Pioneer 10 rx error and tx frequencies?

"Craig Markwardt" wrote in
message news

"ralph sansbury" writes:
[ Markwardt: ]
You can speculate as much as you like on what "could" be,
but it is
irrelevant. The point is, the DSN documentation describes
what is
*actually* implemented (See DSN 810-005 Rev E), and so no
speculation
is required. And from the documentation, we know there is
no link
from the exciter or transmitter assemblies to the receiver.

This is not true. There is no description of the
transmitter/exciter item in TRK-2-25.

Since I cited DSN 810-005 above, your comment is irrelevant.

There is no description "of" TRK-2-25 in DSN 810-005,
Rev.E 202,Rev.A.
If you are referring to something else then quote it or tell
me the lines in the link and give me the link url
Even if there is no link from the receiver system to the
oscillator connected to the transmitter amplifier connected to
the antenna
there could be a transmitter exciter oscillator in the receiver
system.


The output of the exciter is not used the
Doppler counter system,

How do you know that is not used in the general receiver
system?

DSN 810-005 Rev E.

Quote the exact phrase you are interpreting in this way.

DSN 810-005 Rev E, Chapter 202, Section 2.


It is not in what I have. Again quote the exact phrase.

I am not a mind reader. The TRK 2-25 record describes when
the
transmitter is off but not necessarily why. Since there
are many
projects competing for DSN time, the most obvious reason is
that there
wasn't time in the schedule for more passes.

Nonsense. To use the time most efficiently transmission
and
reception would be going on simultaneously.
If reception was going on anyway, what time saving would
be
gained by turning off the transmitter?

If there is no time in the schedule for the corresponding
downlink,
then there is no reason for an uplink. That is not nonsense.

I stand corrected. Yes that would be a reason for doing so.


If I gave a list of exact times, what more would that
prove? Nothing.
If you don't believe my numbers now, you won't believe a
list of times
either, so what's the point?

Without some sense of the pattern of these numbers I dont
dont
know if they could be consistent with my claim or not. (You
have
said you rejected and filtered out many other numbers that
were
not consistent with other numbers in the same fields at other
times. Something of the sort might be going on here).

The pattern is simple. About 30% of the time,



there[were] complete downlink
sessions with no contemporaneous uplink records. During those
same
periods, the uplink transmitter (and exciter) at the receiving
station
are *off*. That is why your scenario is ridiculous.


Do you also have records that show another site in view of the
spacecraft was not transmitting at the same time that the
receiving
station was receiving?

What do you mean by contemporaneous uplink records?
You are suggesting that the transmitter/exciter was coded off
for long periods of times "complete sessions" of
several hours perhaps.
Give me an example of one such session. Is there in the
87037t071
file I have been able to obtain
an example of such a session and is this an unfiltered file?
If your example shows the changing doppler shifts
due to the motions of the earth and that there were no
other sites that were in view of the craft and transmitting
at the same time then I would be very surprised.


Or even if
the transmitter being off or on refers to the assumed
transmission site at the
time of reception and not the reception site.

Each record contains a station ID, so contrary to your
statement, the
associated station is unambiguous.

And so is the assumed transmission site for the associated
station and time.

No. The receiver records are stamped with the receiving
station's ID.
Thus, during the passes in question, the *receiving station's*
transmitter is off. Why are you being so pig-headed?

Why are you being so vague and making additions as you go
along?


False. DSN 810-005 contains a wealth of supporting
documentation.
What is quite evident is your continual failure to read any
of the
documentation.
I have read the documentation and nothing in it supports
your claim.

Doubtful. See above.

Again I have 810-005, Rev.E 202,Rev.A in front of me p4-15
where
section 2 General Information and section 1 Introduction are on
p4.
If you are referring to some other document please quote the
parts explaining
what is meant by the transmitter/exciter on or off item in the
receiving data record.
Please also clarify the pattern of on off times and give links
for documentation of
contemporaneos uplink records
Also for the fourth time what is the difference between file
87037.dat and 87037t071.dat?

"ralph sansbury" writes:

"Craig Markwardt" wrote in
message news

"ralph sansbury" writes:
[ Markwardt: ]
You can speculate as much as you like on what "could" be, but it is
irrelevant. The point is, the DSN documentation describes what is
*actually* implemented (See DSN 810-005 Rev E), and so no speculation
is required. And from the documentation, we know there is no link
from the exciter or transmitter assemblies to the receiver.

This is not true. There is no description of the
transmitter/exciter item in TRK-2-25.

Since I cited DSN 810-005 above, your comment is irrelevant.

There is no description "of" TRK-2-25 in DSN 810-005,
Rev.E 202,Rev.A.

I never claimed DSN 810-005 described the TRK-2-25 data format. What
I *did* claim was that DSN 810-005 describes the actual implementation
of the DSN tracking system. If you were to ever carefully examine the
cited document portions, you would see that (a) there is no
transmitter or exciter in the receiver system, and (b) there is no
link between the exciter (or transmitter) and the receiver system.


Even if there is no link from the receiver system to the
oscillator connected to the transmitter amplifier connected to
the antenna
there could be a transmitter exciter oscillator in the receiver
system.

No there couldn't. Besides the fact that it would be stupid (i.e. why
have a whole RF transmitter system embedded within your receiver?),
the configuration of the receiver system is actually documented in DSN
810-005, and thus you are wrong.



Do you also have records that show another site in view of the
spacecraft was not transmitting at the same time that the
receiving
station was receiving?

Yes, of course. Examples include 1987/07/15 17h, 1987/09/27 12h,
1987/11/04 17h, and so on (hours are rounded down). These are
tracking passes of at least an hour duration, with excellent quality
tracking data, and no other uplink stations in view of the spacecraft
at the time of downlink. There are 102 such passes in the total
1987-1994 arc, and 200 if shorter duration passes are allowed. Such
tracking sessions directly invalidate your scenario.

Also for the fourth time what is the difference between file
87037.dat and 87037t071.dat?

Why do you keep asking? I don't know about, or care about, your CD.
The on-line NSSDC files from 1987-1988 are valid. The filenames often
contain date information, for example 87037t071 probably means 1987,
day numbers 37-71, but the vital information is in the ATDF records.

CM

"Craig Markwardt" wrote in
message news

"ralph sansbury" writes:

snipped previous non answer to my request for quotes that show
the transmitter/exciter on or off in the receive records could
not refer to
such an oscillator in the receive system or link to such an
oscillator in
the transmission system disconnected to the transmitter
amplifier.

But assuming that transmitter/exciter off in the receive
system
meant that the transmitter at the receive site was off
Do you also have records that show another site in view of
the
spacecraft was not transmitting at the same time that the
receiving
station was receiving?

Yes, of course. Examples include 1987/07/15 17h, 1987/09/27
12h,
1987/11/04 17h, and so on (hours are rounded down). These are
tracking passes of at least an hour duration, with excellent
quality
tracking data, and no other uplink stations in view of the
spacecraft
at the time of downlink.

How do you know 'no other uplink stations were in view of the
spacecraft
at this time' and were or were not transmitting?
(I should think there would always be two sites with overlapping
views
of the spacecraft) And what do you mean by 'excellent quality
tracking data'? I suspect that the Doppler shifts in these passes
are
not exactly consistent with the just previously obtained shifts
when the transmitter
was on to the extent another transmission site was moving
differently than the
receive site.
I suspect also that if no other sites in view of the
spacecraft were transmitting
and the receive site transmitter was off that the received
oscillations in the expected
band were just noise.
Perhaps the local oscillator in the PLL set at a frequency
determined by 30 minutes of
previous bonafide received sky oscillations would accept a few
minutes or more of such noise
and still say not go 'out of lock'. This has to be clarified.
If most of the 102 hour duration receptions when the receive
site transmitter was off
and the other in view site was not transmitting, was clearly
noise( part of the 20percent garbage reception frequencies that
you filtered
out) then this correlation between no transmission by in view
sites and garbage reception by in view receiver
sites would conclusively prove my claim.




There are 102 such passes in the total
1987-1994 arc, and 200 if shorter duration passes are allowed.
Such
tracking sessions directly invalidate your scenario.






Also for the fourth time what is the difference between
file
87037.dat and 87037t071.dat?

Why do you keep asking? I don't know about, or care about,
your CD.
The on-line NSSDC files from 1987-1988 are valid. The
filenames often
contain date information, for example 87037t071 probably means
1987,
day numbers 37-71, but the vital information is in the ATDF
records.

CM

"ralph sansbury" writes:

"Craig Markwardt" wrote in
message news

"ralph sansbury" writes:

snipped previous non answer to my request for quotes that show
the transmitter/exciter on or off in the receive records could
not refer to
such an oscillator in the receive system or link to such an
oscillator in
the transmission system disconnected to the transmitter
amplifier.

But assuming that transmitter/exciter off in the receive
system
meant that the transmitter at the receive site was off

Your comments are completely irrelevant. The DSN documents describe
exactly where the transmitter and exciter are in the system (i.e. in
the uplink stage), and the Doppler tracking data contain a record
which indicates when these units are turned off.

Let's review, since I'm signing off from this thread.

1. There are more than a hundred downlink observations where the
transmitter at the downlink station is off, no other uplink station
is in view of the spacecraft, and yet a high quality carrier signal
is detected. Basis: my examination of original Pioneer 10 ATDF
Doppler tracking records (1987-1994), detection of a strong Doppler
signal in those records, while in the same records the transmitter
and exciter system are indicated to be off. The Doppler solution
residuals for these records have the same excellent quality as the
rest of the records.

This fact utterly rejects your scenario.

2. Your supposition that there is an "exciter" in the receiver system
is incorrect. Basis: DSN 810-005 Ch 202, Section 2, describes the
downlink and uplink systems. The figure and prose description in
this section shows that the transmitter and exciter are in the
uplink system, not the downlink. When the transmitter and/or
exciter are off, there is no way for an uplink to occur, since
there is no other path to the uplink antenna system (basis: again,
DSN 810-005). Furthermore, to have "another" exciter within the
downlink system would be stupid, since it would be redundant: it
would have to be immediately converted back from RF to electronic
signals, which would introduce extra noise.

3. Your supposition that there there is a link from the transmitter /
exciter to the Doppler counting system is incorrect. Basis: DSN
810-005 (same section), which shows that there is no such linkage.

4. It is worth pointing out that the reference frequency used for
Doppler counting is derived from the CSS (channel-select
synthesizer), which in turn is derived from the station's FTS
(frequency and timing subsystem), and no "exciter" or "transmitter"
is involved in this process. Basis: DSN 810-005 (same section).

5. A record of all uplink sessions is kept with the data, in the form
of specialized Doppler "ramp" records. For the observations in
question, these records show that at the time of downlink, there
are no other uplink stations that could have transmitted to the
spacecraft. Basis: my examination of all spacecraft uplink "ramp"
records, from all stations, at the times of downlinks for the
Pioneer 10 data in question.

6. Your speculations about the carrier loop "accept[ing] a few minutes
or more of ... noise and still say not go 'out of lock'" are
unfounded. First of all, the observations in questions are one
*hour or more*, not a few minutes (basis: see #1 above). Second of
all, the loop bandwidths are selectable, documented to be between
1-3000 Hz (basis: DSN 810-005 Ch 204, Table 1). So in the presence
of noise, the loop will lose lock within one second or faster
(basis: PLL theory, time constant ~ 1/bandwidth). Your claim is
therefore erroneous.

I have provided a substantial basis for all of my claims. You, on the
other hand, can only provide wild speculations with no substantiation.
With all of this direct evidence in contradiction of your scenario,
the burden is now on you to provide some strong and explicit evidence
of your claims. Good day.

CM

Craig,
Give me one or two complete examples. You have
specified times when I suppose
an hour+ session begins in 1987(7/15/11h and 11/04/17h) and the
transmitter exciter
is coded off but there is no documentation that one of the other
sites in veiw of
the spacecraft is not transmitting.
Also there is no documentation as to the quality of this
signal.
So before you sign off please give us the data items for these
times
about the transmitter being on or off (and its frequency and its
power
should also be zero or blank)as well
as the doppler count and the doppler noise or residual as well as
data from
the other sites at these times that indicate neither of these
were transmitting.
Ralph
"Craig Markwardt" wrote in
message news

"ralph sansbury" writes:

"Craig Markwardt"
wrote in
message news

"ralph sansbury" writes:


But assuming that transmitter/exciter off in the receive
system
meant that the transmitter at the receive site was off

Your comments are completely irrelevant. The DSN documents
describe
exactly where the transmitter and exciter are in the system
(i.e. in
the uplink stage), and the Doppler tracking data contain a
record
which indicates when these units are turned off.

I agree that the documents dont indicate there is a
transmitter
exciter in the receiver system or a link that could be on or off
between
the transmitter exciter in the transmitter system to the receiver
system.
But it is not completely clear that this is not a possibility.


Let's review, since I'm signing off from this thread.




1. There are more than a hundred downlink observations where
the
transmitter at the downlink station is off, no other uplink
station
is in view of the spacecraft, and yet a high quality carrier
signal
is detected. Basis: my examination of original Pioneer 10
ATDF
Doppler tracking records (1987-1994), detection of a strong
Doppler
signal in those records, while in the same records the
transmitter
and exciter system are indicated to be off. The Doppler
solution
residuals for these records have the same excellent quality
as the
rest of the records.

Give me one complete example. You have specified times when
I suppose
an hour+ session begins in 1987(7/15/11h and 11/04/17h) and the
transmitter exciter
is coded off but there is no documentation that one of the other
sites in veiw of
the spacecraft is not transmitting.
Also there is no documentation as to the quality of this
signal.
So before you sign off please give us the data items for these
times
about the transmitter being on or off (and its frequency and its
power
should also be zero or blank)as well
as the doppler count and the doppler noise or residual as well as
data from
the other sites at these times that indicate neither of these
were transmitting.

This fact utterly rejects your scenario.

There are no facts just repeated evasions and generalizations.


2. Your supposition that there is an "exciter" in the receiver
system
is incorrect. Basis: DSN 810-005 Ch 202, Section 2,
describes the
downlink and uplink systems. The figure and prose
description in
this section shows that the transmitter and exciter are in
the
uplink system, not the downlink. When the transmitter
and/or
exciter are off, there is no way for an uplink to occur,
since
there is no other path to the uplink antenna system (basis:
again,
DSN 810-005).
Yes but the figures and prose do not say that the exciter
could be
connected to the transmitter but not to the receiver system.

Furthermore, to have "another" exciter within the
downlink system would be stupid, since it would be
redundant: it
would have to be immediately converted back from RF to
electronic
signals, which would introduce extra noise.

If they can transmit and receive at the same time I am sure
they have
good ways of preventing noise in the receiver system.


3. Your supposition that there there is a link from the
transmitter /
exciter to the Doppler counting system is incorrect. Basis:
DSN
810-005 (same section), which shows that there is no such
linkage.


I agree that this suggests but it does not prove that there
is no such linkage.


4. It is worth pointing out that the reference frequency used
for
Doppler counting is derived from the CSS (channel-select
synthesizer), which in turn is derived from the station's
FTS
(frequency and timing subsystem), and no "exciter" or
"transmitter"
is involved in this process. Basis: DSN 810-005 (same
section).

No but the same FTS could be providing a frequency to
another
oscillator in the receiving system to be used to compare with the
sky frequency
received.


5. A record of all uplink sessions is kept with the data, in
the form
of specialized Doppler "ramp" records.

Where by the way is an explanation of this item "ramp"
records?

For the observations in
question, these records show that at the time of downlink,
there
are no other uplink stations that could have transmitted to
the
spacecraft. Basis: my examination of all spacecraft uplink
"ramp"
records, from all stations, at the times of downlinks for
the
Pioneer 10 data in question.


Again. show the ramp records at all three sites for
1987(7/15/11h and 11/04/17h)
and the documentation as to the meaning of ramp records.


6. Your speculations about the carrier loop "accept[ing] a few
minutes
or more of ... noise and still say not go 'out of lock'" are
unfounded. First of all, the observations in questions are
one
*hour or more*, not a few minutes (basis: see #1 above).
Second of
all, the loop bandwidths are selectable, documented to be
between
1-3000 Hz (basis: DSN 810-005 Ch 204, Table 1). So in the
presence
of noise, the loop will lose lock within one second or
faster
(basis: PLL theory, time constant ~ 1/bandwidth). Your claim
is
therefore erroneous.

I dont understand your explanation

If you are correct here then I would expect no clear sky
frequency in the
small band or range that was being received before the
transmitter was actually
turned off and there were no transmissions at the same time from
other sites.


I have provided a substantial basis for all of my claims. You,
on the
other hand, can only provide wild speculations with no
substantiation.
With all of this direct evidence in contradiction of your
scenario,
the burden is now on you to provide some strong and explicit
evidence
of your claims. Good day.

Its not easy if you withold the data that would substantiate
them.
For example you snipped my claim that the 10-20 percent of the
received
doppler counts that you filtered out as being inconsistent with
the rest
of the doppler counts might be highly correlated with the
transmitter
being coded off.
Also I wonder if there is not some other item in the tracking
data record that
would confirm the 1bit item transmitter on/off was always valid
eg item 116,
Transmitter/Exicter Frequency, or item 96 transmitter power
indicator.
If consistent doppler data is being received when these items
all indicate that the
transmitter was not transmitting at the same or different site at
the same time as
the data was being received then I would suspect there was
something wrong with
these indications.

"George Dishman" wrote in message
...

"ralph sansbury" wrote in message
...

"George Dishman" wrote in message
...

No, the basic idea is to test _all_ frequencies in the band
and find the highest. That way, if the signal is not where
it was expected, it still gets found.

Of course and I am saying the same thing but that the
principle is to
calculate each possibility and the method is to calculate
lots of
them at once and see which frequency or range is best.

Ok,

Craig says that he has 102 cases between 87 and 94 of at least
one hour
sessions when the transmitter/exciter at the reception site is
coded off(1)
and no other sites are transmitting and that the data is
good(consistent
with earlier and later dopplers and same average doppler
residuals?).
This would seem to clearly negate my claim and the need to
test
the doppler shifts predicted by this assumption against the
observed
doppler shifts.
The problem is that he does not want to provide the data that
shows
this. Maybe you could persuade him to be more forthcoming and to
provide
an excel spreadsheet of, if not all 117 items for each of the
hundreds of thousands
of minutes for which this data exists,at least some of this.
If the meaning of the item 26 is exactly that the transmitter
power at
the receiving site was off during these hour sessions and Craig
has
really looked at transmitter records at other sites for the same
time,
then it would seem pretty clear that my claim is wrong.
It would be impossible to randomly mistakenly put in a 1 bit
for 102 times
60 more different times which would indicate the transmitter was
off presumably and the
data received was good.Maybe the code '1' is given in the next
bit
position due to some sort of systematic error. This would be easy
enough
to check since the next bit position should be zero (and the
previous position
if '1' would indicate out of lock and the error residual would be
large?)
Will check the data for 87.





The zero crossing are always there but they will be a
fraction of a degree early or late compared to the
reference signal.

means that the local oscillator may be shifting back and
forth over a small range of frequencies

No, it will shift back and forth over a small range of
phase but that means the frequency of the reference
tracks the frequency of the received signal.

I dont see how you know it is due to a change in phase and
not to
a change in frequency

Phase is the integral of frequency so you can't have a
change of one without a corresponding change in the other,
but that is also the key to the operation of a PLL. Over a
long time, the phase moves back and forth but it doesn't
drift systematically one way. Such a systematic drift would
represent a frequency error so if there is only random
jitter in the phase, the _average_ frequencies are identical.

So you are saying that the zero crossings of the received
signal
may be a little late or a little early and if the size of these
differences
is small enough then the local oscillator frequency is the true
frequency
But if such variations are larger than a specific fraction of the
local oscillator period on average which may include some larger
than one period misses, then the local oscillator does not
represent
a true underlying frequency.
If there is a systematic increase or decrease then you have a
valid indication of a change in frequency consistent presumable
with the motionsof earth and spacecraft.

and I dont understand how many wrong zero
crossings and wrong lack of zero crossings you need before
you
conclude that you haven't locked onto a true carrier
frequency?.
Wrong means a local oscillator zero crossing that does not
correspond to an observed zero crossing.

The signals are normally in step, cycle for cycle. If the
phase error gets too big (think 1/2 a cycle), the loop will
'slip' along to the next cycle and then lock on to that.
Have you ever handled corrugated roofing? Think of sliding
one sheet over another. It is hard to pull up until it has
moved half a corrugation then it easily falls into the next.

Lets stick to what actually happens here. I think you are
making
it more complicated than necessary. When the reference and local
oscillator frequencies match and
the zero crossings all correspond then the average of the
products of the amplitudes is
..5 and if there is no match the average of the products is less
and closer to zero the worse the match..
So a first guesstimate of the chi square error sum of squares
over the sum of squares around
the mean zero is that this ratio is 0 when the average of the
products is .5
and that this ratio increases to 1 as the average of the products
decreases to 0.
If this than you can perhaps associate a probability
distribution with
values between .5 and 0 and values above a specified threshold
like .35.??


In a Type I PLL, there is a steady phase error which
gives a steady rate-of-change-of-frequency on the LO
which matches the drift of the received signal.

Could you give a specific example?

Suppose the LO runs at 500 Hz in the absence of an external
control, the quiescent condition, and changes frequency
by 100 Hz per volt (it's digital but I'll explain it as if
it was analogue to make it easier). Suppose the phase
detector gives an output of 1 volt per degree of phase
error (see the diagram below for what 'phase error' means).

If the receiver signal appears at 700 Hz, then the LO needs
2V to pull it from 500Hz to 700Hz so the loop will be stable
with a permanent phase error of 2 degrees. The phase is the
integral of the frequency so frequency is the derivative of
phase. The frequency error is the derivatve of the phase
error and since the phase error is constant, the frequency
error is zero. It should be obvious from the diagrams below
that a constant phase error represents two signals of the
same frequency with one shifted left or right in time
compared to the other.

In a Type II, the error is integrated so there is zero
phase error when the rate of change of frequency of the
LO matches that of the received signal as well as the
frequencies matching.

Again a specific example?

I'm going to have to draw waveforms :-( I'll leave out the
noise to start with.

Suppose this is the actual craft signal (it's supposed to be
a sine wave but triangles are easier to draw) first and the
the LO (after division by 256) second:


Sig /\ /\ /\ /\ /\
\ / \ / \ / \ / \ / \
----\----/----\----/----\----/----\----/----\----/----\----
\ /| \ / \ / \ / \ /| \ /
\/ | \/ \/ \/ \/ | \/
| |
|-| phase error |-|
| |
LO | /\ /\ /\ /\ | /\
|/ \ / \ / \ / \ |/ \ /
--\----/----\----/----\----/----\----/----\----/----\----/-
\ / \ / \ / \ / \ / \ /
\/ \/ \/ \/ \/ \/


In a Type II, there is an integration is the control path so
frequency of the LO will be increased at a rate that depends
on the phase error. A slight increase of LO frequency will
cause the bottom waveform to start drifting to the right so
it 'catches up' with the top waveform. The control circuit
is designed so that as the phase error reduces the rate also
reduces so the phase error would reduce to zero exponentially.

It is conceivable that the local oscillator frequency
zero
crossings correspond
only a small fraction of the total to the received
oscillations

No, there has a one-to-one correspondence between the
carrier
cycles and LO cycles when locked because otherwise the mean
product of the sine waves goes to zero.

I thought you said that the correspondence would be one to
one
if the mean of the carrier-LO product was .5 but could be
some
lower value above say .35 and still be considered enough
above
noise to say that "lock" was achieved.

Multiply the reference and the received signal. I'll do the
same phase error first:

Sig /\ /\ /\ /\ /\
\ / \ / \ / \ / \ / \
----\----/----\----/----\----/----\----/----\----/----\----
\ /| \ / \ / \ / \ /| \ /
\/ | \/ \/ \/ \/ | \/
| |
|-| phase error |-|
| |
LO | /\ /\ /\ /\ | /\
|/ \ / \ / \ / \ |/ \ /
--\----/----\----/----\----/----\----/----\----/----\----/-
\ / \ / \ / \ / \ / \ /
\/ \/ \/ \/ \/ \/

Product
/\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\
--\-/--\-/--\-/--\-/--\-/--\-/--\-/--\-/--\-/--\-/--\-/--\-/
V V V V V V V V V V V V

Note that the product is mostly positive but sometimes negative
so the mean is just above zero. Now the in-phase condition:

Sig /\ /\ /\ /\ /\
\ / \ / \ / \ / \ / \
----\----/----\----/----\----/----\----/----\----/----\----
\ /| \ / \ / \ / \ /| \ /
\/ | \/ \/ \/ \/ | \/
| |
| |
| |
LO | /\ /\ /\ /\ | /\
|/ \ / \ / \ / \ |/ \
----\----/----\----/----\----/----\----/----\----/----\----
\ / \ / \ / \ / \ / \ /
\/ \/ \/ \/ \/ \/

Product
\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\ /\
\ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ / \
---V----V----V----V----V----V----V----V----V----V----V----

Note that the product is never negative. The mean is higher
and in 0.5 of the peak if both sine waves are considered as
amplitudes of 1.

The test is applied to the simple mean of the product.

Should have figured out by now what this means in terms of
the error sum of squares between the observed and LO voltages
and so a statistical measure of error comparable to other
sorts
of things where this measure is used.
...
You still dont seem to able to connect the dots and
neither am I
Lock is arbitrary and yet it is associated with a criterion
in terms of
an
error sum of squares. What is the formula relating these two
things?

Consider the product terms shown above and think what will
result when you multiply the LO by pure noise. Regardless
of the LO, the product will be equally positive and negative
so the mean will be zero.

What it means is that you only get a lock indication if
the signals are in phase, not just at the right frequency.
In fact you could translate the threshold of say 0.35 into
a maximum rms phase error but I won't bother as I don't
know the actual value used, these are just illustrative.


The threshold of some specific value like .35

George

"ralph sansbury" wrote in message
...

"George Dishman" wrote in message
...

"ralph sansbury" wrote in message
...

"George Dishman" wrote in message
...

No, the basic idea is to test _all_ frequencies in the band
and find the highest. That way, if the signal is not where
it was expected, it still gets found.

Of course and I am saying the same thing but that the principle is
to
calculate each possibility and the method is to calculate lots of
them at once and see which frequency or range is best.

Ok,

Craig says that he has 102 cases between 87 and 94 of at least one hour
sessions when the transmitter/exciter at the reception site is coded
off(1)
and no other sites are transmitting and that the data is good(consistent
with earlier and later dopplers and same average doppler residuals?).
This would seem to clearly negate my claim and the need to test
the doppler shifts predicted by this assumption against the observed
doppler shifts.

It would indeed.

The problem is that he does not want to provide the data that shows
this. Maybe you could persuade him to be more forthcoming and to provide
an excel spreadsheet of, if not all 117 items for each of the hundreds of
thousands
of minutes for which this data exists,at least some of this.

I'm not aware how much effort this would entail but in
many cases I too have gone to considerable effort in
writing descriptions and providing information, only to
have you ignore it or discount it saying I didn't
understand the subject. You have a certain knack for
annoying people so I can understand why Craig is
unwilling to do more. Looking back in the thread, I see
he gave three examples,

"Craig Markwardt" wrote in
message news
Yes, of course. Examples include 1987/07/15 17h, 1987/09/27 12h,
1987/11/04 17h, and so on (hours are rounded down).

so perhaps you could look at those. It's unlikely that
checking a larger sample would tell you any more than
those. Did you get your program to read the files yet?

If the meaning of the item 26 is exactly that the transmitter power at
the receiving site was off during these hour sessions and Craig has
really looked at transmitter records at other sites for the same time,
then it would seem pretty clear that my claim is wrong.

I know Craig would not say they were off without
carefully checking first. He is a professional and
very careful not to claim anything he cannot back up.

It would be impossible to randomly mistakenly put in a 1 bit for 102
times
60 more different times which would indicate the transmitter was off
presumably and the
data received was good.Maybe the code '1' is given in the next bit
position due to some sort of systematic error. This would be easy enough
to check since the next bit position should be zero (and the previous
position
if '1' would indicate out of lock and the error residual would be large?)
Will check the data for 87.

The software that wrote these tapes and the programs
that analyse them have been used for decades for dozens
of missions. There is no chance whatsoever that such a
basic flaw would go unnoticed.

I dont see how you know it is due to a change in phase and
not to a change in frequency

Phase is the integral of frequency so you can't have a
change of one without a corresponding change in the other,
but that is also the key to the operation of a PLL. Over a
long time, the phase moves back and forth but it doesn't
drift systematically one way. Such a systematic drift would
represent a frequency error so if there is only random
jitter in the phase, the _average_ frequencies are identical.

So you are saying that the zero crossings of the received signal
may be a little late or a little early and if the size of these
differences
is small enough then the local oscillator frequency is the true frequency

Almost, I am saying that they may be a little late or early
but there must be the same number in the reference as in the
actual signal in a long integration therefore the frequency
is the true frequency.

But if such variations are larger than a specific fraction of the
local oscillator period on average which may include some larger
than one period misses, then the local oscillator does not
represent a true underlying frequency.

Again, almost. If there is a very large error of more
than half a cycle time displacement for several cycles,
then the reference can 'slip' by getting one cycle out
of step. Before and after the frequency will be exact
but the total over the integration period will be out
by exactly 1Hz (due to the way the counters work) and
this is sufficiently obvious in the recorded results
that it could be identified by the analysts. This is
described in the Anderson paper.

If there is a systematic increase or decrease then you have a
valid indication of a change in frequency consistent presumable
with the motionsof earth and spacecraft.

Right, systematic changes such as the frequency drift
caused by the acceleration component of the motion of
the sites is tracked by the oscillator so does not
cause slips.

and I dont understand how many wrong zero
crossings and wrong lack of zero crossings you need before you
conclude that you haven't locked onto a true carrier frequency?.
Wrong means a local oscillator zero crossing that does not
correspond to an observed zero crossing.

The signals are normally in step, cycle for cycle. If the
phase error gets too big (think 1/2 a cycle), the loop will
'slip' along to the next cycle and then lock on to that.
Have you ever handled corrugated roofing? Think of sliding
one sheet over another. It is hard to pull up until it has
moved half a corrugation then it easily falls into the next.

Lets stick to what actually happens here. I think you are making
it more complicated than necessary. When the reference and local
oscillator frequencies match and
the zero crossings all correspond then the average of the
products of the amplitudes is
.5 and if there is no match the average of the products is less
and closer to zero the worse the match..

The mean of the product is 0.5 if they are in phase,
-0.5 when 180 degrees out of phase and in general
0.5 sin(p) for a phase error of p. Provided the
system is receiving a valid signal the out of phase
condition cannot persist as the loop always pulls the
oscillator towards zero error.

If you multiply a sine wave by noise on the other hand,
then you are equally likely to get plus and minus results
so the mean will be zero. To get a steady positive value
requires two things

a) it must be tracking a real signal, not noise
b) the rms phase error must be within some small margin

So a first guesstimate of the chi square error sum of squares
over the sum of squares around
the mean zero is that this ratio is 0 when the average of the
products is .5
and that this ratio increases to 1 as the average of the products
decreases to 0.

I wouldn't like to comment on chi square values, the
system only uses a simple mean of the products and
indicates lock when that is sufficiently high.

If this than you can perhaps associate a probability distribution with
values between .5 and 0 and values above a specified threshold
like .35.??

For a lock indication threshold of 0.35, the phase error
would be arcsin(0.35/0.5) which is about 45 degrees. If
the reference signal is at exactly the same frequency as
the craft signal and remains in les than 45 degrees phase
error for several seconds then the lock indicator would go
on. The wrong frequency or too poor a signal-to-noise ratio
or too much phase error would all switch it off.

Please remember this value is only illustrative, I don't
know what would be used and it may even be dynamically
adjusted.

Happy New Year
George

"George Dishman" wrote in message
...

"ralph sansbury" wrote in message
...

Craig says that he has 102 cases between 87 and 94 of at
least one hour
sessions when the transmitter/exciter at the reception site
is coded
off(1)
and no other sites are transmitting and that the data is
good(consistent
with earlier and later dopplers and same average doppler
residuals?).
This would seem to clearly negate my claim and the need to
test
the doppler shifts predicted by this assumption against the
observed
doppler shifts.

It would indeed.

The problem is that he does not want to provide the data
that shows
this. Maybe you could persuade him to be more forthcoming and
to provide
an excel spreadsheet of, if not all 117 items for each of the
hundreds of
thousands
of minutes for which this data exists,at least some of this.

I'm not aware how much effort this would entail

If he has done what he says he has then he should be able to
back it up to some extent. Especially when what he says is
contrary to the general DSN policy and to common
sense. That is the efficient way of scheduling transmit and
receive
times to a specific spacecraft like Pioneer 10 is to transmit and
receive at the same time from one site and then at a later time
when the return signal is expected at the site in view of the
spacecraft
for a sufficient length of time at this time, etc.
So why would they turn the transmitter off during some of these
receive sessions for hours at a time? Craig says he is not a mind
reader but there should be some logs or something to explain
this!

but in
many cases I too have gone to considerable effort in
writing descriptions and providing information, only to
have you ignore it or discount it saying I didn't
understand the subject. You have a certain knack for
annoying people so I can understand why Craig is
unwilling to do more.

Yes it is annoying when people point out one's mistakes.
and you and Craig also are human and make mistakes like
the rest of us mortals. And are a little obnoxious sometimes too.
And we can all be irritable in such cases.
But at least you know I appreciate your expertise as I do
Craig's also.

Craig says that the data he
has from his programs applied to the doppler data files
proves that the transmitter at the receiving site was off
when the receiver was receiving good data.
I am questioning this. It could be transmitter/exciter off
does not mean that the transmitter is off in the transmission
system although as he points out this is not indicated in one
of the documents that briefly describes the two systems and
shows only one transmitter exciter.
This is a reasonable argument but not conclusive.
It could be that another transmitter at another site or the
same site is on but he says that he has data to show this
was not the case.
It could be that the code 1 in bit no. 237 that he
observes is misplaced by the coder(bit no. 236 is 1 when
the receiver loop lock indicator is 'out of lock' ) or Craig's
program.
( He acknowledges that the data tape is full of garbage that
had to be filtered out. Perhaps code '1' in no.237 is also
garbage.)
It could be that the data he says is good is not good
since the criterion used to determine 'good' is not stated.
I realize that such objections to one who is convinced
of the validity of the speed of light delay extrapolation, seem
absurd and not worth the time to examine.
He does not want to examine them, ok but he should provide
the data he claims and do so hopefully in an accessible form
eg an excel spreadsheet.


Looking back in the thread, I see
he gave three examples,



"Craig Markwardt" wrote
in
message news
Yes, of course. Examples include 1987/07/15 17h, 1987/09/27
12h,
1987/11/04 17h, and so on (hours are rounded down).

so perhaps you could look at those. It's unlikely that
checking a larger sample would tell you any more than
those. Did you get your program to read the files yet?

I have read the files and extracted some of the fields into
32 bit words but have not yet figured out how to format these
words
to import them to an excel file eg as numbers or ascii numbers
separated by commas
with an ascii code 13 (carriage return) to signify the end of a
row.
Maybe you or some of your colleagues know the best way to do
this?.



If the meaning of the item 26 is exactly that the
transmitter power at
the receiving site was off during these hour sessions and
Craig has
really looked at transmitter records at other sites for the
same time,
then it would seem pretty clear that my claim is wrong.

I know Craig would not say they were off without
carefully checking first. He is a professional and
very careful not to claim anything he cannot back up.

It would be impossible to randomly mistakenly put in a 1
bit for 102
times
60 more different times which would indicate the transmitter
was off
presumably and the
data received was good.Maybe the code '1' is given in the
next bit
position due to some sort of systematic error. This would be
easy enough
to check since the next bit position should be zero (and the
previous
position
if '1' would indicate out of lock and the error residual
would be large?)
Will check the data for 87.

The software that wrote these tapes and the programs
that analyse them have been used for decades for dozens
of missions. There is no chance whatsoever that such a
basic flaw would go unnoticed.


Nonsense. Craig is filtering out "garbage" now so it wasnt
'noticed' before
If a an inconistency in the speed of light assumption was
noticed
they may have changed the data to remove this in some cases
but not in others.




I dont see how you know it is due to a change in phase
and
not to a change in frequency

Phase is the integral of frequency so you can't have a
change of one without a corresponding change in the other,
but that is also the key to the operation of a PLL. Over a
long time, the phase moves back and forth but it doesn't
drift systematically one way. Such a systematic drift would
represent a frequency error so if there is only random
jitter in the phase, the _average_ frequencies are
identical.

So you are saying that the zero crossings of the received
signal
may be a little late or a little early and if the size of
these
differences
is small enough then the local oscillator frequency is the
true frequency

Almost, I am saying that they may be a little late or early
but there must be the same number in the reference as in the
actual signal in a long integration therefore the frequency
is the true frequency.

But if such variations are larger than a specific fraction of
the
local oscillator period on average which may include some
larger
than one period misses, then the local oscillator does not
represent a true underlying frequency.

Again, almost. If there is a very large error of more
than half a cycle time displacement for several cycles,
then the reference can 'slip' by getting one cycle out
of step. Before and after the frequency will be exact
but the total over the integration period will be out
by exactly 1Hz (due to the way the counters work) and
this is sufficiently obvious in the recorded results
that it could be identified by the analysts. This is
described in the Anderson paper.

You are saying I think that
upward zero crossing in the received oscillation that
correspond exactly or are not more than a quarter of
a cycle off and vary this way back an forth so that
when the local oscillator has had 10 upward zero crossings
that the received frequency has had 9 or 10 or 11 etc and then
is more than half a cycle later than that in the local
oscillator and this continues for a specific number
of cycles(eg 2) and then reverts back to the previous
pattern of upward zero crossings that is more similar to
that of the local oscillator then the total number of
such cycles in the receiver oscillation will be one cycle
less than the received and this is specified somewhere
(item 76 ,'total lippled cycles during count') ? as 1 cycle
or 1 Hz.

If there is a systematic increase or decrease then you
have a
valid indication of a change in frequency consistent
presumable
with the motionsof earth and spacecraft.

Right, systematic changes such as the frequency drift
caused by the acceleration component of the motion of
the sites is tracked by the oscillator so does not
cause slips.

and I dont understand how many wrong zero
crossings and wrong lack of zero crossings you need
before you
conclude that you haven't locked onto a true carrier
frequency?.
Wrong means a local oscillator zero crossing that does
not
correspond to an observed zero crossing.

The signals are normally in step, cycle for cycle. If the
phase error gets too big (think 1/2 a cycle), the loop will
'slip' along to the next cycle and then lock on to that.
Have you ever handled corrugated roofing? Think of sliding
one sheet over another. It is hard to pull up until it has
moved half a corrugation then it easily falls into the
next.

Lets stick to what actually happens here. I think you
are making
it more complicated than necessary. When the reference and
local
oscillator frequencies match and
the zero crossings all correspond then the average of the
products of the amplitudes is
.5 and if there is no match the average of the products is
less
and closer to zero the worse the match..

The mean of the product is 0.5 if they are in phase,
-0.5 when 180 degrees out of phase and in general
0.5 sin(p) for a phase error of p.


If the peaks are 1 and -1 in the received and reference
oscillations then the average products of in phase oscillations
are
0 plus (+1)*(+1) plus 0 plus( -1)*(-1) divided by 4 = .5
and the average products of systematically out of phase
oscillations are -.5.
and the products of 90 degree out of phase oscillations is zero.
So you can represent the average of the products as .5cos(p)
so when p=0 then .5 times 1 and when p=90deg then.5 times zero
and when p=180 deg then .5 times -1 etc.
If the received frequency is sometimes one frequency and just
as often
another or sometimes the same frequency in one phase and just as
often another
then we are receiving noise.
That is the average of the products of the cases above is zero.
However if the average of the products is closer to .5 and enough
closer to .5 so that it would be unlikely to result from a
uniform
distribution of the phase error "p" then we can with that degree
of
confidence conclude our result is not noise.
I think we are getting closer to a correct statistical formula
for the
degree of validity of the received signal.



Provided the
system is receiving a valid signal the out of phase
condition cannot persist as the loop always pulls the
oscillator towards zero error.

If you multiply a sine wave by noise on the other hand,
then you are equally likely to get plus and minus results
so the mean will be zero. To get a steady positive value
requires two things

a) it must be tracking a real signal, not noise
b) the rms phase error must be within some small margin

So a first guesstimate of the chi square error sum of
squares
over the sum of squares around
the mean zero is that this ratio is 0 when the average of the
products is .5
and that this ratio increases to 1 as the average of the
products
decreases to 0.

I wouldn't like to comment on chi square values, the
system only uses a simple mean of the products and
indicates lock when that is sufficiently high.



If this than you can perhaps associate a probability
distribution with
values between .5 and 0 and values above a specified
threshold
like .35.??

For a lock indication threshold of 0.35, the phase error
would be arcsin(0.35/0.5) which is about 45 degrees. If
the reference signal is at exactly the same frequency as
the craft signal and remains in les than 45 degrees phase
error for several seconds then the lock indicator would go
on. The wrong frequency or too poor a signal-to-noise ratio
or too much phase error would all switch it off.

Please remember this value is only illustrative, I don't
know what would be used and it may even be dynamically
adjusted.





Happy New Year
George



Maybe we can resolve all of this in the new year
Happy New Year to you too
Ralph

"ralph sansbury" wrote in message
...

"George Dishman" wrote in message
...

"ralph sansbury" wrote in message
...

Craig says that he has 102 cases between 87 and 94 of at
least one hour
sessions when the transmitter/exciter at the reception site
is coded
off(1)
and no other sites are transmitting and that the data is
good(consistent
with earlier and later dopplers and same average doppler
residuals?).
This would seem to clearly negate my claim and the need
to
test
the doppler shifts predicted by this assumption against the
observed
doppler shifts.

It would indeed.

The problem is that he does not want to provide the data
that shows
this. Maybe you could persuade him to be more forthcoming
and
to provide
an excel spreadsheet of, if not all 117 items for each of
the
hundreds of
thousands
of minutes for which this data exists,at least some of
this.

I'm not aware how much effort this would entail

If he has done what he says he has then he should be able to
back it up to some extent. Especially when what he says is
contrary to the general DSN policy and to common
sense. That is the efficient way of scheduling transmit and
receive
times to a specific spacecraft like Pioneer 10 is to transmit
and
receive at the same time from one site and then at a later time
when the return signal is expected at the site in view of the
spacecraft
for a sufficient length of time at this time, etc.
So why would they turn the transmitter off during some of
these
receive sessions for hours at a time? Craig says he is not a
mind
reader but there should be some logs or something to explain
this!

but in
many cases I too have gone to considerable effort in
writing descriptions and providing information, only to
have you ignore it or discount it saying I didn't
understand the subject. You have a certain knack for
annoying people so I can understand why Craig is
unwilling to do more.

Yes it is annoying when people point out one's mistakes.
and you and Craig also are human and make mistakes like
the rest of us mortals. And are a little obnoxious sometimes
too.
And we can all be irritable in such cases.
But at least you know I appreciate your expertise as I do
Craig's also.

Craig says that the data he
has from his programs applied to the doppler data files
proves that the transmitter at the receiving site was off
when the receiver was receiving good data.
I am questioning this. It could be transmitter/exciter off
does not mean that the transmitter is off in the transmission
system although as he points out this is not indicated in one
of the documents that briefly describes the two systems and
shows only one transmitter exciter.
This is a reasonable argument but not conclusive.
It could be that another transmitter at another site or the
same site is on but he says that he has data to show this
was not the case.
It could be that the code 1 in bit no. 237 that he
observes is misplaced by the coder(bit no. 236 is 1 when
the receiver loop lock indicator is 'out of lock' ) or Craig's
program.
( He acknowledges that the data tape is full of garbage that
had to be filtered out. Perhaps code '1' in no.237 is also
garbage.)
It could be that the data he says is good is not good
since the criterion used to determine 'good' is not stated.
I realize that such objections to one who is convinced
of the validity of the speed of light delay extrapolation, seem
absurd and not worth the time to examine.
He does not want to examine them, ok but he should provide
the data he claims and do so hopefully in an accessible form
eg an excel spreadsheet.


Looking back in the thread, I see
he gave three examples,



"Craig Markwardt"
wrote
in
message news
Yes, of course. Examples include 1987/07/15 17h,
1987/09/27
12h,
1987/11/04 17h, and so on (hours are rounded down).

so perhaps you could look at those. It's unlikely that
checking a larger sample would tell you any more than
those. Did you get your program to read the files yet?

I have read the files and extracted some of the fields
into
32 bit words but have not yet figured out how to format these
words
to import them to an excel file eg as numbers or ascii numbers
separated by commas
with an ascii code 13 (carriage return) to signify the end of a
row.
I did write a small program that did seem to work

main(int argc, char *argv[]){
char ch=13;// this asci number indicates end of line.
ofstream out("C:\\test1.csv");//the csv suffix means comma
separated variables
//the extra \ is needed to signify that the following \test1.csv
is correct.
if (!out){cout "cannot open file.\n";
return 1;}
out 10","123.23ch;
out.close();
return 0;}



Maybe you or some of your colleagues know the best way to do
this?.


It seems that you dont have to write the numbers in quotes and
that the
excel program accepts them so long as the commas are written as
ascii
characters and the end of line asci code is written at the end of
a row.(??)




If the meaning of the item 26 is exactly that the
transmitter power at
the receiving site was off during these hour sessions and
Craig has
really looked at transmitter records at other sites for the
same time,
then it would seem pretty clear that my claim is wrong.

I know Craig would not say they were off without
carefully checking first. He is a professional and
very careful not to claim anything he cannot back up.

It would be impossible to randomly mistakenly put in a 1
bit for 102
times
60 more different times which would indicate the
transmitter
was off
presumably and the
data received was good.Maybe the code '1' is given in the
next bit
position due to some sort of systematic error. This would
be
easy enough
to check since the next bit position should be zero (and
the
previous
position
if '1' would indicate out of lock and the error residual
would be large?)
Will check the data for 87.

The software that wrote these tapes and the programs
that analyse them have been used for decades for dozens
of missions. There is no chance whatsoever that such a
basic flaw would go unnoticed.


Nonsense. Craig is filtering out "garbage" now so it wasnt
'noticed' before
If a an inconistency in the speed of light assumption was
noticed
they may have changed the data to remove this in some cases
but not in others.




I dont see how you know it is due to a change in
phase
and
not to a change in frequency

Phase is the integral of frequency so you can't have a
change of one without a corresponding change in the
other,
but that is also the key to the operation of a PLL. Over
a
long time, the phase moves back and forth but it doesn't
drift systematically one way. Such a systematic drift
would
represent a frequency error so if there is only random
jitter in the phase, the _average_ frequencies are
identical.

So you are saying that the zero crossings of the
received
signal
may be a little late or a little early and if the size of
these
differences
is small enough then the local oscillator frequency is the
true frequency

Almost, I am saying that they may be a little late or early
but there must be the same number in the reference as in the
actual signal in a long integration therefore the frequency
is the true frequency.

But if such variations are larger than a specific fraction
of
the
local oscillator period on average which may include some
larger
than one period misses, then the local oscillator does not
represent a true underlying frequency.

Again, almost. If there is a very large error of more
than half a cycle time displacement for several cycles,
then the reference can 'slip' by getting one cycle out
of step. Before and after the frequency will be exact
but the total over the integration period will be out
by exactly 1Hz (due to the way the counters work) and
this is sufficiently obvious in the recorded results
that it could be identified by the analysts. This is
described in the Anderson paper.

You are saying I think that
upward zero crossing in the received oscillation that
correspond exactly or are not more than a quarter of
a cycle off and vary this way back an forth so that
when the local oscillator has had 10 upward zero crossings
that the received frequency has had 9 or 10 or 11 etc and then
is more than half a cycle later than that in the local
oscillator and this continues for a specific number
of cycles(eg 2) and then reverts back to the previous
pattern of upward zero crossings that is more similar to
that of the local oscillator then the total number of
such cycles in the receiver oscillation will be one cycle
less than the received and this is specified somewhere
(item 76 ,'total slipped cycles during count') ? as 1 cycle
or 1 Hz.

Craig's 2002 notes said he did not know how to interpret these
values
but it seems that this is how they should be interpreted.

If there is a systematic increase or decrease then you
have a
valid indication of a change in frequency consistent
presumable
with the motionsof earth and spacecraft.

Right, systematic changes such as the frequency drift
caused by the acceleration component of the motion of
the sites is tracked by the oscillator so does not
cause slips.

and I dont understand how many wrong zero
crossings and wrong lack of zero crossings you need
before you
conclude that you haven't locked onto a true carrier
frequency?.
Wrong means a local oscillator zero crossing that
does
not
correspond to an observed zero crossing.

The signals are normally in step, cycle for cycle. If the
phase error gets too big (think 1/2 a cycle), the loop
will
'slip' along to the next cycle and then lock on to that.
Have you ever handled corrugated roofing? Think of
sliding
one sheet over another. It is hard to pull up until it
has
moved half a corrugation then it easily falls into the
next.

Lets stick to what actually happens here. I think you
are making
it more complicated than necessary. When the reference and
local
oscillator frequencies match and
the zero crossings all correspond then the average of the
products of the amplitudes is
.5 and if there is no match the average of the products is
less
and closer to zero the worse the match..

The mean of the product is 0.5 if they are in phase,
-0.5 when 180 degrees out of phase and in general
0.5 sin(p) for a phase error of p.


If the peaks are 1 and -1 in the received and reference
oscillations then the average products of in phase oscillations
are
0 plus (+1)*(+1) plus 0 plus( -1)*(-1) divided by 4 = .5
and the average products of systematically out of phase
oscillations are -.5.
and the products of 90 degree out of phase oscillations is
zero.
So you can represent the average of the products as .5cos(p)
so when p=0 then .5 times 1 and when p=90deg then.5 times zero
and when p=180 deg then .5 times -1 etc.
If the received frequency is sometimes one frequency and
just
as often
another or sometimes the same frequency in one phase and just
as
often another
then we are receiving noise.
That is the average of the products of the cases above is
zero.
However if the average of the products is closer to .5 and
enough
closer to .5 so that it would be unlikely to result from a
uniform
distribution of the phase error "p" then we can with that
degree
of
confidence conclude our result is not noise.
I think we are getting closer to a correct statistical
formula
for the
degree of validity of the received signal.


Now a certain number of upward zero crossings or cycles
is needed to establish the phase. I suppose two is enough
and the third zero crossing would confirm the previous phase
or indicate some degree of variation etc.
So if you have two thousand zero crossings or a thousand
upward zero crossings this is like a sample of 999 and
the average of the products of reference and 10 received
amplitudes
every cycle is like a sample of 999 or perhaps 9990?

Provided the
system is receiving a valid signal the out of phase
condition cannot persist as the loop always pulls the
oscillator towards zero error.

If you multiply a sine wave by noise on the other hand,
then you are equally likely to get plus and minus results
so the mean will be zero. To get a steady positive value
requires two things

a) it must be tracking a real signal, not noise
b) the rms phase error must be within some small margin

So a first guesstimate of the chi square error sum of
squares
over the sum of squares around
the mean zero is that this ratio is 0 when the average of
the
products is .5
and that this ratio increases to 1 as the average of the
products
decreases to 0.

I wouldn't like to comment on chi square values, the
system only uses a simple mean of the products and
indicates lock when that is sufficiently high.



If this than you can perhaps associate a probability
distribution with
values between .5 and 0 and values above a specified
threshold
like .35.??

For a lock indication threshold of 0.35, the phase error
would be arcsin(0.35/0.5) which is about 45 degrees. If
the reference signal is at exactly the same frequency as
the craft signal and remains in les than 45 degrees phase
error for several seconds then the lock indicator would go
on. The wrong frequency or too poor a signal-to-noise ratio
or too much phase error would all switch it off.

Please remember this value is only illustrative, I don't
know what would be used and it may even be dynamically
adjusted.





Happy New Year
George



Maybe we can resolve all of this in the new year
Happy New Year to you too
Ralph

"ralph sansbury" writes:
"George Dishman" wrote in message
...

"ralph sansbury" wrote in message
...
[ ... ]
The problem is that he does not want to provide the data that shows
this. Maybe you could persuade him to be more forthcoming and to provide
an excel spreadsheet of, if not all 117 items for each of the hundreds of
thousands
of minutes for which this data exists,at least some of this.

I'm not aware how much effort this would entail

If he has done what he says he has then he should be able to
back it up to some extent.

Your statement is ironic, since I have backed up my arguments with
facts and numbers, more numerous to describe in detail here.

You really are in no position to say what I "should" do, especially
since I have already accomodated you in a number of ways. Examples:
analysis with removal of assumption of propagation speed c; or
assuming light travel time is 1-2 s; both of which failed miserably.
Meanwhile you continue to speculate on wildly implausible scenarios.

If you do not trust my results, that is your right, but then don't
*also* expect help from me. You have access to at least two years of
raw data: go ahead and analyze it yourself if you like.

Especially when what he says is
contrary to the general DSN policy and to common
sense.

This is not a DSN policy. Indeed, it is only efficient to transmit
when another downlink session is scheduled. When none is scheduled,
there is no reason to transmit, and so it is actually *inefficient* to
transmit. Your supposed "policy" fails when one considers there is
more than one spacecraft served by the DSN, so no one spacecraft can
schedule all resources for itself.

CM