Single pixel observing with the mock spectrometer

    Single pixel observing with the mock spectrometer supports up to 7 independent frequency bands. Each band can supply: summed pols (1spectra) , polA,B (2 spectra), or full stokes (4 spectra) with 16 to 8192 channels per spectra. 12 bit A/D samplers running in complex sampling mode provide a maximum bandwidth for each spectra of 172.032 MHz.  A low pass digital filter allows decimation of the 172.032 MHz spectra from 1 to 1024 in steps of 1. A 4*fftlen polyphase filter bank improves the channel to channel isolation.  The output power spectrum  can use 2 or 4 byte integer data as output. Each spectrometer band (1,2,or 4 spectra) can record data at about 70 MBytes/second. Running all 7 spectrometers at the maximum i/o would give an aggregate data rate of 490 Mb/sec. To support this data rate, each frequency band is connected to a separate 5 terrabyte disc array.
 
    The parameters common to all frequency bands are:

• Number of bits for output
• Spectral integration time

    The parameters that are independent for each band are:

• configuration: summed pols, polA,B, stokes
• number of channels
• bandwidth divisor (decimation)

Some gotcha's:

• Single pixel observing does not support the winking cal synchronized to the mock buffers.
• Each mock spectrometer box has two fpga's (sp0 low band, sp1 high band) each capable of processing a complete (all pols) 172MHz band. Single pixel observing only uses sp1 (the high band) from each of the 7 mock boxes.
• The baseband mixer before the mock boxes provides:
• to sp0:
  
• 0 to250 MHz low pass filtered IF.
• IF CFR can be 150 to 250 MHz
• to sp1:
• 0 to 400 MHz low pass filtered IF
• IF CFR can be 250 to 400 MHz.
• Wideband single pixel observing uses the 1 Ghz IF.
• When mixing from the 1st IF (1=2 Ghz) to the 2nd IF (200-400 MHz) 3 Overlapping filters are used:
• 1000 to 1500,1250 to 1750,1500 to 2000
• The 2nd Lo is placed IF2 MHz away from the center freq in IF1. Since each filter is 500 MHz wide you would like the 2nd IF to be at least 250 MHz to put the 2nd LO outside  the 500 MHz filter.
• If we used the lower band sp0 IF=175, then the 2nd LO would be inside the 500 MHz filter for the IF1 to IF2 mixing.
• The winking cal synchronized to the mock buffers  is only generated from the lower sp0 fpga (this patch on the board is not connected to sp1).
  
• Since single pixel uses sp1 (hi band) there is no winking cal output signal generated.
  
• In 32 bit stokes mode, the fpga drops 1 fft between each accumulated spectra. If 400 ffts are accumulated, the wall time between each accumulated spectra will be 401 fft times.
• Since it is an fft based spectrometer, the integration times must be multiples of the time for 1 fft (fftlen/clockFrequency). The user interface will find the integration times closest to the ones requested.
• The single pixel mock spectrometer currently (24mar10) uses a 325 MHz IF center frequency. This is different than the 250 or 275 MHz IF frequency used by the wapps. This difference precludes observing with the wapps and the mock spectrometer simultaneously (this  will eventually be fixed).
  
• nov10: I've added the clock and losynth as settable parameters in cima (svnwork version). You should be able to set the 2nd IF to 250 MHz. This is on the edge of the i,q phase generating chip for this band. We need to check whether this is really going to work before telling  people to use it..
  

More info on the mock spectrometer can be found here.