NIRCam Detector Readout

The JWST NIRCam detectors are read out non-destructively multiple times per integration either through one or 4 outputs simultaneously.

NIRCam detectors use MULTIACCUM readout mode

NIRCam detectors operate in MULTIACCUM mode; this means that the charge in each pixel may be read out multiple times as it accumulates during each integration. The readout process is non-destructive, leaving charge unaffected and in place—charge is not transferred between pixels as in CCDs. After each integration, the pixels are read out a final time and then reset pixel-by pixel, releasing their charge which is converted to data number (DN).

NIRCam repeatedly cycles through all pixels in the full frame or a selected subarray, either reading their charge or resetting them as commanded. Reading and resetting take the same amount of time: 10 µsec per pixel plus small overheads at the end of each row and frame.

Each detector readout delivers a frame of data. By selecting one of the NIRCam readout patterns, users may either save all individual frames or averages of groups of frames calculated onboard to reduce data volumes and enable longer integrations. Fitting "up the ramp" to multiple frames in an integration decreases the effective read noise, increases the dynamic range, and allows cosmic rays to be identified and rejected.

Readout of the full NIRCam detector (2048 × 2048 pixels) is (nearly always) performed with 4 outputs simultaneously (Noutputs = 4), each delivering a stripe of data (2048 pixel rows × 512 pixel columns), and taking 10.73677 s altogether. Smaller subarrays are read out more quickly, even though most are read out through a single output (Noutputs = 1).  Noutputs is pre-defined for most subarrays, but users are given a choice between Noutputs = 1 or 4 in the grism time-series observing mode.

Figure 1. Readout of a NIRCam detector

Readout of a NIRCam detector

NIRCam detectors may be read out through 4 outputs simultaneously. Each row is read out in turn, one pixel at a time, beginning with the pixels along the bottom marked with small, open, black squares. The orange border represents reference pixels, 4 pixels across (not to scale), that are insensitive to light but are used to correct for bias offsets and 1/f noise. Arrows show the default scan directions. Charge is read in place, non-destructively (it is not transferred between pixels as with CCDs) until an integration is complete.
Figure 2. Readout of a NIRCam subarray

Subarrays with fewer than 2,048 columns are always read out through a single output. Reference pixels are included in subarrays that are located along one or more detector edges (they need not be, and not all are). An exception is the multistripe subarrays used in .NIRCam Short Wavelength Grism Time Series observations, which are described in more detail in NIRCam Multistripe Subarrays
Figure 3. Readout of all 8 NIRCam short wavelength detectors

Readout of all 8 NIRCam short wavelength detectors

Orientations of all 8 short wavelength detectors in the field of view. Modules A and B are shown with all detectors labeled (A1–4 and B1–4). The 4 readout quadrants are shaded and readout directions labeled as in Figure 1. "First" marks the first pixel read out for the case of a single output for each detector. Coronagraphic occulting masks are shown as projected onto the detectors when in use.
Figure 4. Readout of both NIRCam long wavelength detectors

Readout of both NIRCam long wavelength detectors

Orientations of the 2 long wavelength detectors with readout directions labeled as in Figure 3.

Notable updates
    Added pointers to information about the multistripe subarrays used for SW grism time series.

  • Updated frame time from 10.73676 s to 10.73677 s
Originally published