JWST MIRI's SLOW mode readout pattern offers fewer detector artifacts and slightly lower detector noise than the "fast mode", making it a good choice for faint source medium-resolution spectroscopy where the sky backgrounds are very low.
Long MULTIACCUM exposures can maximize signal-to-noise in detector noise-dominated regimes. SLOW 1 mode is therefore the default readout mode for medium-resolution spectroscopy (although FAST mode will be available for bright targets.) Just like in FAST mode, Ngroups > 4 is required (and Ngroups > 10 is suggested) for SLOW mode in order to optimize the calibration procedure, particularly cosmic ray removal.
Following MIRI readout terminology, SLOW mode obtains nine samples of a pixel (Nsample = 9; t1 = 23.88992 s). The first sample is ignored and the remaining eight samples are averaged to output a single result before being returned to the ICDH (ISIM Command and Data Handling). Nsample = 9 is a fixed parameter in SLOW mode and cannot be altered by the observer. The user can alter the number of groups (Ngroups) and integrations (Nint). All groups will be stored and downloaded.
From a pixel’s point of view, the SLOW mode pattern starts with a read-reset frame, followed by Ngroups − 1 read-only frames while integrating.
The focal plane electronics (FPE) dwells at each pixel for Nsample = 9 (90 µs in total). However, the FPE will only record the final eight measurements to allow for ample pixel settling time. Those eight samples will be averaged in the FPE signal chain and reduced to 16 bits before being sent to the ICDH. The time between sampling each pixel is t1 = 23.88992 s because the entire sensor chip assembly (SCA) is read before returning to a particular pixel. The integration ends with a read-reset after the specified number of Ngroups; this integration timing pattern may be repeated Nint times for an exposure.