[mmaimcal] Matching of frequency profiles for ALMA and ACA correlators

[Stéphane Guilloteau]

I would support all the arguments of Darrel, and add one more

- since this is the maximum frequency resolution, what are the expectation 
that the Signal to Noise be
strong enough that we can actually be dominated by the systematic effect of 
the channel shape mismatch ?

This is specially important with respect to argument 3 of Darrel.  The 
usual case of strong enough narrow lines is that of maser
emission, but in this case, ACA is unimportant...

         Regards,

                 Stephane

A 14:10 20/06/2005 -0700, Darrel Emerson a écrit :
>Are at least some of the following true?
>
>1. If we're prepared to give up a factor of ~2 in ACA correlator frequency
>resolution, then there's no problem, because we can reweight terms
>of the ACA FX correlator in order to equalize the frequency response with
>the main ALMA array sinc response.  Especially true since we'll usually
>want to grade responses (Hanning or whatever) to reduce frequency sidelobes.
>
>2. Doesn't the ACA correlator support a greater number of channels
>than the main ALMA array anyway (I'm not at all sure if it does or not)?
>If so, then combined with (1) above that's the end of the story.
>
>3. How often is the ultimate in ACA frequency response really going to be
>needed?  Could it be that in most cases where extreme frequency resolution
>and maximum number of channels are needed, the angular extent of the
>source is such that the combination of ACA and main ALMA may not
>be so important?  In other words, is there any tendency for spectrally
>narrow emission not to be spatially very extended (the reverse of the Orion
>CO spectrum.)  I don't know if this is true.
>
>  My feeling, which may be probably completely wrong, is that it'll be
>sufficiently rare that we run into a significant  imaging problem because 
>of this,
>since in most cases we can sacrifice FX resolution to equalize the frequency
>responses, that it's not even worth the effort of making simulations to 
>quantify
>things further.
>                        Cheers,
>                                      Darrel.