[time-nuts] M12+T First Light

[Tom Van Baak]

> Isn't that because the 58503A was tuned for S/A
> being on ?  After they shut S/A I found that
> I had to change timeconstants for optimal performance
> on my PLL loop.

I don't know for sure. It does seem that the HP
SmartClocks are overly conservative with their
time constants. But then again, for a commercial
product with wide environmental specs you have
to be a little conservative. You've probably seen
a 10811 do a frequency jump; either at random
or due to sudden temperature change, vibration,
voltage change, or 180 degree turn-over. S/A was
probably part of it but I don't how much relative
to all the other things the PLL has to correct for.

In any event if one were to roll their own GPSDO
using a very fine hand-picked OCXO, or Rb, or
even Cs, in a protected stable environment and
tweak the PLL to the max I'm sure you'd get
better results than a SmartClock.

But not that much better, I predict. Even if you
had a perfect OCXO and PLL you'd very soon
run into the noise floor of a real-time L1-only
GPS receiver.

I guess my point was that if one is measuring
their cesium performance using GPS as the
long-term time reference, then averaging raw
1 PPS measurements from something like a
sawtooth corrected M12+ will give at least the
same precision as any fancy GPSDO, maybe
even better. Does this make sense?

/tvb