Subject: agenda for today's meeting


   GBT Scientific Working Group Meeting of February 14, 1995 
 
      4 PM EST


                   
                AGENDA

  I.  General Project News (Bob Hall)

 II.  Spectrometer News (Ray Escoffier)
	  Spectrometer Data Rates (see comments below)
	  Spectrometer Filters (see information below)

III.  140-foot tests
         Receiver Update (Ron Maddalena)

 IV.  Refraction, etc. (see GBT Memo 112)

  V.  Surface Accuracy (see GBT Memo 119)

  VI.  Use of the active surface (see GBT Memo 120)



If there are any topics you would like discussed please contact me:
jlockman@nrao.edu
       

	*	*	*	*	*	*	*	*


Spectrometer Data Rates

	At a recent meeting it was determined that by splitting the 
data four ways, the spectrometer 
hardware could support the following data rates for spectroscopy:
32 independent IF inputs, 1024 spectral channels per IF input, and
a minimum integration time of about 20 ms, or 4096 channels per IF
at 79 ms.  

Rick Fisher comments:
Pulsar data rates are more difficult to state because of all the
combinations and permutations.  The key number is the 1.2 us per
data point.  Further increase in data speed would require a major
redesign, which is outside of the ground rules for puslar use of the
spectrometer.  

David Nice comments: 
	True, but a simple re-scaling of these numbers gives some optimism
that this could be a useful pulsar search machine (which is the
pulsar application for which the data rate is critical).  If we
have 2 IFs instead of 32, and if we are willing to drop down to,      
say, 256 channels, we get an integration time of about 0.25 ms, similar
to that used in the present spectral-processor-plus-external-workstation
configuration (albeit with only half the channels of the spec.proc.
system).  The bottleneck for pulsar searching in the correlator
will be getting data out of the machine entirely and onto tape,
which may require more powerful computing than NRAO is willing or
able to supply.
 	My point is:  yes, this is actually a potentially useful pulsar
search machine BUT the factor-of-four speedup alluded to in Rick's
note (and discussed extensively at our meeting in December) is
extremely important for this application.   


 *	*	*	*	*	*	*	*	*	

Spectrometer Filters:
	IF filters for the new spectrometer need to be purchased soon.
We have a choice of several filter bandwidths at varying price.  The
table listed below is the bandwidth to the 1-dB point.  Note that because
of the reduced chip clock rate (100 MHz down from the hoped-for 125 MHz)
the maximum BW of the new spectrometer is 800 MHz, and the second 
"wide-band" bandwidth is 200 MHz (down from 250).  The choice of filter
establishes the fraction of the band which is usable for spectroscopy.  
We are leaning toward the 94% filters just because they provide extra 
bandwidth, but they are somewhat more expensive than the other filters.
The initial spec for the maximum bandwidth of the spectrometer was 700 MHz.

  % BW   MHz BW   #poles   
  ====   ======   ======   
   90     716       10     
   92     732       12     
   94     748       14     

   90     179       10     
   92     183       12     
   94     187       15