[time-nuts] Best counter setting for ADEV?

David davidwhess at gmail.com
Tue Oct 2 13:46:41 EDT 2012

In the case of just the sawtooth error, I thought it was caused by the
limited resolution of the counter generating the output pulse.  If the
counter clock was 100 MHz and not phase locked to GPS time, then
depending on how far off it is in frequency, the output pulse would
wander an additional but predictable 10ns.

On the other hand if it was phase locked, then the output pulse could
be stuck at one offset which could not be average out which might be
even worse.

On Tue, 2 Oct 2012 13:10:04 -0400, "Bob Camp" <lists at rtty.us> wrote:

>Almost any receiver that's labeled "timing receiver" will put out some sort
>of saw tooth correction. The saw tooth is a result of the math, so no
>there's not a cheap way to get rid of it. By the time you do a receiver that
>does the phase lock thing, you have pretty much built a GPSDO.
>-----Original Message-----
>From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
>Behalf Of David
>Sent: Tuesday, October 02, 2012 1:04 PM
>To: Tom Van Baak; Discussion of precise time and frequency measurement
>Subject: Re: [time-nuts] Best counter setting for ADEV?
>Is there a list of GPS timing receivers that provide the sawtooth
>correction message or implement sawtooth correction internally?
>I assume there is a design compromise that prevents economically phase
>locking the GPS receiver clock to the GPS signal to remove that
>contribution to timing error.
>On Tue, 2 Oct 2012 09:18:35 -0700, "Tom Van Baak" <tvb at LeapSecond.com>
>>Correct, all GPS timing receiver boards have jitter, sawtooth, or random
>wandering of some sort, on the order of tens of nanoseconds. This is normal.
>And so if you use a counter to compare the OCXO 1 Hz with the GPS 1PPS, a
>TIC resolution of 1 ns or 500ps is sufficient. I would say 25 ps is
>>In many cases (e.g., Motorola Oncore series) the sawtooth correction
>message itself has a granularity of 1 ns. So again, a 25 ps measurement is
>especially overkill given a correction granularity of 1 ns. Depending on the
>receiver applying the correction will improve the average timing performance
>by, say, a factor of 3.
>>See: http://www.leapsecond.com/pages/m12-adev/
>>As for your averaging question, yes, the OCXO will move during the average.
>This is normal. That's why too long an averaging interval is problematic.
>Depends on the quality of the OCXO. And if the averaging interval is too
>short, you pick up too much GPS jitter. Depends on the quality of the GPS
>receiver. There is no perfect answer; instead you choose something between
>too short and too long.

More information about the time-nuts mailing list