Detector characteristics: Gain, readout noise

The camera system uses a Canon 500D camera body, with a 15Mpix CMOS sensor. Detector characteristics were measured and used to derive the system sensitivity and optimal photon-noise limited exposure times and ISO setting.

The detector noise was measured by differencing two short exposure dark frames at 800 ISO. The RMS deviation between the two images is divided by sqrt(2) to compute the readout noise.
The Gain was measured by differencing two images taken with a background level of ~1200 ADU (after subtracting bias), using green pixels only. A quadratic subtration of readout noise is performed to isolate and measure photon noise. Scaling to other ISO values is done assuming linearly of gain with ISO setting.

ISO 100 ISO 200 ISO 400 ISO 800 ISO 1600
Readout Noise [ADU] 10.8959 11.6364 13.9445 19.8761 32.2658
Gain [e-/ADU] 1.36 0.68 0.34 0.17 0.085
Readout Noise [e-] 15.8 7.91 4.74 3.38 2.74
(RON=photon noise) level [ADU] 161.5 92.08 66.11 67.16 88.49

The last line of the table shows the count level for which readout noise is equal to photon noise. Exposures should be sufficiently long to ensure that the background counts are above this level to ensure photon-noise limited performance.

Sky background level

The sky background level was measured under dark sky (no moon), with the 85mm lens at F1.2 and ISO 800 setting. Values are given in the table below for each of the 3 color channels. The blue pixels have the smallest counts. The minimum exposure times to ensure photon-noise limited performance are given by combining the sky background count levels with the previously derived minimum count level to ensure photon-noise limited sensitivity. The table below shows that under dark conditions, this exposure time ranges from 6.5 sec at ISO 1600 to 190 sec at ISO 100.
ISO 100 ISO 200 ISO 400 ISO 800 ISO 1600
Minimum exposure time (no Moon) 190.0 sec 54.2 sec 19.4 sec 9.9 sec 6.5 sec

Dynamic range as a function of ISO setting: optimal exposure time per frame

The dynamical range is the ratio between noise and saturation for a single pixel. All values given for a 1hr observation, assuming that the exposure time is chosen such that readout noide = sky background photon noise under dark conditions, and that readout time is much less than exposure time.

ISO 100 ISO 200 ISO 400 ISO 800 ISO 1600
Single frame exposure time 190.0 sec 54.2 sec 19.4 sec 9.9 sec 6.5 sec
Number of exposures (per hr) 18.95 66.4 185.6 363.6 553.8
Saturation level (e-/frame) 22282.2 11141.1 5570.6 2785.3 1392.6
Saturation level (e-/hr) 0.422e+06 0.740e+06 1.0339e+06 1.013e+06 0.771e+06
Dynamical range (1 hr) 6542 11471 16026 15702 11951

The best dynamical range is achieved by co-adding exposures taken at ISO 400 or ISO 800, with indivual exposures of approximately 20 sec (@ ISO 400) and 10 sec (@ ISO 800). In practive, this optimal exposure time cannot be sustained for a long survey: at 20 sec per exposure, 10 hr observation per night, the shutter lifetime (rated at 90000 exposures) corresponds to 50 nights of observation. Single frame exposure time is therefore a compromize between shutter lifetime and dynamical range . The imaging system currently operates at ISO 100 with exposure times longer than 200 sec in dark time to optimize shutter lifetime, and operation without shutter is being explored.

Dark current level

The dark current for this detector is extremely low, and difficult to measure. It is several orders of magnitudes lower than the sky. An upper limit on the dark current is obtained by measuring the flux on a moonless night when the camera is pointing down. This gives an upper limit of 0.0044 e- / sec / pix (to be confirmed by more accurate measurement), measured at temperature = 3 C. With this value, it would take a 55 days long exposrue to saturate the detector at ISO 100. The dark current level is about 300x lower than sky background on a moonless dark night.

Saturation limit

At ISO100, 120sec, saturation is reached on the best frames for mV~10.5 (approximate).

Links

http://astrosurf.com/buil/50d/test.htm : Quantum efficiency of Canon DSLR (C. Buil)