327 Mhz birdies and their Azimuth dependence

Sections:
Intro
Processing the data
Dynamic spectra of the azimuth swings
Average/PeakHold spectra for each az swing:
The time dependence of the birdies
The azimuth dependence of the birdies
Length of the scattering elements
Summary

---

Introduction (top)

• The azimuth was moved between -90 and 270 degrees at .3 degrees/second. This was done 5 times (each direction counted as 1 swing). A single swing took 1200 seconds.
• A 25 Mhz bandwidth was centered at 320 Mhz with 2048 channels giving a 24 Khz channel width (after hanning smoothing). The data was sampled once a second.
• The alfa motor controller was turned off 2 minutes after the start of the first swing. The dome air conditioner units were left on.

---

Processing the data: (top)

• For each azimuth swing a bandpass was constructed from the 1200 spectra. A fit to a bandpass was needed rather than an average since some rfi are constant and would remain in the bandpass average. The bandpass was created by:
• Sort the 1200 points by their total power and then take the 240 smallest points (20%). These should have the smallest contribution from the sky. Compute a median bandpass from these 240 points.
• Fit a 12th order harmonic and a linear polynomial to this median bandpass using corblauto(). This iterates the fit throwing out outliers.
• Divide each of the 1200 spectra by this fitted bandpass. This should remove most of the IF bandpass.
• For each spectra compute the median total power and subtract it. This should remove the source contribution.
• A residual ripple remained so a 6th order polynomial (using a robust fit that threw out outliers) was fit and removed from each spectra.
• PolA and polB were averaged.
• The entire data set of 5 az swings (2048 channels * 1200smps * 5 swings) was checked using a robust rms. It  iterated throwing out outliers and ended up using 86% of the points. The routine computed an rms/mean of .01. The expected value was .006. So the bandpass correction was ok.

---

Dynamic spectra of the azimuth swings:  (top)

•  Swing 1 -90 to 270 degrees (.gif) :
• There are a number of frequencies that are time variable and come and go together. They all have modulation on them. The strongest of these are:
• 308.7, 309.9, 311.07, 313.1. 314.3, 317.5, 318.9.
• Some other signals that are modulated and tend to be there all the time are: 312.31, 320 , 321.9 Mhz being the most prominent.
• The radio was tuned into the IF signal and the following radio stations were heard (in the IF):
• 312.31 is the 3rd harmonic of 104.1 (FM).
• 321.9 is the 3rd harmonic of 107.3 (FM). This is the strongest fm station.
• 317.5 was heard to be 107.3 (FM). This may mean that the other birdies (308.7,309.9.. ) are also intermods of 107.3 since they turn on and off with 317.5 .
• The wiggling line starting near 315. Mhz at the bottom of the plot is probably the ac unit birdie.
• The birdies at 320.9, 328.1 that start at the bottom of the plot and turn off around az of -50 are from the alfa motor controller. It was turned off at az of -50.
• The large narrow spikes (327 Mhz 30*Tsys and 317.4 20*Tsys) only lasted for a few seconds. The spikes that are broader (and show the modulation ) were there more often.
• You can see a number of weak combs in the image. They will appear stronger when the channel width is narrowed (since they tend to be narrow).
•  Swing 2 270 to -90 degrees (.gif) :
• This image was taken top to bottom in time.
• There are 3 sets of azimuth where a number of birdies occurred together.
•  Swing 3 -90 to 270 degrees (.gif) :
•  Swing 4 270 to -90 degrees (.gif) :
•  Swing 5 -90 to 270 degrees (.gif) :
• There are about 7 sets of azimuths where the a number of birdies occurred together.

---

Average/Peak spectra for each az swing:  (top)

    For each azimuth swing an average spectrum (over the 1200 spectra) was computed as well as a peak hold spectrum (the largest value in each frequency bin was kept). Colors were used to differentiate between azimuth swings. The plots show
the average spectra for the az swings (.ps) (.pdf):

• Top avg.: The average spectra for each swing. The largest birdies are at
• 308.351 (16*Tsys)
• 313.51    (3.8*Tsys)
• 326.97 (.88 Tsys)
• 2nd avg.: This blows up the vertical scale on the top plot
• 3rd Pk hold: This has a peak hold for each az swing (take the largest value in each freq channel from the 1200 spectra). The largest values are:
• 308.352 (1600*Tsys) this signal has gone off the vertical scale
• 326.97 (300*Tsys) this signal has gone off the vertical scale.
• 314.3    (19.3*Tsys).
• 4th Pk hold blowup: The vertical scale on the 3rd plot is blown up.

The time variability of some of the signals caused a large difference between the peak hold and the average values (308.35 peak: 1600, average: 19).

---

The time dependence of the birdies:  (top)

    The plots show the birdie power versus time for the 50 strongest birdies (.ps) (.pdf): Some of the signals have gone off the vertical scale (check the peak hold plots above for the max values).

    Some of the birdies look  periodic in time.  This could be a true time periodicity or it could be an azimuth periodicity (since we went thru the same azimuth 5 times). The peaks that are symmetric about a dashed line are most likely a function of azimuth. The dynamic spectra (above) show that a large number of birdies are occurring at the same time.

---

The azimuth dependence of the birdies:  (top)

        17 signals were found to have a definite azimuth dependence. For each of these signals, the 5 azimuth swings were over plotted (using a different color) versus azimuth angle. For a true azimuth dependence the 5 strips repeated with azimuth.
    Some of these birdies at different frequencies showed the same azimuth dependence. They are either the same signal or coming from the same tower.

    The plots show the azimuth dependence:

• The 17 birdies that show an azimuth dependence (.ps) (.pdf). Each frequency is in a separate plot. The table below summarizes the values.
• The 10 different azimuth dependencies (.ps) (.pdf) :  The above plot showed that there were 10 separate azimuth dependencies. The 5 az swings were averaged for each of these and then plotted versus azimuth.  Each of the frequencies was normalized to its maximum value (so they would all fit on the same plot). Some azimuths positions that could be significant are:
• (2.86, 122.86, 240.86) : Corners of the triangle with the main and tiedown cables.
• (62.86, 182.86, -57.14):  Azimuths perpendicular to the sides of the triangle.
• ([-87.2, 92.86], [32.86, 212.86], [152.86, -27.14]):  These directions are perpendicular to the main cables (T12,T4,T8).

---

Length of the scattering elements:

---

Summary:  (top)

• The dynamic spectra show that there are many birdies that varying in time  together. They look like intermods in the system. A peak hold of about 20 seconds was done on the output of the 327 preamp (before the filter). The power levels that were measured should not be driving the 1st amp into compression. The time variability in the dynamic spectra show that maybe 20 seconds was not long enough.
• 17 signals were found to have a definite azimuth dependence. Some other signals looked to have an azimuth dependence but not for all of the 5 swings. This was probably because they were time variable.
• Of the 17 az dependent signals, 10 of them had independent azimuth profiles.
• The signals with azimuth dependence showed an oscillation in strength near their peaks. The spacing of this oscillation measures the length of the element that is scattering the radiation into the telescope.
• If would be a good idea to measure the azimuth dependence of known signals. We could put a transmitter in the control room, visitor center, optical lab, etc. and then do some az swings looking at these birdies. It will probably be different for different transmitting antennas and different recievers (this can be tested).
• The table below summarizes the info on various birdies.

home_~phil