links
Intro
Some terminology
Processing the data
plotting the results
Summary

plots
the results of the computations (.ps) (.pdf):

---

Intro

       Pulsar data cal files are used to compute tsys vs freq for each pulsar observation (more info). Monthly movies are then  made  of tsys vs freq. The movies showed that the bandpass shape is changing within a single day. The data from oct22 was used to examine this motion.

•  oct22 tsys vs freq movie .avi
• the rf bandpass filter limits the freq to 2205 - 2340 MHz (about 135 MHz of data)
• With 256 channels over 172.032 MHz, the channel width is 670 KHz.
• The spectral sampling varied from 32 usecs to 500 usecs.
• there were 259 calfiles spanning 01oct22 to 27oct22
  

---

Some Terminology:

• let Gx(f) be the frequency dependence of a particular  part of the rcvr chain
• let Tx   be the temperature contribution from a part of the rcver Chain.
• TsysCalOn=((Tsky  + Tscat + Tomt*Gomt  + Tcal*Gcal  + Trcv)*Gamp + Tiflo)*Giflo
• TsysCalOff=Tsys=((Tsky  + Tscat + Tomt*Gomt  + Tamp)*Gamp + Tiflo)*Giflo
• where:
• Tsky is temp from sky, Tcmb .. assume Gsky=1
• Tscat is temp from scattered radiation.. assume Gscat=1 over our narrow sband band
• Tomt is temp from omt . Gomt is the frequency dependence. This can be a standing wave or resonance.
• Tcal  is cal temp. Gcal is the frequency dependence
• Tamp is lna contribution. Gamp is the frequency dependence
• Tiflo is any other Tsys contribution after the lna. Giflo is mainly bandpass filters.
• TcalDefl=TsysCalOn - TsysCalOff= (Tcal*gcal*Gamp + Tiflo)*Giflo
• Tsys/TcalDefl=((Tsky + Tscat  + Tomt*Gomt + Tamp)*Gamp + Tiflo)*Giflo/((Tcal*Gcal*Gamp + Tiflo)*Giflo)
• If we ignore Tiflo (see the lna has lots of gain) then:
• Tsys/TcalDefl=(Tsky + Tscat  +Tomt*Gomt +Tamp)/(Tcal*Gcal)
• This gives Tsys in CalUnits.
• To convert to degK multiply by degKPerCalUnit.. this contains Gcal so the cal freq dependence should cancel (i think)
  
• TsysK=(Tsys/TcalDefl )*(DegKPerCalUnit
  

---

Processing the data

the following processing was done for each 90 sec winking cal file:

• the calon,caloff spectra were separated
  
• The calOn, caloff spectra were averaged for polA and polB
• the average cal deflection was computed (calOn  - calOff)
• The Average calOn,CalOff spectra were divided by the cal deflection and then multiplied by the calValue(f) in degK.
• dividing by the caldeflection removes the bandpass shape after the  cal injection.
  

---

Plotting the results:

      The plots show the results of the computations (.ps) (.pdf)

• Page 1 spectra vs freq:
• the green, black, and red  vertical lines are drawn for reference.
  
• topFrame: over plot all tsys spectra for polA
• 2ndFrame: over plot all tsys spectra for polB
• 3rdFrame:polA ,polB average tsys spectra for the entire month
• blue is polA,purple is polB
• BottomFrame: rms/mean by freq channel for the october spectra.
• If the band pass is shifting around, then the rms/mean should show were the variation is largest.
  
• blue is polA, purple is polB 
• The black an red lines are the frequencies that had the largest variations  (excluding rfi )
• black: 2271
• red: 2305
• green 2240:this is a reference frequency that did not have much variation.
• Page 2: freq channel total power vs sample
• The 3 flagged frequency channels were plotted.
• top: Freq channels vs sample
• green:2240MHz, black:2271MHz, red:2305MHz
• the y axis is Tsys in that channel
  
• The green reference channel shows that Tsys jumped around sample 50 to 150.
• Bottom: remove the reference channel from 2271and 2305
• After removing the reference channel, the 2 freq channels are out of phase.
• when 2305 goes up, 2271 goes down.
• this could happen if the band ripple was shifting from 2271 to 2305 and then back.
  
• Page 3: Plot the 2 freq channels vs elevation and hour of day.
• Top: plot the 2  freq channels vs elevation
• There does not seem to be much correlation with elevation.
• Bottom: plot the 2 freq channels vs hour of day
• there is a strong correlation with hour of day.
• before 5am:
• 2271 higher than 2305
• around 7am the level in 2305 increases while the level of 2271 decreases.
• the motion  reverses towards the end of the afternoon
• This is most likely related to temperature.
• Page 4: Plot the 2 frequency channels of the cal deflection.
• The cal deflection has the electronic gain and  filter shape (it is removed from Tsys when we divide by the cal deflection)
• Top Plot: divide the 2 freq channels by the reference channel at 2240 MHz to remove the the electronic gain and filter shape.
• This freq did not show much variation in the rms/mean of the Tsys spectra
• The two freq channels do not show any cyclic variation with sample
• bottom: plot the relative change of each freq channel vs hr of day
• the mean value from each trace of the top plot was used to normalize the data
• there is not variation of level with hour of day.
• This shows that whatever is causing the variation with frequency is occurring before the cal injection.

---

Summary

• The winking cal files from the sband pulsar observations were used to compute the tsys spectra for the month of oct22.
• Tsys movies showed that there is a bandpass ripple shift that is occurring (primarily in polB)
• The ripple shift is a function of the hour of day.
• it starts to move after 5am, reaches a peak in the early afternoon, and then returns in the later afternoon.
• This is probably being caused by a temperature variation.
• The cal deflection spectra were examined and no shift in levels were seen between the 2 max freq channels
• this says that the problem is occurring before the cal injection.
• Before the cal injection is:
• the omt
• the horn
• the sky
• I doubt that the sky or scattered radiation could cause such a large change in Tsys.
• This must by a standing wave or resonance in the omt/horn.
  

---

processing: x101/221027/chkripple.pro