Cooling the Quickcam
CCD dark current, and therefore dark noise, is a function of temperature. The cooler you get the CCD, the lower the dark current will be. Although there is a temperature at which dark current is essentially zero, there is a much higher temperature at which the dark current is "good enough" for your needs. It is easy to think that this point will be when the dark noise is below the level of digitization for a given exposure, and if one had a 16-bit converter, or even 14, or 12-bit, this might be true, but with a 6-bit converter, the value of a signal which is below digitization is still 1.5% of your total signal. Add any light to the image, and you bring that 1.5% up to where it is digitized.
The CCD output amplifier has a very large change in efficiency with changing temperature, and a comparison of the difference between a cooled and an uncooled frame will reveal nothing of any value. The dark signal's only interest is as it pertains to the SNR in an image. Making a dark frame at ambient might show a black background and white hot pixels, but if you pointed at an object, the signal would not rise to the point of digitization.
Knowing that there is dark current in the device, subtracting a dark frame which has a background which is at 0 ADU will be the equivalent of not subtracting a dark frame at all. The noise is there - if your dark frame doesn't show it, your dark frame is no good.
- Duration = 60
- WhiteLevel = 112
- Contrast = 185
The frame above was made with the camera at ambient. The settings were chosen to give a background level of around 32 - 44.
- Duration = 60
- WhiteLevel = 86
- Contrast = 185
The frame above was made with the camera at temperature. The settings were chosen to give a background level of around 32 - 44, as in the previous frame. The difference in the whitelevel shows the change in amplifier DC offset and gain due to the lower temperature, as well as the reduction in dark current in the device..
This reduction in dark current is due to a brute-force cooling by two 38 Watt Peltiers, with water cooling. The dark current is reduced, but not eliminated.
How much does this gain us? I was surprised by the results.The following images are a final image, and one of the nine frames which went into it. In this case, the camera was at ambient. Winter ambient, around 20F, but ambient.
- Duration = 240
- WhiteLevel = 77
- Contrast = 249
The following images are of the same object, but this time with the camera cooled. Unfortunately, the f-ratio has changed from 3.3 to 4.8, so it isn't exactly apples to apples, but it gives an indication. The above image was combined from nine frames, while the image below was made from six (the clouds came).
- Duration = 240
- WhiteLevel = 86
- Contrast = 255
The uncooled camera can make pretty good images, if one processes the raw data with the peculiarities of an uncooled camera in mind. Exceptional data is replaced by typical data, then the frames are averaged, bringing the SNR back up.