John C. Polasek wrote:
On 27 Jun 2006 01:11:49 -0700, "George Dishman"
wrote:

John C. Polasek wrote:

I tried to show you in my flowgraph paper on my website, that Doppler
isn't even in it. The round trip times are too small.

Even for a 20 hr. round trip the Doppler change in beat is a
fractional change df/f0 = 2e-13 or 4.8e-4Hz, and after 8 years (the
other end of the chart) you have the same values 2e-13 and 4.8e-4Hz.
There would be nothing to plot.

The chart shown would be the same if you used the station clock,
without the Doppler returned signal. The only thing that can
reasonably cause that is a continuously increasing station clock rate
or a mystery acceleration which we can rule out.

John, it isn't that simple. A linear increase in the
station maser clock rate would cause an increase in
the transmitted frequency in 1994 compared to 1987
and that would in turn cause an equal fractional
increase in the returned frequency. The measured
value is the difference between that and a reference
also generated from the maser (albeit at another
site[*]) but that difference is then measured using
a timebase derived from the same maser.

I think you are talking aboug deltaF on the graph,

No, I understand your graph to show the difference
between measured Doppler and the Doppler predicted
based on the modelled speed of the craft. I don't have
an issue with that, I think you have it right. What I am
talking about is the method used to make the
measurement and how it would be affecte by clock
drift.

and we both agree
it's negligible. My argument is that the station or returned (either)
frequency vs the synthetic frequency in the model that makes the
anomaly.

There is no "synthetic frequency" in the model. The
expected shift is calculated from the modelled craft
speed and the measured transmit frequency.

What that means is that a simple change of rate
cancels out. I don't think your flowchart illustrates
that point and it is very important in any consideration
of clock rate variation.

George
[*] The secular rates will be matched via synchronisation
to the international standard.

How can I say it again without being repetitious? Of course the return
frequency is bootstrapped off the station clock and their difference
essentially nulls out. I just pointed that out above, to the effect
that the return differences are just so much noise, in the big
picture.

You must know that I am talking about all real, maser-verfiied clocks
that accelerate compared to the artificial clock in the model which
for several reasons must have a constant value.

The point I am drawing your attention to is that there is
no "artificial clock" in the model, the speed is applied to
the transmit frequency which is measured/generated
against the maser at most 20 hours before reception.

The result is the ramp
function on the chart.

I went on at some length about how the fictional clock can only have
one proper book value. Even today it would be assigned the same
value, just as Cs33 would still have 9,192,731,770 to define one
second. What is there to check against?

In other words, as shown on the graph, the model's frequency is f0 and
the station clock's is f0 plus f0*H*t, leaving f0*H*t as the input to
the graph. .

If the station clock is [f0 plus f0*H*t] at the time of reception
then it is compared against [f0 plus f0*H*(t-tr)] where tr is the
round-trip time, not f0 as you seem to be suggesting.

George