[time-nuts] Best GPS 1PPS Accuracy

[Tom Van Baak]

Said,

The CNS-II outputs a raw hardware sawtooth corrected
1PPS and this is by design; it's not a mistake. There is
or may be a GPSDO version of the CNS-II clock but I
don't need it. What I (and some of the VLBI sites) use
the CNS for is to monitor the long-term performance of
masers and cesium clocks.

[ Rick can jump in quickly to correct me if needed ]

In this application you don't need a 10 MHz GPSDO;
you don't even need a "clean" 1PPS derived from a
disciplined OCXO. Over hours or days, the raw 1PPS
from the receiver has all the information, has all the
stability, that you need and it is not "polluted" by any
disciplining algorithm or jitter and wander of the OCXO.
Does this make sense or shall I explain a bit more?

I think one can make a case that for *long-term timing
purposes* anything resembling a GPSDO will perform
slightly worse than the bare GPS engine upon which
it is based. It has to. Consider:

On the ADEV plot, to the left of the crossover point,
the GPSDO performance is essentially that of the
OCXO minus a bit for system noise minus a bit due
to the "drag" of the GPS engine at short tau.

It follows then that to the right of the crossover, the
GPSDO performance is essentially that of the GPS
engine minus a bit for system noise minus a bit due
to the "drag" of the OCXO at long tau.

So if you don't need a CW RF output frequency, and
only need long-term timing, and are post processing
all the TIC data anyway, then the OCXO -- or anything
else that's part of a disciplining system -- has no value
added. Might as well use the best raw 1PPS you can
get; which is what that version of the CNS-II does.

/tvb

  ----- Original Message ----- 
  From: SAIDJACK at aol.com 
  To: time-nuts at febo.com ; tvb at leapsecond.com 
  Sent: Thursday, December 14, 2006 19:21
  Subject: Re: [time-nuts] Best GPS 1PPS Accuracy


  In a message dated 12/14/2006 16:46:10 Pacific Standard /tvb writes:
    The beauty  of a GPSDO is that you're only changing
    the DAC every few minutes so  there's plenty of time
    to look carefully at the quality of your 1 Hz  samples
    before you commit the DAC update.

    It would be a more  difficult problem if you needed to
    update the DAC in  realtime.

    /tvb

  Hi Tom,

  one thing I noted in the PTTI paper on the CNS Clock II block diagram that was recently published here was that the CNS clock II is outputting a raw but sawtooth corrected version of the 1PPS.

  ftp://ftp.cnssys.com/pub/PTTI/PTTI_2006.pdf

  This 1PPS will suffer from the jitter not recognized by TRAIM that we are talking about.

  I wonder why they did not use the OCXO output DIV 10E6 to generate a clean 1PPS, since this is so much more stable as you mentioned?

  Or is there a mistake in the block diagram?

  bye,
  Said