L-Narrow Cal Values
New (better) cal measurements!
(Placed here 28 Oct. 2002)
|The Results||Data Description||The Plots|
|Freq [Mhz]||Pol A [K]||Pol B [K]||Error||Freq [Mhz]||Pol A [K]||Pol B [K]||Error|
On 28 January, 2002, the L-Narrow noise diodes (aka cals) were measured using the sky+absorber technique. As no cold load large enough to accommodate the L-band horns is available at Arecibo, measurements were only made of the noise diode values on the sky and with the absorber in place. This leads to considerable uncertainty, particularly with our calculations of the total radiation scattered into the sky observations. Additionally, the match of the signal from the sky and absorber into the amplifier (measurable by the reflection coefficient, gamma) is also uncertain. As a result, we cannot get a 100% uncertain measurement of the cals.
Measurements were taken with each correlator board having 25 MHz bandwidth, with 256 channels/board and 9-level sampling. The boards were centered 25 MHz apart, and after each 10s on+off observation of the noise diode, the boards were stepped by 100MHz. The entire 1200 - 1600 MHz range was measured 9 separate times on the absorber and 21 separate times on the sky. For consistency, the radar blanker (to eliminate signal from the FAA radars at 1330 and 1350 MHz) was turned on during the observations.
To avoid any errors which could arise due to a difference in gain/temperature measurements across the dish, the telescope was help stationary during the observations. As a result as a result, 4 of the 21 sky measurements showed some continuum emission as a source passed overhead. All four of those data sets were removed for our analysis.
In the data reduction process, the data was split into 64 channel (6.25MHz) sections and the mean value, versus frequency and polarization, was found. As a check of the measured values, the new cal values were then applied to all L-narrow "calibration" observations taken between August 15, 2001, and May 22, 2002 to obtain a gain curve at 1300, 1375, 1415, and 1500 MHz (the proscribed observation frequencies for the calibration routine). These curves were then compared with the L-wide gain curves, obtained by analyzing all L-wide calibration data taken during the same period. At 1300 and 1375 MHz both L-band curves matched to within 5%. At 1415 MHz, though, the discrepancy between the two gain curves was as high as 13%, while at 1500/1550 MHz the difference was 10%. (To reduce problems from poorly understood source fluxes, any gain measurement which was <6 K/Jy or >14 K/Jy was not included when determinined the average gain values.)
The actual gain curve which is used for the L-narrow receiver is a gain curved obtained with the L-wide receiver (info). As a result, the clear choice is to obtain a fit to the cal values which heavily weights the cal values "obtained" using the comparison of the L-narrow and L-wide calibration data, described in the last paragraph. With this in mind, a 5th order Lagrangian polynomial was fit to the cal values measured both using the sky+absorber method and using the values obtained in the comparison between the L-band receivers. To insure a higher importance was given to the values obtained from the receiver comparisons, those values weights were artificially made 10x the weight of the other points for curve fitting purposes. The final curve, along with the fits and errors, is given below.
As a final check, ON+Off position switched data was obtained on a standard calibration source (B2041+170) on 23 May, 2002. The source was tracked as it set for 1.25 hours. The data was taken with 25 MHz boards, 1024 lags/board/polarization, with the 1290 - 1500 MHz band covered in two observation, or 8 minutes of observing time. The FAA radar blanker was on for all observations, but the 1340-1355MHz range was not observed to avoid RFI. Again, the results of these observations are shown below.
Temperatures used for the calculations are: Tsky + Tscatter=20 K; Tabsorber=292 K
More info on the sky+absorber technique is avilable
How Do I Convert Flux Values for Data Taken After 15 August, 2001 For Which...
In IDL there are a couple of routines which provide cal values. Probably the easiest to use is the "calget1" routine. You need to use this routine to figure out what the old and new cal values are, and scale accordingly.
Example 1: You want to convert data for an object at 1300MHz. The output of the print command is your scaling factor.
Example 2: In this case, you have your data in a variable (i.e. "b"), and you and to get the conversion factor for subcorrelator 1.
In AIPS++ you should use the ao.getcalval routine in the same manner (rcvnum=6, freqin=1300, calnum-0, day=150/01, year=01/02, pol=0/1)
Here use need to figure out both the old and new gain values and the old
and new cal values.
Example 1: You want to convert data for an object at 1300MHz. You already know the azimuth and zenith angle of your observations (in this case, use az=205, za=11.2). Again, the output of the print command is your scaling factor.
Example 2: In this case you have your data stored in, i.e., the variable "b". Again, you are intested in converting the data in subcorrealtor 1, and the output of the print command is your scaling factor.
In AIPS++ you should use the ao.getcalval and ao.getgainval routines in the same manner (rcvnum=6, freqin=1300, calnum-0, day=150/01, year=01/02, pol=0/1 az=205, za=11.2)
What if the 5-10% error is just plain too high to be acceptable for my observations?
If you really need a more accurate measurement, your only choice is to get SEFD measurements over the part of the dish which you are doing your astronomical observations. Since SEFD does not rely on the cal values, your SEFD measurements will be as accurate as the flux of your source and the difference in positions across the dish of your source and the caalibrator sources (plus random noise, of course).