Individual exposure times
An individual exposure should have a minimum length such that the sky noise dominates the instrumental noise, which then becomes negligible. Astronomers refer to this state as being sky-limited.
The minimum exposure time depends on
To determine the minimum exposure time, you first have to determine the readout/dark noise. Take a bias image (zero seconds integration time) and measure the rms fluctuation in the image. Or take the read noise from the spec sheet. If it is given in electrons, convert it to ADU by dividing it with the gain. The dark current noise can be neglected if the camera can be cooled down sufficiently.
Then, measure the sky brightness minus the bias level in an image of your target. The square root of the difference provides an estimate of the total noise in the image. This number should preferrably be a least two times (better: three to four times) as large as the instrumental noise alone. If not, increase the exposure time until it is.
Under light polluted skies most images will be sky limited. The darker your sky gets, the longer you should expose.
Total exposure time
A light polluted sky does not mean that one cannot obtain deep images. The increased nightsky brightness leads to larger sky noise, hence one must take more images than under a darker sky. If the sky is n times brighter than the sky in a different location, then the sky noise is sqrt(n) times as large. One must then expose n times as long to reach the same depth (assuming that transparency etc are the same).
The image depth you achieve is called the limiting magnitude. There are different ways of measuring it. A common approach is to determine the magnitude of a point source that has an integrated S/N of 3 or 5 over the background noise in the image. The measurement error in the magnitude of an object is approximately given by N/S, i.e. the inverse of the S/N. For example, the magnitude of a source detected with S/N = 5 will have an uncertainty of 0.2.
As a rule of thumb, to increase the depth of your coadded image by 1 magnitude (a factor of 2.5), you must expose 10 times as long. To overcome the sky noise alone, you must already integrate 6.25 times longer (the square of 2.5). The rest comes from various instrumental effects and other uncertainties.
For example, if you reach mag 21 in 1 hour, you will need 10 hours to reach mag 22 under the same sky. Under a 10 times darker sky, it will take you 1 hour to reach mag 22.
Exposure times for colour filters
Many astrophotographers spend in total about as much time for RGB as they do for the L filter. For example, the LRGB exposure time ratios read 3:1:1:1. This is suboptimal, as the L image will record details for which then no colour information is available. In the final images this is then clearly visible as the faint levels lack many or even all colours. When imaging, one should aim for roughly equal exposure times in LRGB, that is 1:1:1:1. If the CCD is known to be much less sensitive in B or in R, then the corresponding filter can receive some more exposure time for compensation.
The subject of colour calibration is covered separately in the discussion of image processing.