Originally posted by simons-photography well having graduated into the more elite circle of my snobby "photographic society" with my recent purchase of a K10D and identical pair of lenses that a fellow member got with his new K10D at the same time as me (coincidence) I was invited down to the pub after the meeting
now one member brought up an interesting point: why do we need a shutter on a digital camera ? and do the digicams/bridge cameras and P&S have one at all ?
we discused it a bit and never really came up with a solution to the enigma as it was pointed out there is the option of just turning it on and off to expose the image,
So I ask you what purpose does the shutter serve in a DSLR ? I thought perhaps that by keeping the sensor in the dark it does not get filled with light or something, undounbtedly keeping it off when not required reduces noise due to heating of the sensor but thats a seperate issue.
please help us resolve dilema
Good question - there are in fact cameras and CCDs that have fully electronic shutters and some that use both!!!
All CCDs can be read out during exposure (hence the video feed to the LCD on digicams and now live view SLRs). However, one problem with electronic shutters on simple CCDs is "smear" or "ghosting" of highlights as photosite charges are shuffled across the sensor during readout as the sensor has to remain switched on to be read. If the exposure time is relatively long (say 1/30) the effect is quite small as the exposure time of the other pixels is minimal, but at faster shutter speeds (1/1000) it can be quite nasty. Its much better to stop the light before readout.
Now there are ways around this. An "electronic shutter" can simultaneously dump the charge from the photodiode into a non-photosensitive "scavenger diode" at a prearranged time simply by changing the polarity of the pins on each sensor row. Readout then takes place from the scavenger diodes so no smear occurs. These devices have the major advantage of requiring no mechanical shutter (lower camera cost and smaller size) and can achieve very high shutter speeds (and flash synch speeds). They are frequently used in phone cameras and video devices.
However the scavenger diode has the same capacity as the photodiode plus there is additional control circuitry at each photosite, so such devices (sometimes called inline transfer CCDs) typically have lower fill factors and well depth than ones without (which are called full frame transfer CCDs).
Whereas the former are widely used in phone cameras and other low end applications, and for high speed video applications, they are unpopular in high resolution dedicated still cameras because of their lower sensitivity and ISO performance. Using a mechanical shutter allows the sensor to have fewer components and more light gathering area, hence better sensitivity.
For video on a digicam it is quite possible to use a FFT CCD as the shutter speed remains quite low and the output is much lower res, and only displayed briefly when viewing, so the ghosting is less noticeable. However decent quality dedicated video cameras use electronic shutters and their sensitivity is less of an issue because they use far lower resolution sensors (with much bigger photosites) and use interline CCDs with high shutter speeds to freeze action on each frame.
So what about CMOS you may ask! Well, they dont show ghosting because its not the full photosite analogue charge thats shuffled along the readout row, only the output digital value from the AD converter.
However CMOS has a different problem: without adding more transistors to each photosite to add clocking synch controls, you cant switch them all "off" simultaneously, so moving objects are strangely distorted across the frame (so a mechanical shutter is still needed). More transistors means a lower fill factor - the main problem with CMOS sensors is crowded photosites (the simplest have 3 transistors - so called 3T sensors) adding more is not a favoured option for high quality devices.
However now that CMOS sensitivity and quality has mostly caught up with CCDs, 4T and 5T designs offer an advantage in very high speed video applications because the data can be read out in parallel across the whole sensor, allowing potentially much higher frame rates or much higher video resolution, or both.
Sorry - bit of an oversimplification but I am sure someone will correct me if I have not got it exactly right.