[time-nuts] How does sawtooth compensation work?

David davidwhess at gmail.com
Tue Jul 19 00:50:45 EDT 2016

On Mon, 18 Jul 2016 21:41:51 -0700, you wrote:

>> Or use the sawtooth compensation value  to control an external variable delay line circuit
>Hi Mark,
>Right, one example is https://datasheets.maximintegrated.com/en/ds/DS1020.pdf or google for silicon delay line. Not sure they're in production still but you can find them at the reseller sites.
>This delay line idea came up in the early Oncore-VP era gps mailing list (pre time-nuts) by someone who first explored sawtooth correction and "hanging bridges"; and it's the method that Rick then chose for his CNS Clock product line. See: http://cnssys.com and http://gpstime.com for details.
>The advantage of the delay line method is that you don't need a nanosecond TIC in the box; you correct for the sawtooth error live on the 1PPS. Very simple and effective. The main GPS feed in my lab is a CNS Clock.
>This disadvantage is that if you already have a TIC connected to your GPS/1PPS, there's not much point in pre-sawtooth correcting with a delay line. The error is something that you're going to correct with arithmetic anyway so there's no need to correct it in pulse phase. Rick's TAC32 software (that many time nuts use) handles integration of serial TIC data (such as hp 53132) along with GPS binary data to provide sawtooth corrected measurements. Several of Rick's papers at the above sites explain this in fine detail.

Was their a specific reason to use an integrated variable delay line
for this versus a DAC, ramp generator, and comparator?

I ask because I have never seen a schematic for PPS jitter correction
which used the later.

More information about the time-nuts mailing list