---

History:

---

Calibration measurements:

---

Miscellaneous:

All measurements by date:

• 200720: check sbh after repair of polA rf and bias cable. PolA bandpass has extra ripples.
  
• 200712: oscillation in polA
  
• 191222: polB Tsys high.
  
• 17jun11: sbh calibration used to measure 1az term of platform.
• 02jul09: Tsys vs freq and cal values over entire band.
• Feb04-Jun04: sbh GAIN CURVES. 
• Feb04-Jun04: System performance of data used for gain curves. 
• 09may03: resonances in the sbh receiver.
• may03-sep03: sbh GAIN CURVES. 
• may03-sep03: System performance of data used for gain curves. 
• 08apr02,11apr02 calibration runs.  Performance: gain,tsys,sefd,pnterr,etc.. 
• 23mar02: cal and Tsys versus frequency.

20jul20 check sbh after rf and bias cable fixed. polA had extra bandpass ripples

    The rf cable from the dewar to the blue cabinet and the bad polA bias cable were repaired. On 20jul20 blank sky was tracked for about 2 hours
(from za=19 at rise to about za=11.. the dec was about 15.25 deg).   Scans were taken with 300 1sec spectra each followed  by a 10 sec cal on,off.
The mock spectrometer was configured as:

• 7 172 mhz bands covering  3000 to 4000 Mhz.
• 1 second dumps

A total of 22 5 minutes ons were taken (3 extra ons at the end had a different configuration). During the 2 hours of observation there was a period of heavy rain.

    The first set of plots show the spectra (.ps) (.pdf)

• page 1: average spectra for 110 minutes of observing.
• Each 5 min on was scaled to Kelvins using the  cals
• no bandpass correction was done.
• Red is polA
• Green is  polB
• Top: band centered at 3290 Mhz
• Bottom: band centered at 3340 MHz
• You can see the high frequency oscillation in the polA bandpass in both frequency bands.
• Page 2: a 5 minute average spectra centered at 3340 Mhz (after the rain finished)
• top: average spectra
• You can see the ripples in a 5 minute  average
• Middle: the autocorrelation (fft of the spectra)
• The x axis is lag delay
• If there was a bad match somewhere in the signal path you would see a spike at that delay
• PolA acf has extra power spread out over 2.5 usecs.. so it is not a problem with a few bad matches.
• Bottom: acf plotted vs round trip distance.
• the x axis is the round trip distance for a reflection (assume vel=c/2 or 150 Meter/usec).
  
• If a reflection was in a cable, then the distance's would be about 30% shorter (vel=.7*c/2)
• Page 3: total power for the 3340 Mhz band.
  
• The total power for each 1 second spectra was computed for the 3340 Mhz band.
• The total power was normalized to the median value over the 2 hours (.. tsys).
  
• This used the raw spectra without the cal corrections
• So gain changes every 5 minutes would not be corrected by using the cal
• Top: total power vs hour of day
• The large bump around 14.9 hours was the heavy rainfall.
• PolA and polB started to diverge around 15.5 hours
• Middle: total power vs za.
• the observation started around 19.5 degrees, came down to 3 degrees and then back out to about 11 deg
• the declination tracked was around 15.25 deg.
• Bottom: blowup of total power vs  time after the rain.
• you can see the change in polA around 15.5 hours.
  
• the noise in polA is much larger than polB.

The second set of plots partially removes the bandpass  of the spectra (.ps) (.pdf)

• At arecibo you would normally do position switching to remove the bandpass and any standing waves.
• Since we only had 5 minute ons, I removed the bandpass using adjacent spectra; spcBpc[i]=spc[i]/spc[i+1] where i  is one of the 22 average spectra.
• This will remove the stable IF bandpass shape, but will not remove any standing waves from noise scattered into the system.
• Page 1: over plots the  bandpass corrected spectra.
• After dividing the avg spectra by the next  5 minute average, i removed a linear baseline.
• Only spectra after the rain were used (starting at 15.38 hours)
• the spectra are offset for display purposes
• You can see that the noise in polAn (red)  is a lot larger than the noise in polB (green)
• Page 2: rms of bandpass corrected spectra
• the rms was computed across each normalized spectra
• 5% of the bandpass on each side was excluded from the computation.
• Top rms vs hour of day
• red is polA , green is polB
• the blue line is the expected rms using the channel width (172mhz/8192, 300 second average, and the division for the bandpass correction).
• The expected value is .0056 Tsys.
  
• PolB is .001  and polB is twice that. We didn't reach the expected value because the bandpass correction was not an exact off position.
• Bottom: ratio rmsPolA/rmsPolB
• the polA rms is about twice that of polB

SUMMARY:

• PolA spectra have  much larger ripples across their  bandpass than polB
• the acf's show that the ripples are not from a few bad reflections in the signal path
• the rms noise across the polA band pass is a factor of 2 worse than that of polB
• The .343 hz spikes in the total power spectra have  gone away.
  
• What to do
• We might try switching things to see where the ripples go
• switch output of dewar to make sure it is in the dewar
• try using a different bias card on polA to see if the bias card has problems
• We are scheduled to replace the lna's in a few weeks..  so maybe we live with this for awhile.

processing: x101/200720/sbhblanksky.pro

12jul20 oscillation in polA

--> 15jul20.. found 2 problems

•  blue cable between postamps and if/lo chassis had a loose connection
• bad bias cable connection polA .. connector needs to be repaired. This is the major problem for polA.
  

    The sbn receiver was   reinstalled in the dome on 9jul20. The 2 sbh lna's had to be replaced. One was an older sbh amp. the other was an old sbn amp.   
using the mock spectrometer. Position switching was done with 3 min on and 3 min off.
Spectra were sampled at 2ms and stored. The total power was then computed for 10Mhz bands (after removing rfi).

href="sbhtotpwr_comb_200712.ps">the abs(fft) of the total power data for a 3min off scan (.ps) (.pdf)

•  The 10Mhz totalpower band was centered at 3205 MHz.
• Black is polA, red is polB
• The freq resolution is about 5 MilliHz
• top frame: 0 to 70Hz
• you can see the comb below 10Hz. 
  
• polA has stronger 60Hz than polB
• Bottom frame: 0 to 10 hz
• the green lines flag every .343 hz
• The comb is mainly in polA.

processing: x101/200712/chksbh.pro

Feb04 to Jun04  fit GAIN CURVES to calib data.  (top...)

    Gain curves were fit to the sband high  calibration gain data using  13Feb04  through 30jun04.  The start of this epoch was after the als were replaced and the dewar was lower .8 inches to bring it into focus. The plots show the gain data (black) and the fits (red) for 3300, 3500, 3700, and 3825 Mhz. These gain equations were installed on 03Jul04 and back data to be valid starting on 13feb04.

• Fig 1 shows the az,za distribution for the data. The fit used a linear fit in za up to za=14. Above 14 degrees terms in (za-14)^2 and (za-14)^3  were included. The fit also included  1az, 2az, and 3az sin, cos terms.
• Fig 2 plots the gain data and the fit to za.  The fit equation is plotted with the sigma for the fit (in K/Jy).
• Fig  3 plots the fit residuals (data-fit) vs za.

feb04 thru jun04 : System performance of data used to compute gain curves.  (top...)

13feb04 (new cals, horn lowered) thru 30jun04 were used to measure the
system performance. This data was then used to compute the gain curves used after 12feb04.

    The first set of plots show the system performance with all frequencies overplotted. The sources are identified
by symbol and the frequencies by color.

• Fig 1 shows the distributions on the dish of the measurements.
• Fig 2 has the Gain in Kelvins/Jansky. This relies on the cals and the source flux. Ths  next plot is Tsys vs za in Kelvins followed by the SEFD  (System Equivalent Flux density) in Janskies / Tsys. At the bottom is the average beam width in arc seconds.
• Fig 3 plots the coma parameter, first sidelobe height below the peak, the main beam efficiency, followed by the main beam + 1st sidelobe beam efficiency.
• Fig 4 has the pointing errors in az,za.

• Fig 1 3300 Mhz Gain,Tsys . 282 points
• Fig 2 3300 Mhz sidelobes,beam efficiencies
• Fig 3 3500 Mhz Gain,Tsys 282 points.
• Fig 4 3500 Mhz sidelobes,beam efficiencies.
• Fig 5 3700 Mhz Gain,Tsys 282 points.
• Fig 6 3700 Mhz sidelobes,beam efficiencies
• Fig 7 3825 Mhz Gain,Tsys 201 points.
• Fig 8 3825 Mhz sidelobes,beam efficiencies

May03 to Sep03  fit GAIN CURVES to calib data.  (top...)

    Gain curves were fit to the sband high  gain data from 15may03 through 30sep03.  The start of this epoch was after the shimmin g of the elevation rails in feb03. The end date was chosen so that the cal had not drifted too much (see warning below). The plots show the gain data (black) and the fits (red) for 3300, 3500, 3700, and 3900 Mhz. These gain equations were installed on 10mar04.

• Fig 1 shows the az,za distribution for the data. The fit used a linear fit in za up to za=14. Above 14 degrees terms in (za-14)^2 and (za-14)^3  were included. The fit also included  1az, 2az, and 3az sin, cos terms.
• Fig 2 plots the gain data and the fit to za.  The fit equation is plotted with the sigma for the fit (in K/Jy).
• Fig  3 plots the fit residuals (data-fit) vs za.

Warning: The sbh cal was drifting in time (see tsys plots). During this period it was relatively stable. Use these curves outside this region with care. The cals were replaced in feb04.

    The routine gainget() or corhgainget() will now return the sbh gain for data after 01mar03 from these
equations. The coefficients can be found in the ascii file  data/gain.datR8 (this is provided in the AO idl
distribution for correlator routines). You can also find a copy of it at AO in /home/phil/idl/data/gain.datR8.

may03 thru feb04 : System performance of data used to compute gain curves.  (top...)

15may03 (after shimming) thru 30sep03 were used to measure the
system performance. This data was then used to compute the gain curves used after 01mar03.

    The first set of plots show the system performance with all frequencies overplotted. The sources are identified
by symbol and the frequencies by color.

• Fig 1 shows the distributions on the dish of the measurements.
• Fig 2 has the Gain in Kelvins/Jansky. This relies on the cals and the source flux. Ths  next plot is Tsys vs za in Kelvins followed by the SEFD  (System Equivalent Flux density) in Janskies / Tsys. At the bottom is the average beam width in arc seconds.
• Fig 3 plots the coma parameter, first sidelobe height below the peak, the main beam efficiency, followed by the main beam + 1st sidelobe beam efficiency.
• Fig 4 has the pointing errors in az,za.

• Fig 1 3300 Mhz Gain,Tsys . 295 points
• Fig 2 3300 Mhz sidelobes,beam efficiencies
• Fig 3 3500 Mhz Gain,Tsys 294 points.
• Fig 4 3500 Mhz sidelobes,beam efficiencies.
• Fig 5 3700 Mhz Gain,Tsys 293 points.
• Fig 6 3700 Mhz sidelobes,beam efficiencies
• Fig 7 3900 Mhz Gain,Tsys 294 points.
• Fig 8 3900 Mhz sidelobes,beam efficiencies

08apr02,11apr02: calibration runs.  (top...)

• Figure 1 show the gain in K/Jy, Tsys in K, SEFD in Jy/Tsys, and the average beam width. Sources are plotted with different symbols. Color is used separate the 4 frequencies.
• Figure 2 plots the coma parameter, first sidelobe height relative to the peak in db's, main beam efficiency, and the main beam plus first sidelobe beam efficiency.
• Figure 3 plots the pointing error in za (versus za,az), pointing error in az (versus za, az), and the total pointing error (sqrt(azerr^2+zaerr^2) versus za, and az. At top is plotted the mean and rms for the zaerr, azerr, and total error.

23mar02: cal and tsys versus frequency. (top...)

home_~phil