| To Dan Johnson Home Page | |||||||||||||||||||||||||||||||||
| Demonstration of thermal noise in a CCD chip. Frames A through E are made using a Nikon Coolpix 995 camera with the lens cap on. The central 300-pixel-square portion of each image is presented. Images are by Daniel Johnson of Evansville, IN |
|||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||
| A | |||||||||||||||||||||||||||||||||
| At left, an 8-second exposure made at 70 degrees F. Nearly all of what you see here is thermal noise--in an ideal camera this frame would be completely dark, since no light entered the camera. |
|||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||
| B | |||||||||||||||||||||||||||||||||
| This frame was also made at 70 degrees F, but this time the exposure was for one full minute. The time-dependent increase in thermal noise is evident. One may try subtracting a dark frame (see below), but even in doing so one is still robbed of the ability to record detail. If one succeeded in subtracting out this noise perfectly, there isn't much room left to record the image. Nearly all of the dynamic range of many of these pixels has been used up. | |||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||
| C | |||||||||||||||||||||||||||||||||
| This image is the result of subtacting another one-minute exposure, also made at 70 degrees F, from the one above. It is a great improvement, but there are stll many hot pixels. This demonstrates the partially random nature of thermal noise. Each pixel has its own particular tendency toward such noise, but with a little variation from moment to moment. Hence no dark frame is ever perfect, and (as we shall see), this is worse at higher temperatures. | |||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||
| D | |||||||||||||||||||||||||||||||||
| Ah, the joys of cold! The camera was left outdoors to cool for 4 hours--the ambient temperature was 30 degrees F. This also is a one-minute exposure, just like B above. The only difference is temperature--and Wow! Now we're approaching a tolerable noise level. | |||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||
| E | |||||||||||||||||||||||||||||||||
| Nirvana! This is the above frame, D, but with another one-minute dark exposure subtracted (the second dark frame was also made at 30 degrees F). Now all is as it should be, and we can start making pictures of the night sky. | |||||||||||||||||||||||||||||||||
![]() |
|||||||||||||||||||||||||||||||||
| F | |||||||||||||||||||||||||||||||||
| Here's a glimpse of possible results--the central 1200 pixels shrunk down to 300x300, temperature 30 degrees F. I've made better M42 photos, but this isn't bad for a shot made under city lights and hazy conditions. It used the Coolpix, 40mm Plossl, Megrez 80 refractor, IDAS LPR filter and William Optics violet reduction filter stacked. Four one-minute and four 8-second exposures were combined using Registar, K3CCD Tools, and Photoshop. | |||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||