Re: ATM Atomic clock stuff...

[Jeff Medkeff <medkeff@c2i2.com>]

At 05:08 PM 12/7/98 -0600, Mitch wrote:

>2. WWV can (and has been) used for this purpose with
>a simple receiver and an attachment to a computer.
>I'm still not sure if WWV can correct for transmission
>delay though. I'll check "Telescope Control" by Trueblood
>and Genet and find out.

WWV doesn't correct for anything. Corrections for transmission time are
usually done in the data reduction phase of the experiment on the basis of
the position of the receiver, the active layer bounce being exploited, and
the number of bounces. These last two items can be discerned from
information broadcast on WWV periodically. (There are systems that can make
these corrections in real time.)

This only applies for high-precision observations such as certain kinds of
occultations, however - most 'mundane' observations such as the astrometry,
photometry, or spectroscopy of routine phenomena are considered to be good
if done to +/- a second or so. In part this is because, e.g., in
astrometry, the most-used reference net (the GSC) has residuals larger than
a second of time would introduce; and so on. In most routine observations
the observer manually gets the time from WWV with no correction, or the
observer might grab the time from his or her wristwatch which was
previously set to WWV, or the observer might grab it from whatever
observatory clocks are available at the observing site.

Which brings up the question - just what are you doing with a program like
TheSky that requires sub-second timing precision? I think you will find
that imprecise algorithms, calculation latencies, and the like will destroy
any gain you might experience from refining your time to better than a
second. And even if it doesn't, the chance that you don't know your
position precisely enough to exploit the data is high, especially if you
are using something like a GPS to get your position.

At 08:08 PM 12/7/98 +0000, Dan Fundo wrote:

>I have checked the National Naval Observatory site on the net against my
>old Radio Shack TimeCube.  The NO site is about 1+ to 2 seconds slow
>most of the time.

The Naval Observatory site has been dead on WWV time for at least the last
year and a quarter of (my) observing runs - I think you mean that the NO
site returns a time to your machine that is one or two seconds slow
compared to WWV. This is a common level of imprecision when using a
garden-variety time protocol. SNTP, which is a rather rigorous method, has
been in use here for about a year. We have kept long-term data on the
residuals between SNTP-supplied system times and WWV clock times corrected
for our position, and find that on the worst nights our SNTP supplied time
is off by about .07 seconds (which is a good deal more accurate timing than
we need). Our worst precision occurs when we are using dial-up internet
connections, and our best occurs when we are using a 10Base-T LAN to
connect to OC-3 backbones. At any rate, if anyone is using SNTP and is
getting time one or two seconds slow most of the time, something is wrong.


--
Jeff Medkeff          | Acting Assistant Coordinator
Rockland Observatory  | Association of Lunar and Planetary
Hereford, Arizona     | Observers, Solar Section

On the web at http://shutter.vet.ohio-state.edu/