CARA Science: CMBR Site Characterization: Background Information

(This is background information for the CMBR site testing page.)

The Python Experiment

The Python telescope, in its configuration for the 1996-1997 season, employs a dual feed 40 GHz HEMT based receiver. The receiver has two corrugated feeds separated by 2.75° on the sky, with separate RF chains, HEMT amplifiers, and backend signal processing. The post-detector output is AC coupled with a cut-off frequency of 1 Hz, and incorporates a low-pass anti-alias filter which attenuates the signal at frequencies above 100 Hz. The RF signal is separated into two frequency bands, centered at 39 GHz and 41.5 GHz with bandwidths of approximately 2 GHz and 5 GHz, respectively. Data from the two bands are combined for atmospheric analysis. The feeds are in the focal plane of a 0.8 m off-axis paraboloidal mirror, resulting in two 1.1° beams on the sky which are swept through 10° at constant elevation at a rate of 5.1 Hz by a large vertical flat mirror. The two beams are at the same elevation and their sweeps partially overlap on the sky. Beam spillover is reflected to the sky by two sets of shields, one set fixed to the tracking telescope, and a set of larger stationary ground shields which also shield the telescope structure from the sun and any local sources of interference. Data are taken for ~30 s while sweeping and tracking a central position on the sky; the telescope is then slewed to another position a few degrees away. Between 5 and 13 pointings are stored in one data file, representing 5 to 10 minutes of observing time. Data were taken over 80 % of the period from early December 1996 through early February 1997.

Data Analysis Technique

In order to differentiate atmospheric fluctuation power from instrument noise, the covariance of the data from the two beams is taken for the portion of each sweep in which their positions overlap on the sky, approximately 6°. Atmospheric brightness fluctuations are correlated between the two beams, while most of the instrument noise is uncorrelated. Thus the signal-to-noise of the correlated fluctuation power can be increased by averaging the covariance over many sweeps on the sky, allowing the atmospheric brightness fluctuation power to be estimated during stable periods when the system is receiver noise limited. The mean covariance represents the mean `snapshot' fluctuation power in the 6° sweep, regardless of the number of sweeps that are subsequently averaged together.

Several instrumental effects are accounted for when estimating the atmospheric brightness fluctuation power:

  1. The fluctuation power is corrected for the effect of the anti-alias filter roll-off and backend electronics delay.
  2. Correlated 60 Hz line noise is removed from the data.
  3. An offset dependent on the position of the sweeping mirror is correlated between the two beams. This is removed by subtracting the component of the signal that is constant on the sky over multiple pointings.
We determine the effect of the 60 Hz line noise filter and stationary signal removal technique on the true atmospheric signal by examining their effect on the data during periods when the data are dominated by atmospheric fluctuations. We calculated that the atmospheric power should be increased by 30% to compensate for the combined effect of these filters and instrumental effects. There is an additional correlated signal due to the stationary ground shield and due to the CMB itself, which is not removed with these techniques. Therefore, the quartiles which we report should be taken as an upper limit of the true atmospheric signal.
Back to the CMBR site testing page.