Spectral Instruments Series 800, User Manual, Page: 31 / 70

3. Running The Cooled Camera
Once the camera head has become thermally controlled it is possible to verify some
camera performance metrics, as described in subsequent sections.
3.1 Image Quality
Now that the camera head is cold, a bias image is a more uniform array of gray
values with a lower-level ramp along columns. The image is otherwise pretty
uniform, or else there is a light leak where the camera head joins the application
or within the application itself (assuming there is no shutter between the camera
and the application). Two bias images taken in succession and subtracted are
flat, although you may be able to notice the increase in noise (because of higher
dark signal) at the side away from the serial register.
Dark images from moderately cooled cameras exhibit a wide variety of patterns
that are typical for different CCDs but may vary in visibility from one CCD of a
type to another of the same type. In deeply cooled CCD cameras, you can expect
fairly uniform dark images. In moderately cooled CCD cameras, dark non-
uniformity is the rule. Many CCDs show wafer-dependent patterns in the dark
signal.
With that said, a 100-second dark image shows a structure similar to the bias
although the background will be higher (depending upon the CCD and the
temperature of the camera). The cold-CCD dark image integrates both internal
dark sources and extraneous light (again assuming no shutter) and it shows up
light leaks in a way that short exposures typically do not. Light leaks are usually
not uniform so they are revealed in dark images. Very low-level light leaks can
emulate dark signal – so be wary.
3.2 Performance Metrics
Continue to avoid exposing the CCD to light. It is time to measure some
performance parameters. The SICCD camera meets a number of primary
performance metrics two of which can be verified without other instrumentation.
These are:
3.2.1 Noise
The camera readout noise is determined by the rms of a region from a bias-only
section of the CCD. Obtain a overscan bias image and note the rms of the
ensemble of pixels in a region of the image that is off of the parallel register.
The camera test report shows the conversion factor from counts to electrons for
each readout speed and attenuation state. Multiply the rms value by the
conversion factor for the settings you are using to determine the noise in
electrons. The result will be the noise reported in the test report ±10%.