Intro:(top)

    Data for model4A was taken between 22jul21 and 09sep21.  The setup was:

• the data for model4A was taken using model3B as the installed model.
• The xband receiver was used to make the model.
• 7 172 MHz bandwidth bands were taken covering 8135 to 9200 MHz.
  
• There was a 142 MHz spacing between the band centers
• the band centers were:8221.00 8363.00 8505.00 8647.00 8789.00 8931.00 9073.00 MHz
• the first 82 crosses on 22jul21 3C273 had a wider spacing. I changed it to the above to get away from the edge of the 1GHz IF filter).
  
• The cross  strips had:
  
• 
  

rcvr	bmWidth	stripLen	stripTm	bms/strip	smpRate	smp/Strip	smp/Bm	Notes
xb	10.Amin	120Amin	120sec	12	10hz	1200	100	mod4A sep21,nocal



• Most sources were tracked rise to set.

Taking the data: (top)

The sources used:

• some sources were observed on multiple dates:
• some days had rain and part of the track had to be repeated.

total power strips for each source

    The plots show the azimuth strip total power vs time for each source track.

• Only the 8505 MHz band is plotted
• some sources that spanned local midnight will be on adjacent day plots.
• Each plot is normalized to the median value.
  
• The data is plotted vs local hour of day.
  
• The gaps in time between az strips are where the elevation strips would be.
• the periodic spikes in the plot is the azimuth driving over the source.
• some days show large jumps in the power. This was rain.
• After the rain, water could remain in front of the horn until it moved to lower el (there is a lip around the edge holding the capton).
• Some jumps in level may  be instabilities in the receiver chain.
  

• dates	sources	link
  210722 210722 210723	3C273 3C58 3C405	(.pdf)
  210723 210724 210724	3C58 3C274 3C405	(.pdf)
  210725 210725 210726	3C461 3C84 3C279	(.pdf)
  210726 210831 210908	3C461 3C144 3C273	(.pdf)
  210909 210910	3C461 3C461	(.pdf)

•  top : total power vs time
• the red lines straddle a jump in power
• the blue lines straddle rain
• middle: blowup of power jump
• bottom: blowup of rain
  

The raw pointing errors: (top)

• the fits were done with the current model (model3B) applied).
• All errors are great circle arcseconds.

Removing some of the bad fits

• fit2dall() did the 2d gaussfits to all of the crosses.
• rain and unstable baselines caused many of the fits to be unusable (especially the weaker sources)
• fit2dfilter() was then called to try and remove some of the bad fits

•  Page 1:  beamwidths
• data from all freq bands is plotted.
  
• Top:  elevation vs azimuth beam widths
• the data was clipped to 8Amin < width < 15Amin . The median value is 10Amin
• the elevation width increases as the az width decreases.
• The outlying widths probably come from bad baselines that were not fit well by the linear el term.
• middle,bottom: bmwidths vs  az,el
• black is az bmwidth, red is the el bmwidth
• The bm width increases at low and high elevations.
• Page 2: pointing errors.
• These are relative to model3B
• black is az error, red is el error
• The data was clipped to bmwidths 8-15 amin and less than 180 Asecs.
• top: az,el errors vs az
• middle: az,el errors vs el
• The el errors are increasing as you move to low elevations (this may be a baseline problem because of the Tsys change. 
• bottom: az, el errors vs elevation using the median value of the 7 frequencies.
  

Plotting the filtered pointing errors

    After removing the bad fits

Plots  of pointing errors after filtering points (.pdf)

•  Page 1: az,za coverage of the data
• the circles are every 10 deg in za (za=0 at the center)
• the blue lines are of constant declination
• The colored + are the locations in az,za of the measured crosses.
• There is little coverage below declination -10 deg (no strong sources).
• Page 2: az,el error (relative to model 3B) vs azimuth
  
• each color is a different source
• The red dotted lines show if the source is rising or setting.
  
• Top: elevation error vs azimuth
• the mean error is 31.5 asecs, the rms is 32.2 asecs
• the horizontal purple lines  are clipping levels that will be used to throw out outliers.
• bottom: azimuth error vs azimuth
• mean error: 8 asecs, rems 25.8 asecs.
• Page 3: az,el errors vs elevation
• top: elevation errors
• bottom: az errors
• the black plot (3C144) pointing errors  jump up after transit.
• Page 4: az,el errors by source
• Each color is a different source
• Top: el errors vs index
• bottom: az errors vs index.
• The purple clipping lines will be used to remove outliers.
• You can see the jump in error at transit for 3C144 (black).
• Page 5: Total errors:
• the az,el errors were added in quadrature.
• top: total errs vs elevation
• middle: total errors vs azimuth
• bottom: total errors by source.
• Page 6: arrow plot of errors vs az,za
• Each color is a separate source
• Each circle is a step of 22.5  deg in za
• The length of the arrow is proportional to the error.
  
• 1 tick mark is 60 asecs.
• You can see that some errors go in a completely different direction than the adjacent crossed. These are bad fits.
• Page 7: arrow plot showing outliers to be removed.
• The purple arrow are crosses that will be excluded.
• Page 8: Raw az,el errors vs azimuth
• The model3B corrections have been removed. These are then the raw errors.
• Page 9: raw az,el errors vs elevation
• top: el errors vs elevation
• middle: az errors vs elevation
• bottom: little circle az errors vs elevation
• this shows that the azimuth encoder is off by about 1400 asecs (.3 degrees).

processing: plt2dfit.pro

Verifying some of the pointing errors

   3C461 and the 3c144 were tracked multiple times to check the repeatability of the measurements.

3C461

    The first track of 3C461 (25aug21) was not a complete track. It was re observed on 09sep21.

The plot shows the pointing errors for multiple tracks of 3C461(.pdf)

•  1st,2nd frames: az error  vs azimuth and elevation
• 3rd,4th frames: el error vs azimuth and elevation
• the parts of the tracks that overlap have errors that repeat very closely.

processing: model4A/chk3c461.pro

3C144

    3C144 (the crab) had a large jump in pointing errors after transit. It was tracked multiple times to verify that the jump was real
The plot shows the pointing errors for multiple tracks of 3C144  (.pdf)

• top : az pointing error vs cross number
• 2nd: az pointing error vs elevation
• 3rd: el pointing error vs cross number
• bottom: el pointing error vs elevation
• The az and el pointing errors track pretty well for the two dates
• The jump in pointing error after transit is seen in both days.
• Both days show the pointing error jumping around closer to setting.
• the sun was at ra,dec: 10:40  and 08:23  so it wasn't too close to the beam (maybe a tripod leg?)\

Fitting the pointing model.

• All errors are great circle arc seconds.
  
• Page 1: fit residual errors by cross number
• each color is a different source
• top: az fit residuals
• bottom: el fit residuals
• The table show the az mean,rms and el mean,rms fit residuals for each source.
• Page 2: az,el fit residuals vs azimuth and elevation
• frame 1,2: az fit residuals vs azimuth and elevation
• frame 3,4: el fit  residuals vs azimuth and elevation.
• Page 3: arrow plot of az,za fit residual errors vs az,za
• each color is a different source.
• 1 tick mark is 30 secs.
• there are still some outliers that probably shouldn't have been included in the fits.
• You can also see that the pointing errors for 3C144 (black) blow up after transit.
• much of the scatter in the residuals is from the weaker sources
• 3C144, 3C461, and 3C405 have less scatter.
• The errors tend to increase for low elevation (see 3C461). The refraction correction is probably  to blame since we are using an approximate model (we are not including the current precipitable water vapor in the computation).
• Page 4: lists the pointing model results:
• the AzF and elF are value when fitting az,el separately. The all values fit everything at once.

The model results:

• The total col has the az,el residuals added in quadrature.
• model 4A
• i'm a little surprised that the mean values are not closer to 0. (probably since it is trying to fit both at once).
• Excluding the weaker sources would improve the rms, but lessen the az,el coverage.
  

<- page up
 home_~phil