[time-nuts] How do I measure oscillator frequency using 1pps?

Brooke Clarke brooke at pacific.net
Sat Jul 9 17:35:19 EDT 2005

Hi David:

I'm typically getting a sigma of 9 ns from my M12T+ with 1000 averages.
I use the GPS 1 PPS as the start signal to the SR620 TI counter and the 
1 MHz output from the Cesium standard as the stop signal.  Note that 
with this arrangement rollover occurs at 1 microsecond, not 1 second, 
but it takes months for that to happen.

If you look at the Time Interval once per day you will have 9 ns of 
sigma on the GPS signal which divided by 86,400 seconds is 1E-13.  But 
the 9 ns number gets divided by SQRT(n) where n is the number of 
readings you are averaging, so in my case of 1,000 averages the 1 day 
accuracy is about 3E-15.

Another option is to use a computer to read the negative sawtooth 
correction from the GPS receiver and use that on a second by second 
basis to correct the 1 PPS TI.


Brooke Clarke, N6GCE
w/Java http://www.PRC68.com
w/o Java http://www.pacificsites.com/~brooke/PRC68COM.shtml

David Kirkby wrote:

> I've now got
> 1) Stanford PRS10 rubidium standard
> 2) Motorola M12+ timing GPS receiver with a 1 pps output.
> 3) HP 5370B time interval counter.
> I'd like to look at the drift of the rubidium before I try to steer it 
> with the PLL. Can anyone explain how to do this with the 5370B?
> I understand how you can measure the drift between two sources at 10MHz, 
> by looking at the rate at which the time interval is changing, but I 
> don't understand how to do it with the 1pps GPS signal, since the jitter 
> on the 1pps will be of the same order (or larger) than the time period 
> of the 10MHz source. If I start Even if you fiddle the cable lengths so 
> the time interval between the 1pps and the rubidium is 50ns, the jitter 
> on the GPS means a value of 1ns could be 1ns or 101ns, and you would 
> have no idea at all.
> Perhaps the way to do it is to use the rubidium as a start, and the 1pps 
> as a stop. Then the TI will be somewhere in the range 0 to 1 second, and 
> the jitter can be averaged out in software.
> I'm basically confused, as I am sure you can tell.

More information about the time-nuts mailing list