[evlatests] Tests of the bandpass phase equalization filter, 26 Sep 2005

[Mark Claussen]

Yesterday (Sep 26, 2005) Mike R. was ready to test 
his implementation of the EVLA equalization of the 
phase of the bandpass in the second stage of the 
FIR filter.  He prepared the requisite files, and
he and Ken did the necessary re-booting and loading
of the proper files and devices.  This was only done
in 14A, as Ken reported.  As Mike expected and Ken
reported, this caused an extra delay for 14A. 

I have looked at some of the data that Ken took after
setting up the filter to do this.  The data were taken
in mode 1A for backend filters 50, 25, 12.5, 6.25, 
3.125 and 1.5625 MHz.  I report here specifically on 
only the 50, 25, and 3.125  MHz filters; these are 
the ones that I have previous data to compare to.

I have made some plots showing a comparison before 
and after the equalization for the 50, 25 and 3.125
MHz backend filters, and comparing the 50 and 25 MHz
after equalization with the bandpass from An 11, in
order to compare with a VLA antenna.  These plots
can be viewed at 

http://www.nrao.edu/~mclausse/evlatesting/plots.html

What is plotted is the bandpass phase table, made
by using the AIPS task BPASS which, given a reference
antenna (in this case I used An 20 for the reference)
calculates a complex correction table over a given
time range.

The results are:  
1) We corrected the 50 MHz bandpass phase pretty well.  
The peak-to-peak phase variations on An 14 after the
correction are ~20 degrees, dominated by the channel
at the low-frequency edge of the baseband.

2) As we go to narrower bandwidths, so that the baseband 
frequency is lower, the correction does not work as
well.  This was expected by Mike, I think, because for
the FIR filter to correct at the low frequency end
(below about 1 - 2 MHz) many more taps would have to be
used in the filter.  At 25 MHz, as can be seen in the 
plot, the correction is pretty good down to about
2 - 3 MHz, and then, although there is some correction,
it is not as good as we would perhaps like it to be.
The 3MHz filter, of course, is dominated by this effect,
since most of the bandpass lies below 2 MHz (obviously).
The plot of the 3 MHz filter still shows some correction
was correctly applied.

I will make some plots of the bandpass phase across
the 12.5 and 6.25 MHz filters, but don't have earlier 
data to make the comparison.  The three that I have 
discussed make the point.

Mark