[evlatests] An interesting observation ...

[Michael Rupen]

I think this is expected. The minimum fshift (the frequency shift 
"fundamental") is 100 Hz, and this is multiplied by the EVLA antenna
number to give the total LO offset for that antenna.  So ea01 has an LO
offset of 100 Hz, while ea28 has an LO offset of 28*100 Hz= 2.8 kHz.
This does eat up a noticeable portion of the band for the narrowest subband 
bandwidths but we decided that it was OK to have a true subband bandwidth of 
~28 kHz (31.25-2.8 kHz) for this narrowest setting.

100 Hz is a true limit, set by the 10 msec "tick" of the correlator:
1/0.01 sec = 100 Hz.

As to whether the lost frequencies occur on the low or the upper edge of
the band, I'd have to think a bit...

         Michael

On 
Wed, 21 Apr 2010, Rick Perley wrote:

>    I'm reducing data taken by Michael a couple days ago, which includes
> observations made with the sub-bandwidth of 31.25 kHz, and 122 Hz per
> spectral channel.  The data are generally lovely (details to reported on
> later), and the bandpasses perfect -- with a curious exception...
>
>    As a scroll through the bandpass solutions, I note that they
> progressively shift 'leftwards' as the antenna number increases -- the
> low-frequency filter cutoff is completely gone by the time antenna 28 is
> reached.  The shift, per antenna, is about 125 Hz.  Sounds to me like
> the fshift is not properly accounted for.
> _______________________________________________
> evlatests mailing list
> evlatests at listmgr.cv.nrao.edu
> http://listmgr.cv.nrao.edu/mailman/listinfo/evlatests
>