[daip] AIPS BPASS

[Eric Greisen]

PAUL wrote:
> Hi,
> 
> I was just wondering if someone could tell me what bpass actually does?
> 
> I've used AIPS many times before to reduce VLA data and had no problems.
> I'm currently working on GMRT data in which the bandpass seems quite  
> variable.
> 
> I have 128 chans @ 610MHz and running BPASS before CALIB means I get  
> 100% good solutions.  However, if I run CALIB without any BPASS, I  
> get a very high failure rate (~90% bad solutions) - even with the  
> specified UVRANG's for my calibrators from the VLA calib manual,  
> WEIGHTIT=1 and WTUV=0.1 for stability (both solmod='A&P' and soltype  
> 'L1R').
> 
> However, I think the AIPS cookbook suggests using BPASS after CALIB  
> and not before.  I've tried averaging channels together using SPLAT  
> (with aparm(3)= 1 and CHANNEL=10 for example) but this only helps  
> slightly in getting better CALIB solutions.
> 
> Also - I've noticed that BPASS actually changes my UVPLT's quite a  
> bit (i.e. the difference in bparm 0 (aplitude .vs. BL)  is quite  
> different when I toggle DOBAND between -1 and +1.

What data are you running CALIB on.  Note that CALIB averages channels
if you are not using a "channel 0" data set.  It now uses ICHANSEL in 
31DEC09.  BPASS actually does a CALIB solution channel by channel.  the 
question then in running BPASS is how you set BPASSP(5) and (10).  If 
your phase stability is good, you can set BPASSP(5) to 1 and avoid the 
record by record normalization which can be noisy.  BPASSP(10) then 
controls whether the bandpass is normalized on the way out.  Frequently 
we recommend BPASSP(10)=3 - again set ICHANSEL carefully and be sure to 
use the latest version of AIPS to get correct use of ICHANSEL.  You can 
then run CALIB applying the bandpass - but think about whether it should 
be a function of time - SOLINT in BPASS controls whether there are 
several times in the BP table and DOBAND controls how those times are 
used.  If you do not normalize the BP at all then it replaces CALIB 
entirely - except that usually one has a seldom observed strong BP 
source and a more observed weaker A&P cal source - so I recommend 
normalizing the BP.

Eric Greisen