Advanced use

Amplitude self calibration
The amplitude calibration ( SELFCAL AMPLI) is a secondary step in the self-calibration process. In general, it should only be attempted if the phase calibration was already excellent, i.e., once you obtained the best solution by adjusting the SELF_TIMES parameter for the SELFCAL PHASE command. If possible or needed, the amplitude self-calibration should use a longer timescale than the phase calibration (typically, SELF_TIMES = 120 s). SELFCAL automatically adjusts the gains so that their mean is 1, in order to avoid changing the flux scale. In practice, it is useless if the expected noise limited dynamic range is less than about 300.

Cases where Amplitude self calibration may be essential
If PHASE self-calibration does not sufficiently improve the image despite ample signal-to-noise, Amplitude self calibration may do it. This situation often occurs when the observations span different dates, so that the relative flux calibration between the separate dates is inconsistent. In this case, a AMPLI self-calibration may be of great help. For compact sources, the flux consistency scale across dates can be also checked and cured using command SCALE_FLUX.

Note however that the resulting improvement does not necessarily produce a higher fidelity image. It removes the inconsistencies, but the selected (average) flux scale has no guarantee to be good. Flux calibration should be independently checked if this situation occurs.

Support restriction, flux threshold
Support restriction in the CLEAN process may be needed to build a simpler model for very complex, extended sources only. Command MASK THRESHOLD can be useful in this respect. Similarly, limiting the flux per pixel in the model (see SELF_MINFLUX) may help, since noise peaks are then ignored. However, the later may fail if the source is too extended: low level brightness can be important for self-calibrating short baselines.