The rms brightness temperature is related to the point source sensitivity
For a typical experiment with configuration BC or CD, tracking the source for 8 hours (but spending 2 hours on calibration) on each configuration, we obtain = 0.11 K and = 0.03 K, for 0.625 MHz channel spacing with the spectral correlator, with the same assumptions as above at 90 GHz. The factor is increased by uniform weighting and tapering: the amount depends on the uv-coverage and on the phase noise as a function of baseline lengths (1 to 3), but tapering also increases the synthesized beam size.
An equivalent way to look at noise in a map is by comparison to the
single-dish sensitivity (given here for position switching)
The mapping sensitivity is roughly related to by