Scientific CMOS (sCMOS) cameras offer a set of high performance features, including extremely low read noise, wide dynamic range, high quantum efficiency, and rapid frame rate within a "mid-range" scientific price bracket. In this note I compare the narrowband imaging performance of the Andor Neo 5.5 megapixel sCMOS camera to my Quantum Scientific Imaging QSI 683wsg 8 megapixel CCD camera. The cameras have similar pixel and sensor dimensions.
Narrowband imaging acquisitions are typically not "sky limited", that is, camera read noise is not effectively buried by sky background shot noise. Hence a camera with lower read noise and lower dark noise, maintained on longer acquisition times, can produce a result with a higher signal-to-noise ratio (SNR) and a lower detection limit.
The table below shows a longer acquisition time SNR comparison of the QSI 683wsg CCD and Andor Neo 5.5 sCMOS cameras, for the binned 1 x 1 and 2 x 2 cases. The sky background is assumed equal to 130 electrons (e-) for a 40 minute exposure with the QSI camera binned 2 x 2, the Takahashi FSQ-106EDX telescope, and the Astrodon 3 nm H-alpha filter. I normalized sky background to the other camera configurations by accounting for sensor quantum efficiency, pixel size, and binning differences.
The table shows the per pixel SNR estimates for two extended targets, a bright target with an intensity 2 magnitudes brighter than the sky background, and a faint target with an intensity 2 magnitudes fainter than the sky background. The table also shows the corresponding bright and faint speed ratio estimates, which are equal to the square of the ratio between equally binned Andor and QSI SNRs for the same targets. The Andor is about 4.1 times faster than the QSI in the binned 1 x 1 faint target case, and about 3.3 times faster in the binned 2 x 2 faint target case. The bright target speed ratios are smaller than the faint ratios since the camera noise contribution to total noise is smaller. The binned 2 x 2 speed ratios are smaller than the 1 x 1 ratios because there is no read noise advantage when binning the Andor camera. The table also shows root-mean-square (rms) sky background noise and camera noise, the latter of which is equal to the quadrature sum of read noise and dark noise.
The SNR estimates for the QSI camera correspond to a single 40 minute acquisition, and the mean of four 10 minute acquisitions using the rolling shutter for the Andor camera. Hence, the total acquisition time for all configurations are the same. A smaller number of proportionally longer acquisitions with the Andor do not significantly increase it's SNR. Acquisition temperatures are -20° C for the QSI and -40° C for the Andor.
|Camera||Binning||Image scale||Sky background noise||Camera noise||Bright SNR||Faint SNR||Bright speed ratio||Faint speed ratio|
|QSI||1 x 1||2.1 arcsec/pixel||5.7 e- rms||8.7 e- rms||11.5||0.48||-||-|
|Andor||1 x 1||2.5||3.8||2.5||17.4||0.96||2.3||4.1|
|QSI||2 x 2||4.2||11.3||14.9||23.8||1.06||-||-|
|Andor||2 x 2||5.1||7.7||5.1||34.8||1.92||2.1||3.3|
The Andor Neo 5.5 sCMOS camera offers other important advantages for longer narrowband imaging acquisitions over other scientific camera technologies, including a high anti-blooming factor, low lag, low non-linearity, high quantitative stability, and no etaloning.
Disadvantages of the Andor camera include the need for liquid cooling to achieve the -40° C temperature and possibly the lack of USB connectivity for control and data.