Antenna Configurations
The choice of antenna configuration can greatly affect the performance
of an array. For example, here is a configuration originally considered
for the core of LOFAR (S. Doeleman, private communication).
Here is are the results of calculations (see
Bowman, Morales, and Hewitt
for a discussion of the method)
of the sensitivity of this configuration
to the power spectrum at z=8 and z=10. The sensitivity is poor because there
are not enough short baselines.
Power spectrum
sensitivity of the LOFAR configuration
calculated for z=8 (z=10;right).
This calculation assumes a system temperature of 300 (520) Kelvin,
a collecting area of 14 () square meters for each antenna element,
a bandwidth of 8 MHz, 101 frequency channels, a square top-hat
field of view
of 31X31 () square degrees, and an integration time of 360 hours.
The window functions are approximated as delta functions to simplify
the computation.
Error bars expected for the above core configuration
are plotted with the power spectrum expected from HI fluctuations
at each redshift without reionization (a useful reference spectrum).
It is interesting to understand how the sensitivity to the power
spectrum changes with different radial distributions of antennas.
Here are four distributions which have uniform density
of antennas on the ground (antennas per square meter),
an antenna density that goes as 1/r, an antenna
density that goes as 1/r^2, and an antenna density that
goes as 1/r^3. In the central regions of the array the
antennas are not allowed to overlap, so they don't exactly
follow the power law distributions in the center.
The array diameter is 1.5 km as is planned for the MWA.
Four possible radial-power-law MWA demonstrator configurations.
Power spectrum
sensitivity of the four radial-power-law MWA demonstrator configurations
calculated for z=10.
This calculation assumes a system temperature of 520 Kelvin,
a collecting area of 18 square meters for each antenna element,
a bandwidth of 8 MHz, 101 frequency channels, a field of view
of 38X38 square degrees, and an integration time of 360 hours.
Error bars expected for the above core configuration
are plotted with the power spectrum expected from HI fluctuations
at each redshift without reionization (a useful reference spectrum).
These results differ somewhat from the results of
Bowman, Morales, and Hewitt
because I assume a square field of view (instead of circular) and
to speed up the calculations I assume delta function window functions
(instead of doing a full numerical convolution).
Last updated 14 August 2005