QSI-532 / QSI-583 CCD camera performances comparison

The serial # of tested QSI-532 is 502148. The serial # of tested QSI-583 is 503318.

General feature and measured opto-electronic parameters

Acquisition: Audela software. The electronic gain is extracted from a photon transfer analysis. Relative quantum efficiency is given by the ratio of integrated signal (circular aperture) on the same stellar images.

Thorium-Argon lamp spectra sampling (spectra taken with an eShel spectrograph)

The 50 µm fiber diameter image is very well resolved with the QSI 583 sampling. This oversampling increase probably the spectral calibration precision and facilitate ThorAr lines identification (more lines taken into account during calibration processing). Info about eShel spectrograph click here.

Lag phenomena

QSI 532 model

QSI 583 model

The CCD lag of KAF-3200ME is significant (QSI 532 camera). Lag is absent in KAF-8300 image (QSI 583 camera).

Non-linearity response

Measurement non-linearity method: the ouput signal (in ADU) from a stable light source is observed as a function of exposure time (or illumination, product of flux intensity and integration time).

QSI 532 model

QSI 583 model

Note: the experimental setup is improved for QSI 532 model measure (the light source is a white LED + a very stable power supply) in comparison to the QSI 583 model very noisy measure (the light source is a simple halogen lamp).

The relative non-linearity relative gain g is represented by the quadratic equation

where S is the measured counts (in ADU) minus bias.

The expected counts (measured counts corrected after linearization) S' is given by (in ADU)

Coefficients of polynomial fits to non-linearity measurements for QSI 532:

a0 = 1.0007,  a1 = -1.8712 x 10^-8,  a2 = -5.0816 x 10^-12

Coefficients of polynomial fits to non-linearity measurements for QSI 583:

a0 = 1.0054,  a1 = -5.2482 x 10^-7,  a2 = 8.9886 x 10^-13

The observed non-linearity amplitude is similar for the two cameras in the explored dynamic range. The change in the gain is nearly 2-3% across the region from 3000 ADU and 63000 ADU, a reasonably low value.

Residual counts error before and after simple quadratic non-linearity correction of the QSI 532 measured signal:

Linearization demo (click on the image for enlarge)

Dark signal

Thermal signal histogram for the same acquisition conditions:

The QSI 583 apparent dark signal in ADU (and thermal noise in ADU) is slightly superior. But if electronic gain difference is considered,after gain normalization (QSI 583 count level x 0.362), the true dark signal (in electrons) per pixel is very similar for the two models (but note also the ratio 1.26 between QSI 532 pixel surface and QSI 583 pixel surface, not taken into account for calculate dark current per surface unit).

Aspect of dark image for the same command temperature (-12°C), the same exposure time (60 s) and the same visualisation threshold:

Back