During September 2000 extensive observations were made to obtain complete characterizations of the polarization properties of all of Arecibo's receiver systems. A memo written by Carl Heiles, et.al (and available here) gives a complete description of the Mueller matrices used and the results. On this page a summary of the L-Narrow results are given.
The LBN system uses a turnstile for its orthomode transducer, which is a relatively narrow-band device; it is tuned to give good dual circular near 1420 MHz, but as one departs from this frequency the polarization becomes dual elliptical. Additionally, the LBN does not have a correlated cal. This means that the Mueller Matrix parameter (psi) has no meaning.
The LBN system is commonly used over a large frequency range. A
turnstile is a narrow-band device, and over this range the polarization
changes from linear to circular and back again. The solid lines in
Figure 1 (below) exhibit the frequency dependence of G, ,
for the 25 MHz band centered at four
frequencies. These particular data were derived from the source
B0017+154; we obtained data for two additional sources, and the results
are very close. The dashed lines are our adopted analytic expression,
which are defined by
G = 0.034 - 1.78X10-4 f15 + 3.27X10-6 f152
= 47.74 - 0.363 f15
= 0.0028 + 1.38X10-5
f15 + 1.31X10-6f152
= 21 + 1.02 f15
where f15 = f - 1415MHz and angles are in degrees.
The linear dependence of on f15$ is just what's expected for a turnstile junction. However, the variation of is remarkably complicated, varying rapidly with frequency as one goes away from the 1400 MHz center frequency, and we do not understand the reason. The scatter in $ for the 1375 MHz spectrum simply reflects the uncertainty in the angle, which is large because is small. Of course, G simply reflects inaccurate relative cal values and not the properties of the turnstile itself.
Fractional polarization (really gaina-gainb)/(gaina+gainb) since the outputs are circular) versus azimuth and zenith angle, by source