QSI532 / QSI583
CCD camera performances comparison
The serial # of tested QSI532 is 502148. The serial # of tested QSI583 is 503318.
General feature and measured optoelectronic parameters

QSI 532 
QSI 583 
Pixel array 
2184 x 1472 pixels 
3326 x 2504 pixels 
Linear array size 
14.85 x 10.26 mm 
17.96 x 13.52 mm 
Pixel size 
6.8 µm 
5.4 µm 
Quantum efficiency @ 656 nm (source: Kodak specification) 
80% 
47% 
Measured gain (inverse) at middynamic 
1.340 e / ADU 
0.485 e / ADU 
Measured readout noise @ 12°C 
11.9 e 
8.7 e 
Measured relative quantum efficiency @ 656 nm 
1.62 
1.00 
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.
ThoriumArgon lamp spectra sampling (spectra taken with an eShel spectrograph)


QSI 516  CCD KAF1603ME 
QSI 532  CCD KAF3200ME 
QSI 583  CCD KAF8300 
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
Saturated 30 seconds exposure. 

QSI 583 model


The CCD lag of KAF3200ME is significant (QSI 532 camera). Lag is absent in KAF8300 image (QSI 583 camera).
Nonlinearity response
Measurement nonlinearity 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 nonlinearity 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 nonlinearity measurements for QSI 532:
a0 = 1.0007, a1 = 1.8712 x 10^{8}, a2 = 5.0816 x 10^{12}
Coefficients of polynomial fits to nonlinearity measurements for QSI 583:
a0 = 1.0054, a1 = 5.2482 x 10^{7}, a2 = 8.9886 x 10^{13}
The observed nonlinearity amplitude is similar for the two cameras in the explored dynamic range. The change in the gain is nearly 23% across the region from 3000 ADU and 63000 ADU, a reasonably low value.
Residual counts error before and after simple quadratic nonlinearity correction of the QSI 532 measured signal:


QSI 532 nonlinearity. 
QSI 532 residual nonlinearity after 2nd quadratic correction. 
Linearization
demo (click on the image for enlarge)
Dark signal
Thermal signal histogram for the same acquisition conditions:


Before gain normalization (signal proportional to ADU) 
After gain normalization (signal proportional to electrons  i.e. true dark signal) 
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:
QSI 532 
QSI 583 before normalization 
a 

QSI 532 
QSI 583 after gain normalization 