Dynamic range is a ratio of the biggest and the smallest signals that can be measured. We are using 16 bit ADC, so maximum signal is 65535 ADC counts. The smallest signal is the RMS signal (noise) of the dark current. It is, primarily, determined by read-out noise of the detector. For CMOS S11639 detector, used in Ariel spectrometer, the RMS noise is ~ 13 to 14 ADC counts. This gives the dynamic range of ~ 5000. It follows, that we can measure reflectivity from 100% to 0.02%. Of cause, 0.02% will be the detection limit (signal =noise); so we can accurately measure ~ 0.1% reflectivity difference at low reflectivity level. One such measurement is shown on Fig.1 (left). The amplitude of the interference fringes in reflectance spectrum is ~0.1% and they are very clear: the filmstack model is fit to the measured data and thickness/n&k can be accurately determined.
For comparison, SonyILX and Toshiba 1304 detectors that are used in many popular visible spectrometers, have DNR ~ 1000. It would be more difficult to make the measurement on Fig. 1 with one of those detectors. On another hand, high quality CCD detector like S10420 has DNR ~ 40K to 50K and one can, potentially, accurately measure 0.01% reflectivity. In practice, a very accurate calibration of fixed pattern noise would be needed to measure a signal at that low signal level.
The SNR provides a measure of the signal quality – it compares the power of the signal to the average power of noise. The largest SNR is achieved for the maximum signal. There are many sources of noise but the large signal is shot-noise limited i.e. the main contribution to the noise is coming from the photons shot-noise.Short noise is defined by the fluctuation of the detected photons and it is described by a square root of the number of photons. For CMOS S11639 detector, the full well electrons capacity is 80ke. Assuming ~ 0.6 Quantum efficiency, we have a maximum SNR ~ 350. It is clear that SNR is significantly lower than Dynamic range. For comparison, Sony ILX detector has ~ 250 SNR and Toshiba 1304 detector ~300 SNR. A high quality CCD S10420 has a maximum of ~ 550 SNR. How can we improve it?