OK, I'm probably going to upset some people here but...
WRT noise in astrophotography you have to understand that the noise apparent in a photo, or sub exposure comes from several different sources
- Sensor noise (dark current, temperature and exposure time dependant)
- Skyglow (always present, stronger during fuller phases of the moon)
- Light pollution (depends on where you're taking the shots)
- Read noise (always there, but varies from shot to shot)
- Others that I've forgotten about for the minute
By far the greatest of these for long exposures in or around major population centres is ... Light pollution.
Unlike most of the other sources of noise, LP is not subtractable, nor does it behave like other random noise, in fact it is better classed as unwanted signal. This and this alone, within 50km of a major city in Australia limit your subs from a DSLR to maybe 5 minutes maximum. You'll not gain a great deal (if anything, you may lose S/N) by going longer.
So, what to do to combat this? You can move to a darker site, temporarily, like what Greg does with his dark sky site. This reduces the N part of the S/N calculation. LP filters also do this to an extent, but are less effective. Or you try to increase the Signal part of the S/N, by using a more sensitive sensor, again this is better at a dark sky site and LP filters can help when at home. Narrowband filters and a nice sensitive CCD makes a huge difference in and around the city.
So, which is better, a cooled CCD or a modded DSLR, I'd go for the CCD everytime.
There's a handy calculator to work out the optimum exposure time based on the CCD and your background levels from a test exposure at
http://starizona.com/acb/ccd/calc_ideal.aspx
It doesn't list DSLR's, but I think one of the QHY cameras has a similar type of OSC chip in it?
Cheers
Stuart