Discussion Forums

baudline analysis of Voyager 1 redux

39 replies [Last post]
sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733

The SETI Institute has spotted the NASA Voyager 1 space probe again. I analyzed this new Voyager 1 data with baudline and blogged about it:

http://baudline.blogspot.com/2012/11/setiquest-voyager-1-redux.html

The data format is different and the signal is weaker than it was when I first analyzed the Voyager data last year.  I look at both polarizations and use a number of tools to analyze the data.  This time the Voyager signal is drifting at -0.58 Hz/sec.

So far away, so cold, so alone. What is it about that little space probe that warms my heart?

jrseti
jrseti's picture
Offline
Joined: 2010-07-22
Posts: 250
Beautiful!The reduction in

Beautiful!

The reduction in signal strength from several days ago may be due to the proximity to the sun? This week (noon-ish on Nov 28, 2012) SonATA was not able to detect Voyager. At the time of observation the Az of the Sun and Voyager1 were the same, and the elevation just 35 degrees different.

To bad we can't gather enough signal to try and decode the signal!

What about Voyager2? Is that a valid target? It is much closer. I'll have to ask the powers-that-be here for their opinion.

-jrseti

jrseti
jrseti's picture
Offline
Joined: 2010-07-22
Posts: 250
Jill has informed me that

Jill has informed me that Voyager2 is too far South to view. Dang!

-jrseti

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
Is Voyager 2 too far south

Is Voyager 2 too far south just this time of the year or all year round?

robackrman
Offline
Joined: 2010-04-15
Posts: 235
voyager 2 declination

Unfortunately, Voyager 2 currently has a declination of -55 degrees which is growing more negative with time.  Given the ~ 40 degree latitude of the ATA, Voyager 2 is always below the horizon.  When you add in the ~ 18 degree artificial horizon of the ATA instrument, the situation is hopeless.

jctarter
Offline
Joined: 2010-08-27
Posts: 17
what warms your heart?

maybe its Voyager1's RTG.wink
thanks for this analysis!
remember that what we are detecting here is the carrier wave from the spacecraft. the information content is in the modulated sidebands several (can't remember) kHz away.
jill

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
RTG = Radioisotope

RTG = Radioisotope Thermoelectric Generator

Yeah, Pu-238 has that special warming quality. (:

Listening to the Voyager 1 press announcement today on http://www.ustream.tv/nasajpl2 I heard:

  • Voyager's Pu-238 power plant is losing 4W/year due to half-life decay.
  • Voyager's instruments will run out of power by 2020.
  • JPL via the DSN talks to Voyager 1 & 2 each day.
  • To communicate they use a combined 70m & 34m dish array or multiple 34m dishes.
  • Voyager is transmitting at a 110 bits/sec rate.
  • Voyager has entered a new region in deep space.
  • "The plasma just sits there like a swamp."

I love that last quote. So Voyager's radio signal is bogged down in a plasma swamp. That could explain some of the signal attenuation.

robackrman
Offline
Joined: 2010-04-15
Posts: 235
voyager telecommunications

I found this interesting resource:

http://descanso.jpl.nasa.gov/DPSummary/Descanso4--Voyager_new.pdf

I may spent a bit of time trying to determine if it would be possible (with a setiQuest style observation) to demodulate and see some structure.

jctarter
Offline
Joined: 2010-08-27
Posts: 17
thanks for this resource

this document is very informative - thanks for passing it along.
jill

gerryharp
Offline
Joined: 2010-05-15
Posts: 365
nice analysis

Hi Siblips

That was a nice characterization you gave for the Voyager signal. you have a great many tools for detailed anaysis.

As for the broadnening, the very last FFT uses a Hamming window to reduce edge effects. This is also why a narrow band signal will always have a small amount of power in adjacnet bins.

If you could recount the power you've seen on the signal as a function of day, this might help us correlate with various disruptions of the array on those dates. The voyager antenna is 3.5 m in diameter and the transmitter had 25 W when it was built (probably less now). So we can put an absolute value on your power measurements on Voyager.

Thanks a and keep thinking!

Gerry

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
The signal from the 2010

The signal from the 2010 Voyager data is significantly stronger than it is in the recent 2012 data. You're right Gerry, I was comparing apples to oranges. The sample rates are different. The Hz/bin resolution is different.  Other parameters in my analyses were different too.

I've gone back and re-ran them through baudline with the goal of making the comparison as equal as possible. Here are two Average plots with Auto Drift enabled, the purple and green curves are the two polarities:

2010 notes and measurements:

  • Strongest 2010 peak is +6.04 dB above the noise floor.
  • Drift rate -0.47204 Hz/sec
  • 546133 sample rate
  • Resolution 0.26042 Hz/bin
  • Duration is about 140 seconds.
  • 16-bit samples decimated with gain to an effective 24-bit samples (ENOB).
  • The gain offset for the green polarity (X) is about +0.5 dB higher than the purple polarity (Y). Why?

2012 notes and measurements:

  • Strongest 2012 peak is +2.43 dB above the noise floor.
  • Drift rate -0.57733 Hz/sec
  • 711.111 sample rate
  • Resolution 0.17361 Hz/bin
  • Duration is about 92 seconds.
  • 4-bit samples (compamp file)

Comparison:

  • The 2010 signal is +3.61 dB stronger than the 2012 signal.
  • I equalized the two different Hz/bin resolutions so they are not a factor.
  • The 2012 signal has a faster drift rate but that should not be a factor with Auto Drift.
  • Despite the 2010 signal having a 50% larger duration it's variance of the noise floor seems less than it should be. Remember variance is proportional to sqrt(N); The larger effective bit quantization could be playing a factor.

Is a +3.61 dB difference more than should be expected? What is the possible cause? Could it be the plasma swamp? Or is it just the sum of a bunch of smaller causes?

robackrman
Offline
Joined: 2010-04-15
Posts: 235
I very much enjoyed...

I very much enjoyed reading your Voyager observation analysis post.  I will dust off my Doppler code to see if a prediction (if I can remember how to run it) matches with your measurement.  Just for fun.
More later...
Rob

robackrman
Offline
Joined: 2010-04-15
Posts: 235
doppler calculation

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
Hey Rob, that is really cool.

Hey Rob, that is really cool. Your -0.566 Hz/s calculation is very close to my -0.5770 Hz/s Auto Drift measurement.

That's a 2% error which isn't bad.  If I knew someone was going to check my work I would of spent some extra effort to make a more accurate measurement! (:

I think if I use a larger FFT and tweak Auto Drift's overlap and quality parameters I can improve my drift measurement accuracy. I'll let you know how it goes. Thanks for calculating and posting that plot.

robackrman
Offline
Joined: 2010-04-15
Posts: 235
error

Some, or all, of the error may be on my side.  The Doppler routine I used is for an RA/Dec object on the celestial sphere - not for a solar system object (which is a more difficult problem).  I used the RA/Dec of Voyager at the time of the observation.  That is close, in this case, but not the correct frame of reference.  I don't know if Voyager has a significant (with respect to our difference) acceleration that should be included.  Let me know if you get an improved number from your analysis.  I will do some more investigation on my side.  Also, the df/dt (which I predicted as the instantaneous value at the beginning) will change a bit during the course of the observation.  Can you confirm for me the exact time window of the observation?

gerryharp
Offline
Joined: 2010-05-15
Posts: 365
voyager errors

Hi

You probably noticed that the signal is relatively weak in the data. I noticed that the signal appears to hop around in frequency as a  function of time. As if it were scintillating. Or maybe the transmitter is dying.

Lets say the signal bounces around randomly over 2 Hz. Over a 100 s observation, the error on the drift calculation would be 2/100 or 2% of the nominal drift. So the noise-induced error could be in the second digit of the measured slope.

Gerry

robackrman
Offline
Joined: 2010-04-15
Posts: 235
re: voyager errors

Right! I agree that determining the weak signal slope may be the significant explanation for the ~ 0.01 Hz differences between baudline, SonATA, and predicted Doppler.  Another item to add to your list (stability and scintillation) is that random noise - given that the signal is so weak - conspires to look like the signal for a short run of pixels. There is also a slight curvature which we are not accustomed to (is not as significant) at lower frequencies.  A ~0.01Hz/s difference does not even add up to a pixel (bin) width (at 1Hz resolution) over a 90 second observation.

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
-0.01 Hz/s drift anomaly

Rob you're correct, there are a lot of factors that could be the source of this -0.01 Hz/s drift anomaly. To test that this drift value is correct I re-ran the recent Voyager 1 data through baudline with an increased 0.17361 Hz/bin resolution and used a Manual Drift correction feature. At that resolution an error of one bin over a 93 second duration is 0.00187 Hz/s which is more than accurate enough to resolve this issue. Baudline's legacy Manual Drift feature was used to eliminate any possible error caused by the far more complicated Auto Drift algorithm.

For this experiment I manually dialed in two drift rates. In red is the -0.577 Hz/s drift rate and in cyan is the -0.567 Hz/s drift rate. Baudline sums up these two drift vectors and the -0.567 Hz/s rate is 0.48 dB weaker.

So I'm convinced that the -0.577 Hz/s measurement is accurate and is not caused by some numerical error such as line dithering or algorithm error.  There are many factors remaining that could be the source of this -0.01 Hz drift anomaly but measurement error of the compamp data is not one of them.

What is the significance of this  -0.01 Hz/s drift anomaly?

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
The only timestamp I have for

The only timestamp I have for the Voyager observation is encoded in the 2012-11-07_21-14-05_UTC.act61.dx1011.id-4.R.archive-compamp file name. I don't know if that is the start or the end of the 93 second data collection.  Jane or Jon may have more accurate time information.

I spent some effort adjusting baudline's controls and trying different Auto Drift algorithms attempting to make a more accurate drift rate measurement. I was unable to do so. My thinking was that the number of possible Auto Drift paths (vectors) was limiting accuracy but that was not the case.  It seems that low SNR is the limiting factor. I am going to stand by the initial -0.5770 Hz/s measurement and give it an error range of ±0.001 Hz/s.

janebird
Offline
Joined: 2010-04-22
Posts: 20
voyager act 61 start time

Here are the statistics for the SonATA detection of voyager for this activity.

Activity Id: 61
Frequency: 8419.628109
Drift: -0.5716677
Width: 0.01085
Power: 24.4
Subchannel: 1157
Activity Start Time: 2012-11-07 21:12:22
PFA: -24.39
SNR: 0.1215

Note that the time in the name of the archive-compamp file is the time the file was created which is at the end of the signal detection step.

robackrman
Offline
Joined: 2010-04-15
Posts: 235
using another method of Doppler prediction

Doppler prediction was made for this plot (below) from JPL Horizon range and velocity data.  This is independent from the previous method of Doppler prediction which had its own internal engine.  The results agree to within 0.001 Hz/s.  The plot is zoomed this time down to a two hour window roughly centered on the observation.  The sigblips measured value, according to this prediction, occurs 17 minutes later.  Jane's value from Sonata -0.572 lies halfway between the two (not shown because I saw Jane's post after I made the plot).  There is a curvature of ~ -0.0015 Hz/s/min,  during the observation.  I don't know if this is large enough to have an affect on SonATA or baudline slope algorithms.  This was fun. 

Here are the JPL Horizon parameters:


Here is the calculation:

sed -n '/\$\$SOE/,/\$\$EOE/ p' 2012-11-07-voyager-1-horizons.txt | grep 2012-Nov-07 | cut -c 83- | awk '{if(NR>1){print NR+1200,8.415e9*(((($1-x)*1000)/3e8)/60)};x=$1}' > 2012-11-07-voyager-1-horizons-doppler.txt

And, the plot:

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
Hi Rob. Hmmm, a 2% error

Hi Rob. Hmmm, a 2% error isn't bad but I expected better performance with baudline's Auto Drift algorithm so I created a test suite of drifting tones of varying SNRs. With equal durations, bin resolutions, and spectral variances I'm able to measure the drift rate within a 0.2% error using a test signal that is 1 dB weaker than the Voyager signal. That's an order of magnitude difference so it's not baudline and something else is going on. For note, increasing the duration and SNR can have a significant effect, drift rate error can drop to 0.01% which seems to be the error floor for the Auto Drift algorithm.

So what could be going on?

  • Synthesized test signals are very different than weak signals from outer space.
  • Drift curvature, the signal starts at -0.5670 Hz/s and ends at -0.5693 Hz/s.
  • Baudline's compamp signal source has 4-bit quadrature samples (~2-bits effective). SonATA's packet signal source has 16-bit quadrature samples.
  • We are near the limit of what can accurately be measured: ±1 bin error * 1 Hz/bin / 100 seconds = ±0.01 Hz/sec error

What is interesting is that the measurement accuracy limit can be increased by an order of magnitude with a synthesized test signal and by another order of magnitude with a significantly better SNR.

Even more interesting is by copy-n-pasting sections of the Voyager data into baudline's Average window I was able to see the drift curvature you mentioned. It was noisy but it tracked the Hz/s curvature rates above but with the same -0.01 Hz offset I've been seeing.

robackrman
Offline
Joined: 2010-04-15
Posts: 235
baudline experiments

Those are good tests/experiments.  I like that you were able to see the curvature by breaking the observation into sections.  It seems like there may be an unexplained small drift in the Doppler calculation (missing term) or in the detection.  Or, a time offset?  It is interesting, however, that the baudline and SonATA drift measurements differ by ~ 0.005 Hz/s (50% of the Doppler prediction to baudline difference) on the same observation data. 

gerryharp
Offline
Joined: 2010-05-15
Posts: 365
very cool!

I'm very impressed with both Sigblips' and Rob's analyses. I'm learning a lot.

The plasma swamp that Sigblips mentioned is certainly a possible explanation for signal broadening. Voyager's frequency will bounce around when it is travelling through clumpy plasma. I wonder if this could be simulated (variable frequency of arrival vs. time).

Here is an interesting thought experiment:

Take a triangular prism made of non-dispersive material, set it on a horizontal piece of flat glass. Starting at the middle of the prism, shine a laser though the prism and measure the light frequency below the glass with a spectrometer. Now, slide the prism with constant velocity horizontally. If you slide such that the skinny end of the prism comes under the laser, then the light will pass through the prism in a shorter time due to the smaller amount of dielectric. As a result, the light waves arrive more rapidly at the spectrometer than they would normally, causing a blueshift. Slide it the other way, and the light spends more time in the prism and you have a redshift.

I think that is right.

Gerry

robackrman
Offline
Joined: 2010-04-15
Posts: 235
accuracy of ATA rubidium time standard tested by Voyager obs ?

At the time I left the SETI Institute, the ATA LOs were trained by a Stanford Research rubidium oscillator with specifications like (or perhaps the same model) shown below.  Is that still the case?  If so, I find Gerry's observation that the Voyager signal appears to "hop around in frequency" interesting because an ~ 8 GHz signal must maintain 11 digits of significance to stay within a 1Hz bin resolution but the rubidium oscillator has a stability of only order 2e-11 over 1 second (~ 2/3 of the time).  This appears to be marginal.

robackrman
Offline
Joined: 2010-04-15
Posts: 235
something interesting

I noticed in Jane's post that SonATA detected the Voyager carrier at 8419.628109 MHz.  This is the Doppler shift from the 8420.432097 coherent downlink frequency (see table below).  It appears that at the time of the ATA observation Voyager 1 was returning a phase-locked signal in response to an up-linked signal from Earth. 

I assume the up-linked signal is Doppler corrected (frequency adjusted) such that Voyager always receives the same frequency signal from Earth, however,  I can't find a document with that level of detail.  I wonder if the 0.01Hz/s offset we seek might have to do with the Doppler difference between the uplink station location and the ATA?

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
I remember reading that the

I remember reading that the DSS (Earth station) corrects for the Doppler drift prior to transmitting. Giving this responsibility to the DSS makes a lot of sense because it simplifies the logic and circuitry on Voyager. So if Voyager 1 was in the two-way coherent tracking mode then it's possible that the +0.01 Hz/sec offset was caused by the DSS.

Do we know if Voyager was in this mode?  Or are we looking at an idle carrier?  How chatty is Voyager?

A square wave would have its first harmonic down -9 dB from the fundamental and I doubt we could see that with only a 93 second integration.

robackrman
Offline
Joined: 2010-04-15
Posts: 235
A square wave would have its

A square wave would have its first harmonic down -9 dB from the fundamental and I doubt we could see that with only a 93 second integration.

see http://ipnpr.jpl.nasa.gov/progress_report/42-111/111Q.PDF

I noticed the first Voyager 1 analysis you did, from the SonATA distribution, is from data captured at a 546133 sample rate.  Depending on the tuning (if roughly centered) it may be possible to recover much of the 25.5 kHz sideband energy.

The second, most recent observation, is hopeless.

If you were to look at that (previous wide-band Voyager 1 SonATA) data again, at reduced resolution (say 100 Hz / bin) the largest sideband harmonic, which may be spread by 160 bps telemetry BPSK modulation, might pop out of the noise (because the energy would be collected into one, or a few, bins).  But watch out for the narrow carrier which at the reduced resolution might fall into the collected sqrt(n) noise.

 

gerryharp
Offline
Joined: 2010-05-15
Posts: 365
problems with uplink

If we're having trouble seeing Voyager because of scintillation, then Voyager will have trouble receiving signals from Earth for the same reason.

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
While scintillation is a

While scintillation is a two-way problem I don't think it is a problem for the uplink. Here is why:

  • basically unlimited power on the 70m DSN uplink
  • if needed the DSN can array the 70m dish with multiple 34m dishes
  • uplink data rate is 16 bps while the downlink rate is either 110 or 160 bps = 9 dB gain

It is not a symmetrical problem.

YNOT2014
Offline
Joined: 2012-12-13
Posts:
How long until Voyager 1

How long until Voyager 1 reaches it's first system? Is it in the foreseeable future or is it far beyond our generation?

robackrman
Offline
Joined: 2010-04-15
Posts: 235
Voyager 1 is traveling at ~

Voyager 1 is traveling at ~ 1/10000 the speed of light.  The nearest star system is > 4 light-years distant.  So, sadly, the answer is tens of thousands of years in the future (if it were aiming at the nearest star -- which it is not).  The Voyager power supply is predicted to diminish to the point where the craft can no longer communicate within ~ 10 years.

jrseti
jrseti's picture
Offline
Joined: 2010-07-22
Posts: 250
Is Voyager1

Is Voyager1 transmitting?
We've have not been able to detect the Voyager signal the last 4 weeks. Internally there was some discussion as to wether this could be due to Voyager1 not transmitting all the time. Does it transmit all the time?
We've been thinking of trying to decipher the information available at http://voyager.jpl.nasa.gov/mission/soe-sfos/integrated_2013.html to see if Voyager is transmitting or not.
The PDF documents they supply have a lot of unfamiliar acronyms. I fould this document that helps: http://www2.jpl.nasa.gov/basics/glossary.php
Any thoughts?
 
 

-jrseti

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
Their twitter stream is

Their twitter stream is posting light travel times which is a round trip measurement which means that JPL is successfully communicating with Voyager.

A big unanswered question is what exactly has SonATA been detecting? The signal is clearly not modulated. The on-line Voyager docs make no mention of what the signal SonATA is seeing might be. This is a mystery.

jctarter
Offline
Joined: 2010-08-27
Posts: 17
Voyager 1

we have never attempted to detect the modulated sidebands from Voyager as far as i can remember (though we did that with Pioneer 10 at Arecibo). we are simply trying to detect the single frequency carrier tone. i'm guessing that the carrier is always transmitting, although it may be suppressed whenever DSN links up and downloads data.
jill

robackrman
Offline
Joined: 2010-04-15
Posts: 235
I notice in the Voyager(s) Sequence Of Events logs referenced by

I notice in the Voyager(s) Sequence Of Events logs referenced by Jon many entries for Voyager I (Spacecraft ID 31) similar to this:

FOR CHG IN X-BD D/L CONFIG
MOD INDEX=41 DATA RATE=HIGH
SUBCARRIER FREQ=22.5KHZ

The higher the modulation index (MOD INDEX) the more power is removed from the carrier into the sidebands. A high data rate (DATA RATE=HIGH) means 160 bps for the current ageing and distant state of the spacecraft (according to the paper I linked to earlier in this thread).

Perhaps when Voyager is in the high data rate mode the carrier is below the noise, but when the transmission is in the low data rate 40 bps telemetry mode (again, see paper), the carrier is detectable (at the ATA)?

gerryharp
Offline
Joined: 2010-05-15
Posts: 365
thanks

Thanks to everyone for helping us track this down. Voyager is about the best fiducial we have, when it works.

As we're upgradidng the feeds, a year from now seeing Voyager might be a piece of cake!

Gerry

jctarter
Offline
Joined: 2010-08-27
Posts: 17
carrier suppression

Rob - thanks for the reminder about the earlier paper you posted with the detailed info on Voyager. This thread is now pretty long, and i'd lost track of that. Jane Jordan was going to collect date/times of successful detections so we can compare with the archived segment logs - that should allow us to decide whether this is indeed the explanation. since these are published at least a few weeks in advance, it should also allow us to schedule observations when they are more likely to be successful.
jill

robackrman
Offline
Joined: 2010-04-15
Posts: 235
Jane's SonATA Voyager 1 Detections

Date/Time for failure to detect Voyager 1 would also be useful for this study.

robackrman
Offline
Joined: 2010-04-15
Posts: 235
Voyager 1 Detection Attempts