much about shorter term stability unfortunately.
More information about the time-nuts