lband wide calibration Sept 01

last updated 19nov01

Calibration runs for the lband wide receiver were done in Aug01 using the chcross routine (see aotm 99-02 sec 3.1). A low frequency set of data was taken (1175,1300,1375, 1415 Mhz) and a high frequency set (1415,1550,1610, 1666 Mhz). The data results are for the average of the two polarizations (stokes I/2).

The southwest quadrant of the reflector had not yet had it's final adjustment when the august data was taken. The surface rms went from 5.5 mm rms (before) to < 1.8 mm after the final adjustment. So the august data had 3/4 of the dish with an rms < 1.8 mm and 1/4 of the dish with an rms of about 5.5 mm. In Sep01 the southwest quadrant adjustments were finished and a few sources were re-measured.

gain at 1415 Mhz.
Tsys recovery after dewar warmup.
Tsys at 1415 Mhz.
Gain as a function of frequency.

Fit to the gain at 1415Mhz. (lbwgaintop)

The first set of figures shows the gain of the telescope at 1415 Mhz using the august data. 1415 Mhz was included in the high and low frequency data sets so there were twice as many points at 1415 than the other frequencies. The gain values were computed using the original cal values measured by gene loria in the lab (circa 1997). The fluxes for the source were taken from fits by chris salter. The "standard" astronomical calibrators included were: B0134+329 (3C48) and B0518+165 (3C138). The figures of the plot show:

Figure 1 is the measured gain versus zenith angle (top) and gain versus azimuth angle (bottom). The sources about 0 degrees azimuth cover the north half of the dish while the sources about 180 degrees azimuth cover the southern half of the dish. The sept data was not included.

Figure 2 plots the gain versus azimuth and zenith angle. The length of the arrow is proportional to the gain (1 tick mark = 5 K/Jy). The angle of the arrow also measures the gain: vertical up is the maximum value, 90 degrees to the right is 25% of the maximum, and vertical down is 50% of the maximum value. Sources with declination < 18 deg (AO latitude) will appear on the upper half (northern part) of the dish. Sources with declination < 18 deg will appear on the lower (southern half) of the dish. Rising sources are on the left (west half of the dish) and set on the right (east half of the dish). The southwest quadrant is in the lower left. A number of tracks across the dish have double arrows. These correspond to the lower and upper frequency data sets (that were taken on different days). How close these two sets overlap shows the repeatability of the measurements. The table at the bottom shows the average over the entire dish (9.51K/Jy) and the averages in 5 deg za steps.

Figure 3 shows the fit to the data. It is a 3rd order polynomial in za with the za^2, za^3 terms only being fit above za=14 deg. It also includes 1az,2az,3az sin and cosine terms.

The top plot is (data - azfitComponents) vs za.
The bottom plot is (data-zaFitcomponents-constant) vs az. The black line is the sum of the 1a,2az, and 3az terms. The blue line is the sum of the 1az and 2az terms. You can see the drop-off in gain for the southern portion of the dish (about 180 degrees). From pitch, roll, and focus (prf) measurements we've seen that there is a 3 azimuth term for the prf and thus the gain. The 1a and 2az terms are trying to compensate for the bad surface quadrant (any za dependence that is not fit for probably also shows up in the 2az term since we have lots of points near az=90, 270 degrees).
Figure 4 is the fit residuals (data -fit) versus azimuth and za. 1 tick mark is .5 K/Jy. The arrow direction is: up-->0 difference. right 90degrees data-fit = 1K/Jy, left 90 degrees data-fit=-1K/Jy. The black 2 arrows in the upper left (B0139+029) point to the right so they are 1K/Jy above the fit. The flux for this source is probably a little high.

Figure 5 top is the (data - fit) by source. The bottom plot is (data - fit) for the sources taken in september after the surface adjustment had completed. The southern sources (B2209+080,B0518+165) in the north part of the dish are about .4/10 =4% above the fit. The northern source B0134+329 is about .8/10=8% above the fit. J0051+177 is about 10% above the fit (although i think its flux is a bit off).

Figure 6 compares the sources common sources taken in Aug. (black +) and Sept. (red *). B0134+329 crosses the south part of the dish while B2209+080 is in the north half. The lines are 2nd order fits versus za to the data. The green line is (fitAfter/fitbefore -1)*100. This is the percent change. B0134+329 has increased by 9% while B2209+08 has increased by 4%.

Figure 7 shows a correction to the august data to try and compensate for the southwest quadrant correction. Assuming the 1 and 2 az terms were trying to correct for the southwest quadrant I subtracted the 1az and 2az terms (with the az dependence) from all of the data. I then added then maximum value of g(1az)+g(2az) to all of the data. The top plot is DataWithCorrection- gainAzFit for the new fit. The bottom plot is DataWithCorrection - gainZaFit-Constant. The 1az and 2az terms of the fit are now zero (since they were removed from the data). The constant term has increased by .15 K/Jy.

Plot 8 top is the residuals (dataCorrected -fit) by source. The bottom plot is (Sep01Data - fit) by source. B2209+080 and B0518+165 are still about 4% high while B0134+329 has moved from 9% high to 4% high. The scatter in B0134+329 has also decreased. The source J0051+177 remain about 10% high.

Using the corrected august data, the gain fit remains about 4% below the data values of sept01.
The zenith angle dependence of the gain out to 15 degrees should be flat. The linear za term is -.1 K/Jy/Degza (1%/degZa). The mar00 model of the pitch, roll,focus errors gave an azimuth average gain loss of .5%/degZa. This included a 2.5 inch focus error versus za. It may be that the pitch and focus have changed.

Tsys recovery after dewar warmup. (lbwgaintop)

The lbw dewar compressor tripped off on 12aug01 at about 10am. The 1st stage dewar temperature went from 13K to 200K. It was back down to 15K by about 14:00. The gain data for this period was not used in the fits. The figures show the dewar, OMT temperatures, and the system temperature computed using the cals during this period.

Figure 1 top plots the system temperature for the 700+ measurements made over August. The many curves are the za dependence of tsys. At sample 345 the dewar temperature warmed up and was back normal around sample 590.

Figure 1 bottom is Tsys versus za for all of the data. Each day is plotted with a different color. The green and blue * (august 12 and 13) occured while the dewar heated up and the cooled back down.

Figure 2 top is the temperature of the 1st stage amplifier. It warmed up at day 224.3 and got back down to 15 K by 224.7.

Figure 2 middle is the temperature of the OMT. It warmed up to 210 degK and didn't get back down to 104K for two days.

Figure 2 bottom plots (TsysData - TsysFit) versus day number where TsysData was measured using the cals and tsysFit was a polynomial fit to Tsys excluding the data where the dewar warmed up.

Tsys remained about 3 Kelvins above the normal value for about a day. Unfortunately there were no measurements for daynumber 225.3 to 225.8. The curve for Tsys looks like it is following the OMT temp rather than the dewar temp for daynumber 224.8 to 226. If we could separate out the contributions to Tsys from the 1st stage amplifier being 1-2K high and the OMT being still hot, we could predict the loss in the OMT and know how much it contributes to Tsys under normal operations. In anycase, when the compressor trips and the dewar heats up, the system temperature for lbw will not be back to normal for about 2 days (even thought the the 1st stage amplifier looks like it is pretty close to normal temperature).

System temperature at 1415 Mhz. (lbwgaintop)

The system temperatue at 1415 Mhz was computed using the cals and then fit to a 3rd order polynomial in zenith angle. The fit was to (polA + polB)/2 (stokes I/2). The days when Tsys was high because of the dewar warmup were not used. The za^2 and za^3 terms of the fit are only applied for za > 14 degrees.

Figure 1 plots Tsys versus azimuth and zenith angle. The length of the arrow is proportional to the system temperature (1 tick mark = 20K). The rotation angle is also proportional to Tsys. Straight up is the minimum value, 90 degrees to the right is were Tmin/T=.75 and down isTmin/T =.5 . The table at the bottom shows the average for all the data and then in 5 degree za steps. The sigma for (data-fit) is about 1K. The sigmas for the coefficients (at the bottom) show that the za^3 coefficient is not significant (value=-.00741, sigma .00887).

The fit for 1415Mhz is:

Tsys(za)=36.32 +.153*za + .17094*(za-14)^2 -.00741*(za-14)^3 degK

The expected za dependence should be flat out to about 18 deg za (where the beam falls off the ground screen) instead of the measured dependence of .15 K/degZa.

Gain as a function of frequency. (lbwgaintop)

Data was taken at 7 separate frequencies. The gain processing described above for 1415 Mhz was applied to each of these frequencies. The correction to the data for the dish alignment was computed and applied separately for each frequency. The figures show the variation of the gain with frequency.

Figure 1 top has the fit to the gain versus za for the 7 frequencies using the uncorrected data. The fit is evaluated at integral values of za. The vertical line at each za is the az dependence of the gain at that za.

Figure 1 bottom plots the gainfit verus za for the 7 frequencies using the data corrected for the southwest quadrant of the dish. The vertical line az dependence has decreased since the 1az and 2az terms have been removed from the data.

Figure 2 top plots the fit constant term versus frequency. The black line is before and the green line is after correcting the data. The 9% change between 1200 and 1700 Mhz would require a 5.5 mm rms surface. The pitch, roll, and focus alignment are also frequency dependent and probably account for a large part of this.

Figure 2.2 plots the linear za dependence versus frequency. It gets about 50% worse moving from 1200Mhz to 1700Mhz.

Figure 2.3 plots the amplitude of the azimuth terms versus frequency. The 3az term arises from the 3 fold symmetry of the platform. The pitch and focus measurements have shown this 3 fold symmetry.

Figure 2 bottom plots the frequency dependence of the phase angle for the azimuth terms. For the 2az and 3az terms I picked the peaks closest to 0 degrees.

Figure 3 is the fits to the uncorrected data.

Figure 4 is the fits to the corrected dat

It would have been nice if we did these measurements after the dish had been completely adjusted. As it stands the corrected fits look to be a few percent below the data taken after the dish adjustments.

The gain fits rely on the lbw cal values if the gain curve is used for different receivers (eg. lbn). For lbw, the fit can be used as a transfer from the flux of the measured calibrators to the unknown flux of the users source. This will work indendently of the cal values as long as they remain constant.

To use the fits to gain, see the idl routines corhgainget (idl routines cor) and gainget (idl routines general).

processing: x101/callb/Sep01/process.pro, inpsav.pro.

plotting  : x101/callb/Sep01/pltgain.pro,pltgaina,plttsys

 home_~phil