The CAT is a 3-element radio interferometer, built as a prototype for a larger instrument (the Very Small Array; see e.g. Jones (1996)). Its three 70-cm horns are mounted on a turntable, with baselines adjustable in the range 0.7-5 m. Operating at frequencies between 13 and 17 GHz, the CAT is sensitive to structures on scales of over a primary beam of FWHM (Robson et al. (1993)). Data are taken in two orthogonal polarisations, but the polarisation axis rotates on the sky as the telescope tracks. In 300 hours of good data CAT reaches a typical rms sensitivity of 10 mJy/beam (equivalent to rms) (O'Sullivan et al. (1995)).
As an interferometer, CAT is not sensitive to large-scale signals or ones which do not rotate with the sky, and thus efficiently filters out unwanted atmospheric emission and groundspill. This is the main reason why CAT can operate from an ordinary sea-level site (while other experiments seek high, dry sites or go into space). Groundspill is also minimised by surrounding CAT with earth banks covered with aluminium shielding.
To combat foregrounds including discrete extragalactic sources (mostly active galaxies) and emission from our Galaxy (thermal, free-free and synchrotron components) CAT uses two strategies: observations are made at three frequencies to remove Galactic emission (whose spectrum differs from the CMB), and we use the Ryle Telescope (RT; see Grainge in these proceedings) which operates at 15.2 GHz to measure and remove discrete radio sources. Using 5 of the east-west RT aerials in a compact configuration (baselines up to 108m), the RT achieves resolution. At this higher resolution, the RT is used to identify sources at 15 GHz by making raster scans of the central of CAT fields, down to a sensitivity of 2 mJy/beam rms, i.e.\ well below the sensitivity of CAT.