Intro

    On 08Nov21 the xband cal was changed to be temperature regulated. On 07dec21 we took winking cal data on blank sky to look at the cal spectra.

Setup:    

• track blank sky: ra=16:00 dec=35:17  (az=311,el=61.72)
• take data for 120 seconds
• leave the 25 hz hardware winking cal on during the 120 seconds
• Dump spectra with the mock spectrometer
• 7x172MHz bands,
  
• 2048 chan/spc, 84KHz channel width
  
• bands spaced by about 142 MHz.
  
• band center freq: 8219.,8363.,8505.,8647.,8789.,8931.,9073. MHz
• 16 bit spectra
• 2 millisec integrations.
• a/d rms set to 30 counts.

---

Processing the data

For each of the 7 bands:

• input the file
• split the data into calon and calOff
• drop the 2ms spectra around each cal transition.
• average the calOn and calOff spectra
• compute calR= (calOn - calOff)/CalOff
• this puts the cal deflection in units of Tsys
• What we have computed is:
• let Tsky be the sky temp that goes directly into the feed,
  
•  Tsc be any scattered radiation into the feed,
• Tcal be the cal temp
• Trcv be the receiver temp (amps.. etc)
  
• Gsc(f) be the freq dependence of the scattered radiation
• Gcal(f) be the cal freq dependence
• Grcv(f) be the rcv temp freq dependence
  
• Gif be the IF band pass dependence
• i'm assuming that the sky freq dependence is relatively flat across the 1GHz.
  
• Ton= (Tsky + Tsc*Gsc +  Tcal*Gcal + Trcv*Grcv)*Gif
• Toff=(Tsky + Tsc*Gsc + Trcv*Grcv)*Gif
• calR=(Ton-Toff)/Toff= (Tcal*Gcal)/(Tsky + Tsc*Gsc + Trcv*Grcv)
  

---

Looking at the data:

    The plots shows the cal spectra an fit (.ps) (.pdf)

• Page 1:  spectra with cal on and cal off
• The 7 spectra covering the xb freq range. No band pass correction has been done.
• black in calOn, blue is with the calOff (Tsys).
• Page 2: calR=(calOn - calOff)/calOff   spectra and acf
• Top: (calOn - calOff)/calOff
  
• black polA, red polB
• the 7 spectra were interpolated to a single contiguous spectra with an 84Khz channel width.
• Dividing by calOff removes the band passes and changes the units to Tsys.
• Bottom: acf of (calon-calOff)/caloff
• The acf was computed for the calR excluding the edges
  
• This can show the spacing for any standing waves (rippled) in the spectra.
• Black in polA, red is polB
• There are spikes at:
• .015 usecs (2.3 meters with n=1)
• .088 usecs (13.2 meters with n=1
• The CalR spectra has a frequency dependence
  
• from the cal
  
• From the division by CalOff spectra.
• Page 3:  Fit to the calR spectra
• The fit used a linear polynomial and an 18th order harmonic function
• black: the data
• blue : the  fit
• The red * were excluded from the fit
• Top polA
• Middle: polB
• bottom: fit residuals
• black: polA rms: .0018 Tsys
• red: polB rms: .0018 Tsys
• If the cal is .3* tsys, then the  fit residual = .006 of cal Value.
  

---

Summary

• The cal value is about .3*Tsys
• The cal varies by about 23% across the band
• ripples are seen in the cal deflection value.
• .015 and .088 usecs delays.
• This differs a little from the ripples seen when looking at the sky (more info)
• a linear with 18th order harmonic fit to the cal deflection has an rms residual of .006 cal units
• The fit is including the ripples in the cal.
• If the ripples are from the cal diode or the insertion into the cal coupler then this is correct.
• If the ripples are from the division by the calOff spectra then they should not be included in the fit.
  

processing: x101/2112107/wcal_fit.pro