[time-nuts] Z3816a and newbie questions

Tom Van Baak tvb at leapsecond.com
Sun Jan 9 04:53:11 EST 2005


> Hello,
> 
> just joined this list a few days ago. I have checked
> out K8CU's (great) pages and a few other references,
> but I may have missed some info.

Welcome to the list.

> Specs I have seen on the Z3801a for 10 MHz output are
> only 1 x 10^-9 per day. Looking at a link to NIST from
> www.realhamradio.com it seems that something around
> 10^-12 might be closer to possible.  I almost always
> have 8 satellites; never seen less than 6.

That 10^-9 number seems wrong; do you remember
where you saw it?

A value 10^-12 per day is more like it -- for either the
Z3801A or Z3816A, or most GPSDO for that matter.

Note that figures like this are for one day averages.
Another way to look at it: consider that most modern
GPS timing receivers can track time to within 100 ns
(10^-7). One day is close to 10^5 seconds so 100 ns
per day is on the order of 10^-12.

What you can't say for sure is that your 10 MHz will
be accurate to 1e-12 any instant you look at it. There
may be instants or short durations when it is worse,
say, 5e-12, or 1e-11, or even more. But, on average,
over a day, it will come out to 1e-12, or better.

> About what accuracy can I reasonably expect from this unit? Is there
> anything in the GPSCon statistics that gives me a better real grasp on
> accuracy of my 10 MHz?

It sort of depends on what you mean by accuracy. Time
accuracy or frequency accuracy? What's your application?
But a rule of thumb is 100 ns time accuracy, all things
considered.

Below that you enter a can of worms: have you measured
your antenna delay, your cable delay; splitter or connector
delays; receiver delay; BNC output delay, and many other
factors.

GPScon can report high-level events that have a bearing
on accuracy; locked or holdover mode, for example. Or
less than 2 SV for too long vs. the normal 5 or 6 SV.

But in general, neither the Z38xx nor GPScon can really
know the accuracy of the output. For this what you do
is measure the 10 MHz output with a more stable local
reference (e.g., cesium standard).

/tvb







More information about the time-nuts mailing list