The correlator is composed of four independent boards, each a full correlator in its own right. One board can observe both polarizations, or just one of the two available from the chosen receiver, at a bandwidth and resolution that is set independently for each board. The signal may be three-level, three-level interleaved, or nine-level sampled. At bandwidths of 12.5 MHz or less, the signal to noise from three-level spectra can be improved by oversampling (i.e. using double-Nyquist sampling). The choice of any of these configurations is set by the ``lagsam'' variable in the Table , which is specified for each correlator board as an argument to the cor_lagc command.
| ``lagsam'' | Configuration | MaxBW per sbc | sbc/Pol | Boards used | Lags/sbc (best resn at 1420 MHz) |
| 0 | 9-level, A pol. | 25 MHz | 2 | 4 | 2048/(2.6 km/s) |
| 1 | 9-level, B pol. | 25 MHz | 2 | 4 | 2048/(2.6 km/s) |
| 2 | DON'T USE | -- | -- | -- | -- |
| 3 | DON'T USE | -- | -- | -- | -- |
| 4 | DON'T USE | -- | -- | -- | -- |
| 5 | 3-level A \& B pol. | 25 MHz | 4 | 4 | 2048/(2.6 km/s) |
| 6 | 3-level interleav. A | 50 MHz | 2 | 4 | 4096/(2.6 km/s) |
| 7 | 3-level interleav. B | 50 MHz | 2 | 4 | 4096/(2.6 km/s) |
| 8 | 3-level interl. A \& B | 50 MHz | 4 | 4 | 2048/(5.2 km/s) |
| 9 | 9-level A \& B pol. | 25 MHz | 4 | 4 | 1024/(5.2 km/s) |
| 0 | 3-level polarization | 25 MHz | 4 | 4 | 2048/(2.6 km/s) |
Double-Nyquist sampling can be used with all configurations, except interleaved, provided the bandwidth is less than or equal to 12.5 MHz. However, its use decreases the maximum bandwidth by a factor of two. 3-level, double-Nyquist, 12.5 MHz bw and below give 4 sub-bands with better resolution than the corresponding 9-level configuration.
9-level operation achieves 98\% of the signal-to-noise of analog correlation whereas 3-level sampling achieves 81\%. One advantage of 9-level sampling is to limit the effects of RFI (usually) to just a few channels.