RF calibration

The goal is to measure the receiver bandpass (RF) to calibrate the source data. For this, a bright calibrator is observed for some minutes. By default, in a first calibration the procedure Select selects the brightest observed source for the RF calibration. The selected RF calibrator can be changed by modifying the variable ``band_source'' as explained in Sect. . If several gain ranges are found, an independent RF calibration is performed for each range.

The procedure pauses to propose fits for amplitudes and phases for each correlator input (as shown in Figure ); by default the solutions for the Narrow correlator inputs are obtained with ``solve rf 12 20 /plot'', where ``12'' is the polynomial degree to fit amplitudes and ``20'' for phases (higher degrees are used for the Widex correlator units).

Figure: Example of the RF calibration of data obtained for the second Narrow correlator input

Changing the polynomial degrees with respect to the ones proposed by the automatic procedure rarely improves the RF calibration. In the pause, ``continue'' (or ``c'') should be typed to accept the solution and store the calibration parameters in the hpb file.

Sometimes the fit of the bandpass looks poor with deviations of a few degrees (in phase) or a few percents (in amplitude) between the measured data and fits are observed. For the time being there are no simple means to reduce these small differences. In general, they have no influence on the results.

Sometimes the procedure crashes (when fitting) because the data of an antenna is flagged. Two solutions exist: (1) Mask (see point 2 in Sect. ) shadow, saturati[on] or redu[ction] flags to use these data for the RF calibration if the signal to noise ratio is good enough. (2) Select another calibrator by typing ``let band_source `a_new_calibrator' '' as indicated in Sect. .

If delays are observed (constant slope in phases vs. frequency for one or more antennas) we may correct for it to decrease the phase decorrelation within scans. By using the commands ``solve delay /plot'' and ``modify delay'' we can correct for this (see Sect. ).

Differences in the LO1ref frequencies result in frequency offsets in the RF amplitude profiles. If the AoD (through the project notes) informs you on the presence of a big LO1ref jump (which is often linked to cable phase jumps) we recommend to calibrate the RF with the strongest phase calibrator observed cyclically with the source, for which the source LO1ref frequency was properly considered.