Question For Craig Markwardt re Pioneer 10 Data

Thanks Craig for the clarification that the transmitter
frequency for all transmissions to Pioneer 10 was the same
and produced by a circuit that multiplied 96 in the early
days;48, later, times a very precise 24 MHz local oscillator
frequency And that this then could produce a very reliable
difference in
the received frequency and the transmitted frequency.
My question: What is ratio of error sum of squares around the
selected frequency to the sum of squares around the mean?
Does item 101 Average Doppler Residual in the NASA data
tape have something to do with the numerator of this ratio
eg 3 times the sq rt of numerator would lead to a 99percent
confidence interval for the true received doppler shift?
Ralph Sansbury


The process of digging the weak received signal out of noise
as I understand it involved the
representation of nanosecond voltage variations as a Fourier
series with the largest weighted sine component of frequency
around 2292MHz and the other sine components much smaller.
The specific phase and frequency is detected using filters,
Fast Fourier Transforms and Phase Locked Loops. And if you
subtracted
the received voltage values at each nanosecond or fraction of a
nanosecond from those predicted by the detected frequency and
phase, you would get a set of numbers that was normally
distributed around zero indicating that these differences were
noise.
Of course if the component of frequency in the expected range
has the same weight as those in other ranges then this would
indicate that it too was noise also.
If the sum of squares of the observed around a predicted set of
values is as great as the sum of squares about the mean of the
set of values then the predicted set of values is worthless and I
suppose some sort of criteria is the basis for saying that the
receptions from Pioneer 10 are now lost in noise.
It would be nice to get a little more clarification on this
point eg What is ratio of error sum of squares around the
selected frequency to the sum of squares around the mean?
Does item 101 Average Doppler Residual have something
to do with the numerator of this ratio eg 3 times the sq rt of
numerator would lead to a 99percent confidence interval for
the true received doppler shift?
Ralph Sansbury





"Craig Markwardt" wrote in
message news





"ralph sansbury" writes:

Hi Craig,
Re the transmitter frequency subtracted from the received
frequency
to the get the doppler shift and motion of Pioneer 10
relative to
the earth
at any specific time.
If the multiplier is exactly 48 for the DCOcase but 96
earlier
and this corresponds to something
specific in the phyical circuit, that would be ok.
(What does it correspond to?)


The hardware has a fixed integer multiplier between the
reference
oscillator and the transmitted frequency. For the VCO the
multiplier
was 96, for the DCO it was 48. This is not a tunable
parameter,
i.e. it is fixed exactly by the electronics and microwave
components
of the oscillator and amplifier.

The "choice" of 48 vs. 96 comes in the modeling software. The
multiplier in the software must match the multiplier used in
the
hardware. There is no subjective choice involved.


Exactly and that is my question???? If the milliHz terms
supposedly
used to show a small anomalous acceleration would have been
changed
by using a different multiplier and there is no independent
reason for choosing
48 or 48.1 etc, then there is a problem!!!!



And, to reiterate, there is no fitting or tuning involved in
the DCO
multiplier.

Another problem is how do we know the transmitter frequency
was
always
exactly the same as the frequency produced by the DCO times
48?

Because that is how the system was designed, tested and
productively
used for more than a decade.

"ralph sansbury" writes:
My question: What is ratio of error sum of squares around the
selected frequency to the sum of squares around the mean?
Does item 101 Average Doppler Residual in the NASA data
tape have something to do with the numerator of this ratio
eg 3 times the sq rt of numerator would lead to a 99percent
confidence interval for the true received doppler shift?

You appear to be asking about the signal to noise ratio. The ATDF
Doppler tracking records do not appear to contain a measure of the
carrier signal to noise ratio. However, data quality measures for the
Doppler data *are* present, including Doppler noise and estimate cycle
slips, which are of course the most relevant quantities for Doppler
tracking.

However, there are published measures of the carrier signal to noise
ratio. For example, Watola (1992) shows a plot from Pioneer 10 on 19
December, 1991, where the signal to noise ratio is 20 dB for a
bandwidth of ~0.2 Hz, which means the mean signal is stronger than the
noise by a factor of 100. The epoch of the measurement is in the
middle of the data arc of published papers, such as Anderson et al or
my own paper. Of course, signal to noise ratio depends on many
environmental and technical factors, not always constant.

The "average Doppler residual" field of ATDF records is not a signal
to noise measure. Rather, it is the difference between the measured
and predicted Doppler frequencies. I.e., the navigation group
prepared a prediction of the Doppler frequency before the track, and
the "residual" field represents the difference between the measured
frequency and the prediction.

CM

References
Watola, D. A. 1992, TDA Progress Report 42-111

"Craig Markwardt" wrote in
message news

"ralph sansbury" writes:
My question: What is ratio of error sum of squares around
the
selected frequency to the sum of squares around the mean?
Does item 101 Average Doppler Residual in the NASA data
tape have something to do with the numerator of this ratio
eg 3 times the sq rt of numerator would lead to a 99percent
confidence interval for the true received doppler shift?

You appear to be asking about the signal to noise ratio. The
ATDF
Doppler tracking records do not appear to contain a measure of
the
carrier signal to noise ratio. However, data quality measures
for the
Doppler data *are* present, including Doppler noise and
estimate cycle
slips, which are of course the most relevant quantities for
Doppler
tracking.

Yes I was afraid of that from looking at some of the papers
you
suggested in your notes. And I dont see how to relate "Doppler
Noise"
and "cycle slips" in the data. Re 100 times
stronger signal over noise to this clearer statistical accuracy
measure.
Maybe obs(i)=sig(i)+error(i) where error(i) is approximately
sig(i)/100
That is when you subtract the voltage changes observed over
time,obs(i), from|
one or a sequence of systematic sine voltages
obtained using Fast Fourier Transform techniques etc, sig(i), the
error(i) would
have this average magnitude, sometimes it would be 1/99 sometimes
1/101 etc.




However, there are published measures of the carrier signal to
noise
ratio. For example, Watola (1992) shows a plot from Pioneer 10
on 19
December, 1991, where the signal to noise ratio is 20 dB for a
bandwidth of ~0.2 Hz, which means the mean signal is stronger
than the
noise by a factor of 100. The epoch of the measurement is in
the
middle of the data arc of published papers, such as Anderson et
al or
my own paper. Of course, signal to noise ratio depends on many
environmental and technical factors, not always constant.

The "average Doppler residual" field of ATDF records is not a
signal
to noise measure. Rather, it is the difference between the
measured
and predicted Doppler frequencies. I.e., the navigation group
prepared a prediction of the Doppler frequency before the
track, and
the "residual" field represents the difference between the
measured
frequency and the prediction.


I was afraid of that also, that I couldn't use the average
doppler residual
But it is good to know one can use the papal Watola measure of
signal to noise at one time
as representative of the signal to noise for years around 1991.
I gather that the Anomalous Doppler Shift measurments are 100
times
greater than noise and that they are systematic over time?.



CM

References
Watola, D. A. 1992, TDA Progress Report 42-111

"ralph sansbury" writes:
I gather that the Anomalous Doppler Shift measurments are 100
times greater than noise and that they are systematic over time?.

Question 1: from a statistical standpoint, that is approximately
correct; the statistical errors in the anomalous acceleration are a
few percent of the measured value. Question 2: yes, it is systematic
over time.

CM

"Craig Markwardt" wrote in
message news

"ralph sansbury" writes:
I gather that the Anomalous Doppler Shift measurments are
100
times greater than noise and that they are systematic over
time?.

Question 1: from a statistical standpoint, that is
approximately
correct; the statistical errors in the anomalous acceleration
are a
few percent of the measured value.

Craig, ..
Yes I see from the papers you refer to that the difference
between the received sky frequency(oscillating voltage vlaues)
and the known transmitted frequency(voltage values) is obtained
digitally as follows: a sequence of oscillating voltages is
received from the sky where the spacecraft is, through a filter
that suppresses frequencies outside the expected range and this
is input to one gate of a dual gate transistor while an
oscillation at a preset frequency is input to the other gate so
that the output of the transistor controlled by both of these
inputs contains the sum ,difference, and both input frequencies.
A resonance tuner picks out the difference frequency and a a
sequence of voltages at this difference frequency.(mixer and
repeated heterodyne up and down conversion etc is the jargon and
the engineering details I am trying to avoid).
This sequence of oscillations is digitized into a set of 1s
and 0s (1 if the analogue voltage is above a certain amount etc)
and an Fast Fourier Transform procedure is used to find the
underlying "sine" pattern of 1s and 0s that most closely fits
this. Since the incoming frequency is constantly changing
slightly because of the motions of the earth etc, the detected
underlying sine patterns will change.
The 100 times greater noise figure in the Watola paper means
then that the differences between the FFT sequence value and the
corresponding observed sequence value is zero, 99 times out of a
100.
Is this what Watola says? I should hope this is quoted
somewhere in the Anderson papers or yours?
Ralph



Question 2: yes, it is systematic
over time.

CM

Hi Ralph,

"ralph sansbury" wrote in message
...

Craig, ..
Yes I see from the papers you refer to that the difference
between the received sky frequency(oscillating voltage vlaues)
and the known transmitted frequency(voltage values) is obtained
digitally as follows: a sequence of oscillating voltages is
received from the sky where the spacecraft is, through a filter
that suppresses frequencies outside the expected range and this
is input to one gate of a dual gate transistor while an

Where did you read that it was a dual gate transistor?

oscillation at a preset frequency is input to the other gate so
that the output of the transistor controlled by both of these
inputs contains the sum ,difference, and both input frequencies.
A resonance tuner

Wrong, we have been over this dozens of times. It is
digitally filtered, there is no "resonance tuner"
involved. The characteristics are fundamentally
different.

picks out the difference frequency and a a

Wrong, it doesn't pick out one frequency, it passes a
complete band of frequencies to the FFT.

http://eis.jpl.nasa.gov/deepspace/ds...05/209/209.pdf

See page 10 (yet again).

sequence of voltages at this difference frequency.(mixer and
repeated heterodyne up and down conversion etc is the jargon and
the engineering details I am trying to avoid).

Instead you are inventing a process that doesn't exist
and describing it in far more (and incorrect) detail
than exists in the published documentation.

This sequence of oscillations is digitized into a set of 1s

The entire band is digitised.

and 0s (1 if the analogue voltage is above a certain amount etc)
and an Fast Fourier Transform procedure is used to find the
underlying "sine" pattern of 1s and 0s that most closely fits
this.

Nope, the amplitude of _all_ frequencies in the band is
calculated and passed on to the next stage without any
judgement. Ralph, it's as if all the weeks I spent
talking you through the DSN documentation by email had
never happened. You seemed to grasp it at the time, why
have you reverted to this grossly inaccurate description
of the process?

George