# Reflector errors: ruze curves.

#### mar 2009

dec04-may04 data
jul08 - feb09 data
mar11 - mar12 data
nov12 - aug13 data
aug13 - jan14 data
jan14  - mar15 data

# Intro

The error in the reflectors will cause the gain to decrease as the wavelength decreases. The ruze formula tells how much this is:
• Gain=Gain0 * exp(-(4*pi*delta/lambda)^2)
• Gain0 is the gain for a perfect reflector (or long wavelength).
• Delta is the rms surface error (which should be random)
• Lambda is the wavelength in the same units as delta.
There are a number of factors that  can affect these measurements:
• The gain is computed using the cal values.If the cals are incorrect, then this will shift the gain on the gain x frequency plot.
• The gain uses the source flux. We use  chris salters fits to flux vs frequency. Some of these fits start to degrade at higher frequencies.
• At lower frequencies, gain errors will shift the surface error a lot (since the gain does not change much with surface error).
• At 5 ghz 10% gain change will move the surface error by about .5 mm
• At higher frequencies, the source extension can start to make a difference
• at 9.2 Ghz the fwm beamwidth is about 30 asecs.
• When comparing the gain of different frequencies, we assume that the different horns illuminate the tertiary equally.
• The optical collimation errors will affect the gain of the higher frequencies (pitch, roll, focus errors of each receiver).
• some of the errors are common between receivers (dome tilt).
• Other will differ by horn (horn position , tilt relative to the tertiary.
• We need to pick a Gain0 (long wavelength where surface errors do not affect the gain) to anchor the fit.
• an antenna with 2762 m^2 is 1 kelvin /Jy
• A 225 spherical aperture has 39760 m^2 giving  14.4 K/Jy.
• This assumes a uniform flux across the area. The edge taper will decrease this:
• (12K/Jy) / (14.4K/Jy) = .83 aperture efficiency
• (11K/Jy )/  (14.4K/Jy)=  .76 aperture efficiency.
• Originally  i had varied G0 for each data set.. so that the curves passed through the highest gain..
• This made comparing things over time difficult.
• Looking at 2014-2015 gain for the 430 dome gave around 11K/Jy.
• I then decided to use this as G0  for all the datasets...
• This assumes that that the illumination of the other horns are similar to 430 (edge taper, etc.. probably not)
• The dashed color lines on page 2 plots are the gain vs frequency from the ruze equation using various surface errors:
• black: rms= 1.5 mm
• red: rms=2.0 mm
• green: rms=2.5 mm
• blue: rms=3.0 mm
• purple: rms=3.5 mm
• for each data set i've included some things that occurred during the time frame.
• In some instances the occurrences defined the start and end of the data set (platform beam breaks, earthquake damages main cable, etc..)
You should probably take these measurements with a grain of salt..
• they probably tell more about the different cal value errors, and illumination patterns than they do about the surface error.
• the higher frequencies measurements should be used to estimate the surface error.

# Ruze curves: data dec04 - 09may05  (top)

The measured telescope gain from 01dec04 to 09may05  was used to compute the ruze curves.
This was after the adjustment of the primary reflector.

• 1411 gain measurements were used (lband thru xband).
• For each frequency, the median gain value for za=[5 to 15] degrees was plotted.
• Gain0 was taken to be 11.K/Jy. This value was picked as the long wavelength value (see intro above).
• The dashed color lines are the gain vs frequency from the ruze equation using various surface errors:
• black: rms= 1.5 mm
• red: rms=2.0 mm
• green: rms=2.5 mm
• blue: rms=3.0 mm
• The 2.5 mm comes closest to the data. Some of the errors in the data are:
• The lband curves do not lay close to the curves..
• cbh (6-8 Ghz) under illuminates the tertiary (so the gain is low).
• xband (8 - 10) ghz. The gain looks a bit high
• The cals could be a bit off.
• For higher frequencies, some of the sources with 15" extent or more are no longer point sources.
processing: x101/sysperf/misc/ruze/ruzecurve_2005.pro

# Ruze curves: data jul08 thru feb09  (top)

The measured telescope gain from jul08 thru feb09 was used to compute the ruze curves.
The plot shows the gain vs frequency with the ruze curves over plotted (.ps) (.pdf):
• 1549 gain measurements were used (lband thru xband).
• For each frequency, the median gain value for za=[5 to 14] degrees was plotted
• Gain0 was taken to be 11.K/Jy. This value was picked as the long wavelength value (see intro above).
• The data lies between 2 an 3 mm. some of the errors in the data are:
• cbh (6-8 Ghz) under illuminates the tertiary (so the gain is low).

processing: ~phil/sysperf/misc/ruze/ruzecurve_2009.pro

# Ruze curves: data mar11 thru mar12  (top)

The measured telescope gain from mar11 thru mar12  was used to compute the ruze curves.
Related events:

• 03feb10: found broken beam on the platform
• 05feb11: finished reinforcing the platform beams.
• 1707 gain measurements were used (lband rcvr  thru xband).
• For each frequency, the median gain value for za=[5 to 14] degrees was plotted.
• Gain0 was taken to be 11.K/Jy. This value was picked as the long wavelength value (see intro above).
• The 2.5 mm comes closest to the data.
• sbw,sbh, and cband lie on this line
• lbw is above all the lines
• cbh is low and xband is above.
processing: ~phil/sysperf/misc/ruze/rusecurve_2012.pro

# Ruze curves: data nov12 to aug13  (top)

The measured telescope gain from 20nov12 to 01aug13  was used to compute the ruze curves.
Related events:

• aug-sep12 : new cal values installed for lbw,xband,cbh,cband.
• 28aug12    : new sbn receiver installed
• 20nov12:   : new sbn cals installed
• 12jul13      : drag link on dome breaks
• 01aug13    : 1st stage sbn amps burned up.
• 1731 gain measurements were used (lband rcvr  thru xband).
• For each frequency, the median gain value for za=[5 to 14] degrees was plotted.
• Gain0 was taken to be 11.K/Jy. This value was picked as the long wavelength value (see intro above).
• The sbn gain values now lie on the same line as the sbw values. The new reciever increased the gain by decreasing the edge taper on the tertiary.
• The new cal values have brought the higher freq measurements closer together.
• cbh is still  a bit low.
• The surface error looks like it falls between 2.5 and 3 mm
processing: ~phil/sysperf/misc/ruze/rusecurve_2013.pro

# Ruze curves: data aug13 to jan14  (top)

The measured telescope gain from 01aug13 to 13jan14  was used to compute the ruze curves.
Related events:

• 01aug13    : 1st stage sbn amps fried. replaced with sbh amp.
• 13jan14     : earthquake damages main cable to tower 8.

• 1125 gain measurements were used (lband rcvr  thru xband).
• For each frequency, the median gain value for za=[5 to 14] degrees was plotted.
• Gain0 was taken to be 11.K/Jy. This value was picked as the long wavelength value (see intro above).
• The sbw gain has decreased from the previous dataset. Looking at the tsys measurements, it looks like the sbw cal has decreased a little.
• The sbn gain has gone up a little. this is probably for the amp switch. gain values now lie on the same line as the sbw values.
• sbn only had 40 gain measurements during this time period (as opposed to 200 - 400 for the other receivers).
• Tsys polB xband jumped down a few percent at the end of aug13. This is probably what increased the xb gain.
processing: ~phil/sysperf/misc/ruze/rusecurve_2014.pro

# Ruze curves: data jan14 to mar15  (top)

The measured telescope gain from 14jan14  to 26mar15  was used to compute the ruze curves.
Related events:

• 13jan14    : earthquake damages main cable to tower 8
• xxjul14      : sbn amp replaced with chalmers amp.
• 26mar15   : hf mesh first raised.

• 2735 gain measurements were used (430, lband rcvr  thru xband).
• For each frequency, the median gain value for za=[5 to 14] degrees was plotted.
• Gain0 was taken to be 11.K/Jy. This value was picked as the long wavelength value (see intro above).
• the lbw gains remain high.
• sbh 3640,3800 Mhz. these gains are too high. probaby a cal problem.
• The sbw gains have moved back up to the nov12-aug13 levels. So the short aug13-jan14 values were probably wrong
processing: ~phil/sysperf/misc/ruze/rusecurve_2015.pro
` home_~phil`