# Intro:

On,off position switching is done to remove standing waves from the resulting spectra. Tracking the same part of the dish for the off should have the same "ground based" standing waves as the on position (this will not work as well for energy that comes from the sky). Typically people will track a source for 5 minutes, and then spend the next 5 minutes retracking the same portion of the dish ( which is an ra/dec position offset  by 5 minutes + the move time).
If the off position does not  match the on position track exactly, how fast does the baseline degrade? This is the question that is looked at here.

# Setup:

project a2340 tracked a 20 degree dec source rise to set for multiple days using the cband receiver. The setup was:
• 7 172 bands covering  4200 to 5200 MHz were used. 172/1024 = 172 kHz resolution was used.
• 900 second scans  at .9 second integrations were taken. There were 9 complete scans.
• The mock spectrometer took full stokes data with 32 bit dumps.

# Processing:

Each 900 second scan was used to emulate on,off position switching.
•  The first N samples were taken as the On position while the next N  samples were the off position.  The offset between the on and the off tracks on the dish where then N*source motion in .9 seconds.
• The offsets used (N) were:
• 1,5,10,20,25,50,100 samples corresponding to great circle angular offsets of .004,.02,.04,.07,.09,.18, and .35 degrees.
• Each  trial had 450 seconds on and 450 seconds off.
• Since the same source was tracked, the main beam power should not have changed.
• For each scan:
• compute the on/off for the 7 different offsets.
• Do this for each of the 7 frequency bands.
• Compute the baseline rms for each on/off spectra (throwing out outliers).
• Record the   measured/expected rms (from the radiometer equation).
• Record the   acf of on/off to get the power vs delay to see if the 1 usec standing wave is present.
The plots show the baseline rms vs on,off distance offsets (.ps) (.pdf):
• Page 1: rms vs great circle offset.
• Each frame is a different frequency band
• black is polA, green is polB.
• The rms starts to increase at .2 degrees. Part of the spread is from rfi.
• Page 2: rms vs little circle offset.
• These are the same plots as page 1 except they are versus little circle offsets (azdif is not multiplied by  sin(za).)
• The offsets do not group as well as the great circle offsets. This is a little strange since ground based radiation might be more sensitive to little circle offsets than great circle.
The image shows the acf's for on/offs (.gifs):
•  The 7 different on,off offsets are separated by dashed lines (the bottom has the smallest offsets).
• Within each set, the acfs for the  5 frequencies (4275 and 4405 had rfi), and 9 scans are shown.
• You start to see a vertical line around 1.03 usecs start in the 4th section (.07 degree offsets). It is  strongest at the top section (.35 degrees).
• A second  reflection is seen at .9 usecs in the top section.
• The lower delays (0-.2) start to have power i the 3rd section.

# Summary:

• 450 sec on,offs which track different parts of the dish were  constructed from a 900 sec track of a source.
• great circle offsets of .004,.02,.04,.07,.09,.18,.35 degrees were tested.
• The rms of the on/off starts to increase around .2 degrees.
• The rms increase correlates better with great rather than little circle offsets.
• the 1 usec standing wave starts to be seen in the .07 degree offsets.
• a .9 usec standing wave is seen at .35 degrees offset.
• The rms increase is probably also a function of az,za and not just the offset distance. It probably correlates with the telescope symmetry.
• standing wave at cband have the same period as lband. The phase moves 4 times faster as the distance changes. It would be a good idea to repeat this at lband.

processing: usr/2327/dec09/chkonoff.pro
`home_~phil`