Access Official Technical PDF

Share Summary:LinkedInEmail

For engineers designing imaging systems to detect dim, sunlit objects in space, sensor sensitivity is only part of the challenge. A useful sCMOS camera must balance read noise, quantum efficiency, exposure time, optical performance, background light and platform motion. Modeling these factors together helps teams compare sensor candidates and understand the trade-offs behind a camera design.

Why sCMOS suits low-light imaging

Scientific CMOS, or sCMOS, describes CMOS image sensors designed for high sensitivity and low noise. In a typical CMOS pixel, a photodiode converts incoming light into electrical charge, and pixel circuitry converts that charge into a voltage. Parallel readout through column amplifiers and analog-to-digital converters supports rapid image acquisition while keeping read noise low.

That combination matters when a target produces a weak signal. Lower read noise can improve the signal-to-noise ratio, particularly when exposure times are short. Other sensor characteristics also affect the result: quantum efficiency determines how effectively incident photons are converted into signal, while dark current contributes charge even in the absence of light.

One evaluated sensor, the BAE Systems CIS2521F, has a 2560-by-2160-pixel array with a 6.5-micron pixel pitch. Its specified read noise is below two electrons RMS, and its quantum efficiency is above 55%. It also provides rolling and global shutter modes, two ADC channels per column for different light levels, and programmable region-of-interest readout. These are sensor-level features; system performance still depends on the optics, operating conditions and camera electronics.

Model the camera as a complete system

A sensor comparison is most useful when it reflects the intended camera and mission conditions. The camera model described in the source combines imager properties, lens parameters, target and background signals, and operational inputs. This allows engineers to explore how design choices affect measures such as minimum detectable irradiance (MDI), centroid accuracy and detection probability.

Relevant imager inputs include exposure time, pixel size, quantum efficiency, fill factor, dark current, read noise, ADC resolution, voltage range and gain. Some properties, including read noise and dark current, vary with temperature and radiation exposure, so assumptions about operating conditions should be explicit.

Optical inputs include field of view, entrance pupil diameter, point spread function and wavelength-dependent transmission. The model can also account for estimated stray-light noise. Background signals from sources such as solar-system and galactic light are considered alongside the target signal. Finally, line-of-sight angular rate and the required signal-to-noise ratio connect the camera design to how the platform will operate.

This system-level approach supports practical trade studies: for example, assessing how aperture or exposure time changes MDI, or how platform motion affects detection. It can also estimate whether closely spaced objects can be distinguished, based on predicted irradiance, image spread and angular separation.

Read noise, quantum efficiency and exposure trade-offs

A comparison of three candidate imagers illustrates why a single sensor specification cannot determine suitability. The evaluated devices were the OnSemi HAS2, MIT Lincoln Laboratory CCID-88 CCD and BAE Systems CIS2521F sCMOS. In the reported comparison, the HAS2 had 55-electron read noise and 45% quantum efficiency at 600 nm; the CCID-88 had 12-electron read noise and 90% quantum efficiency; and the CIS2521F had two-electron read noise and 55% quantum efficiency.

Under the modeled assumptions, the CIS2521F provided better MDI performance than the other candidates for exposure times in the 10-to-100-millisecond range, despite the CCD’s higher quantum efficiency. The result highlights the role of read noise when measuring weak signals over relatively short exposures: higher quantum efficiency does not necessarily compensate for substantially higher read noise in that operating regime.

The trade-off can change as exposure length increases. Dark current becomes more influential over longer exposures, and the evaluated CCD had lower dark current than the sCMOS device. For applications that need a long effective integration, the analysis considered stacking multiple short frames when registration accuracy is sufficient. That approach depends on the ability to align the frames; it is not a universal substitute for a long exposure.

Centroid accuracy is another consideration. In the reported simulation near a fifth-magnitude star, CCD and sCMOS centroid errors were close, differing by less than 0.03 pixels. The comparison assumed different sensor temperatures: minus 40°C for the CCD and room temperature for the sCMOS. The limiting noise sources also differed, with read noise limiting the CCD case and shot noise limiting the sCMOS case.

Account for stray light and dynamic range

Stray light—unwanted illumination reaching the detector, including light not rejected by the baffle—can reduce the useful dynamic range available for a target. Evaluating this effect requires more than a sensor’s nominal well capacity. The pixel well, amplifier behavior, ADC range, gain and offset all influence which signals can be represented without losing useful information.

Optical analysis tools can estimate stray-light levels, which can then inform camera-level modeling. Comparing dynamic range across exposure times and gain settings helps reveal whether background illumination may obscure the target signal. The model can also examine how platform angular rates affect the design trade space.

Design for the space environment

Optical and electronic performance must be considered alongside mechanical, thermal and radiation requirements. The described camera design uses a modular lens and baffle arrangement, with field of view and entrance pupil diameter adaptable to mission needs. Its lens design includes radiation-hard glass elements and venting between elements to address pressure differences during the transition to space vacuum.

The electronics separate the sensor board from the processing board, supporting modular testing and sensor replacement. The processing architecture includes radiation-mitigation features, while the camera interface and configuration are intended to support different mission requirements. Thermal analysis considers solar input and Earth albedo, with passive heat conduction and baffle treatments among the design factors.

Radiation response also requires evaluation rather than assumption. At the time described in the paper, single-event latch-up threshold testing for the CIS2521F was being planned, with a custom board intended to monitor device voltages and currents during heavy-ion exposure. Those plans should be understood in their original context, not as a statement of completed test results.

Practical implications for camera selection

For engineers and technical decision-makers, the central lesson is to select an imager against the complete operating scenario. A low-read-noise sCMOS sensor can be advantageous for short-exposure detection of dim objects, but quantum efficiency, dark current, stray light, motion, thermal behavior and radiation response all remain part of the design decision. A camera model that combines these parameters helps expose the trade-offs before hardware choices are fixed and supports more informed requirements for the sensor, optics and operating conditions.

Access Official Technical PDF

About The Author