[daip] 4IF mode

[Eric Greisen]

 > 
 > I have data obtained in the 4 IF mode (8 channels of 3.125 MHz). Channel 1
 > gets scrapped (online I assume) and replaced with the "channel 0". This
 > leaves 7 channels to work with. I have heard that end channels should not be
 > used (due to high noise). Should I thus scrap channel 8, after bandpass
 > calibration, thus only leaving 6 of the original 8 channels ?
 > 

Original channel number 1 is at zero lag and is completely invalid.
That is why it is replaced.  All other channels should in principle
have equal S/N, but experience shows that the outer ones may be
somewhat more noisy - i.e. original channels 2 and 8.  Do not throw
them away however - do a proper BPASS and calibrate them.  BPASS can
be asked to tell youu about average closure error which will let you
evaluate how bad the edges are.  When you get to imaging, the latest
versions of IMAGR offer an option using BOXFILE to weight the channels
differently.  If you think the outer channels are twice as noisy give
them 1/4 the weight for example.  You could estimate the noise by
imaging each channel separately initially - or make a more crude
estimate from the BPASS reports (BPASSPRM(2,6,7) control this).

Eric Greisen