Shortest useful subs - how short?

[sharkmelley]

For any of you who haven't seen this.....here's the definitive answer

I would argue that it is not the definitive answer because sky background glow and thermal noise are not taken into account. A different conclusion can be reached in the presence of strong sky background glow.  I'm sure I've seen an analysis somewhere that included the combination of all these effects in one formula.

Mark

[RobH]

I would argue that it is not the definitive answer because sky background glow and thermal noise are not taken into account. A different conclusion can be reached in the presence of strong sky background glow.  I'm sure I've seen an analysis somewhere that included the combination of all these effects in one formula.

Mark

A bit further in the thread skyglow and it's effect are taken note of Mark.

Thermal noise isn't an issue with a good CCD, but is obviously a problem with a DSLR. All other things being equal (which of course they never are), the conclusion is inescapable that long subs are better, and this is borne out by the experience of very many of us.

When shooting broadband subs, especially for colour data, I find that with the system I use these days, at F5.3 and binned 2x2, I quite quickly burn out star cores, so this obviously limits my sub length, as does skyglow, which is quite pronounced to the south for me.

With narrowband though, I regularly go to 30 minute subs.

If you can find a link to the analysis you mention that would be interesting to see for everyone I would think.

[sharkmelley]

If you can find a link to the analysis you mention that would be interesting to see for everyone I would think.

Here's a good example:  http://www.eso.org/~ohainaut/ccd/sn.html

Unfortunately it is all equations but it does consider the difference between read noise dominated cases and sky background dominated cases.

I find it mildly amusing when amateurs confidently doubt what the professionals take for granted.

For imaging faint objects it is often the case that wideband (L or RGB) imaging is sky background limited and longer exposures make no noticeable difference whilst for narrowband imaging, read noise is the limiting factor and exposures should be as long as possible.

Hope that helps!

Mark

[RobH]

Thanks Mark.....fairly straightforward to understand and precisely what I've found to be the case, especially with a fastish system that's also binned. For my rig, binning gives me an image scale of 1.14arcseconds/pixel, which is about my limit due to local seeing, so makes sense, but with luminance subs, it means that after about 6 minutes, I gain very little, if anything.

[davew]

You can always try the sub exposure calculator here - http://www.ccdware.com/resources/

Dave.

[glowingturnip]

so per that maths doc, the only noise that increases (ie decreases s/n) if you add many short exposures rather than take one exposure for the full time, is the read-out noise.  Every other source of noise is just a function of total exposure time.

As a humble wideband uncooled DSLR user, what does the read-out noise look like for a DSLR and how much should I be worrying about it compared to all my other sources of noise ?

I'm actually willing to bet that my dark current noise, specifically the amp glow I get in the corners of the picture in longer exposures, isn't linear either, but increases in rate with exposure time (hand-wavy pseudo science- amp glow noise is proportional to amp temperature and amp temperature is increasing with t).  Any picture I take with my DSLR beyond about 5 mins has some ferocious amp glow on it.  Then of course guiding error probability increases with exposure time too.

I am starting to think that different rules apply to someone like me with a modest rig and a dslr than someone with a really well-guided rig and cooled ccd, and that many shorter (ish !) exposures is better for me than fewer long ones.  For narrow band, cooled ccd on a great rig, clearly longer exposures will be better.

[dph1nm]

what does the read-out noise look like for a DSLR

Depends on your camera model, but modern DSLRs can have *very* low read noise (1-2e-), better than many astro CCDs. After all, if you are taking 1/60s shots in the dark then read noise, not dark current, is your main issue, so camera manufacturers have every incentive to improve it.

NigelM

[kirkster501]

I have always found two minutes as short as practical with a DSLR.  Five minutes is the maximum I'd ever go up to and I would allow 15s between subs to let the sensor cool down.  This may be a crock of BS but thats my experience with Canon DLSR's.  Your mileage may vary, to coin a phrase....!

[LouisJB]

I'm a bit confused about stacking, I don't seem to get any noticeable hot pixels or noise from long exposures (by long I mean 6 or 7 mins) on my DSLR (CMOS based) even at ISO 1600. For example a 7minute dark still looks black all over.

I still stack, but mainly it seems to help contract etc because most of the noise is coming from light pollution from what I can tell and so stacking seems to give a better SNR.

am I missing the point of darks etc or is this more useful for noisy CCDs?

[LouisJB]

contract = contrast, must learn to type...

[Physicist13]

I recommend having a read of Craig Starks articles on s/n http://www.stark-labs.com/craig/articles/articles.html

Note that, in general, "noise" refers to random fluctuations.

Patrick