Intro

160909: running vlbi and puppi together.

    bm442 was scheduled to run the vlbi backend together with puppi (to look at a pulsar). We tested the mode on 09sep16.

The setup was:

• 327 receiver with 302 to 352 rf filter
• 750 Mhz IF1 used for vlbi
• IF1 had 500 Mhz centered at 750 Mhz IF freq.
• vlbi setup required :
• 750 Mhz IF1 corresponds to sky frequency of 510 MHz
• they took 512 Mhz BW.
• This required the 1st LO to be: 510 + 750 = 1260 Mhz (high side lo).
• With 510MHz rf centered at 750Mhz IF1 (and the band is flipped from the high side lo..):
• 327 Mhz was centered at  750 + (510-327)=933 MHz IF1
• the sky 302 - 352 Mhz band gets mapped to:
• IF1 908 to 958  .. this band is also in decreasing frequency from the high side 1st lo.
• standard Puppi 327 Mhz rcvr setup
• puppi samples at 100 Mhz
• We normally center the 327 band at 750 Mhz
• this folds down to 0 to 100 Mhz . 7 100 MHz folds ends up flipping the band..
• So the 0 to 100 Mhz data is in increasing frequency (1 flip from highside lo , 1 flip from 7 alias flips).
• puppi 327 rcvr setup for this experiment.
• 100 Mhz bandwidth baseband sampling.
  
• the down conversion (from under sampling ) work for any 100 Mhz band on n*100 to (n+1)*100 band edges.
• So 900 to 100 Mhz downconversion is just as good at 700-800 Mhz  default that is normally used.
• to go from 900-1000 to 700-800 takes 2 extra flips, so the final 0 to 100 Mhz is also increasing frequency.

• freq band	location of 302-352 Sky signel		notes
  	low freq	high freq
  sky	302	352
  900-1000 IF1	958	908
  000-100 Baseband	42	092

• Location of baseband data..
• We split the 100 Mhz into 32 100/32 =3.125 Mhz channels (still complex voltage).
• each of the 8 gpu's received 4 of these to record.
  
• The software assumed that  510 Sky freq mixed to 750Mhz IF1 had the sky power.
• The actual power was 200 Mhz above that (950).
•  So we need to subtract 200 Mhz from the header freq to get the correct values (- since high side 1st lo).
• puppi puts the sky Cfr in the middle of the 1st 3.125 Mhz channel above the center of the 100 MHz (gpu06 chan1 .cnt from 1)
  
•  

  	3.125  channels	hdr Freq center 4 chan	Actual freq center 4chan
  gpu01	1-4	464.68750	264.68750
  gpu02	5-8	477.18750	277.18750
  gpu03	9-12	489.68750	289.68750
  gpu05	13-16	502.18750	302.18750
  gpu06	17-20	514.6875	314.6875
  gpu07	21-24	527.18750	327.18750
  gpu09	25-28	539.68750	339.68750
  gpu04	29-32	552.18750	352.18750

  
  
•  Aliasing..
• We are using the 302-352 Mhz rf bandpass filter as the anti aliasing filter.
  
• This filter does not fall off very fast.
  
• This is a problem at the high side.. since the aliasing will start at 358.3125 Mhz
• The aliasing on the low side starts below 258.5625 .. the rf filter has fallen off by this time..
  

• This data was taken as a cal measurement (not baseband mode).
• top (polA,polB)
• bottom: stokes u,v
• You can see the power increase on the right edge of the band. This is caused by the slow rolloff of the 302-352 bandpass filter.
• I was transmitting a birdie at 320Mhz during the data acquisition (to check on the freq locations).

• We were transmitting a 320 Mhz birdie
• A 2^20 length fft was done over the 3.125 Mhz channel (freq res: 2.9802322 Hz)
• The x axis is the offset from 320 Mhz.