Good target signal vs avoiding star saturation

[NigeB]

Hi All

I've started doing some imaging of M33, and am trying to figure out an appropriate exposure time. I'm trying to avoid too many saturated stars, and with a 240 second subexposure in L, I'm getting about 100 saturated stars in the frame, as shown in one of the JPEGs below, which has a maximum histogram stretch applied. This looked acceptable to me.

Obviously this leads to low signal levels in the image - the brighter emission near the core has a level in raw data of around 1500, compared to background level near the field edge of around 900-1000. Dark frame levels are around 300 ADU for the same exposure time and CCD temperature, and read noise is around 1.4 ADU according to Atik spec.

I could push the exposure time up, but the other image I've attached shows a single frame stretched to bring out the galaxy, and I'm finding that when I stack frames in DSS, the results look fairly reasonable (for me, at least).  I'm interpreting this as being consistent with the numbers above which suggest that at 240 seconds, the read noise contribution is pretty negligible. I'm attributing the granularity in the second image to shot noise, and that could be improved in a single sub by increasing the exposure time, but this would also increase the number of saturated stars.

So the question is: does this exposure time for subs seem sensible, or would you go longer/shorter?

I've put a raw fits sub on Dropbox here if anyone is really interested...

https://www.dropbox.com/s/1ojhby87qv9ggvu/M33_021116_001_L.fit?dl=0

 

Thanks

Nigel

 

[ChrisLX200]

This is a problem you run into with the Sony-chipped 460 (and my 490) - a shallow well depth of only 17-18k. It's easy to saturate star cores with this. If getting colour into the stars is really important to you then take a separate set of short RGB subs and layer them onto the image to replace the saturated stars. I would certainly not restrict length of exposure because of this effect though because you need the longer integration time to pull out the faint details, and I routinely used 20minute (narrow band) and 10-15min (broadband) using my 490EX.

ChrisH

[sharkmelley]

As a very rough guide, measure the standard deviation of noise in a bias frame and the standard deviation of noise in a star-free background area of one of the lights.  If this noise is at least 5x bigger than the bias frame noise then your subs are already long enough.  Just take lots of them.

Mark

[NigeB]

Hi Chris, Mark,

Many thanks to you both for your replies - that's really useful. Chris - I'd not thought about the separate RGB layering approach to restoring colour in the stars - I'll give that a try. Mark - I think you've convinced me that the signal level is just too low: I get a std dev of about 15 in the bias frames, and around 60 in blank sky regions of the lights, so it's clear that to get the best results I've got to drive up the exposure time, accept the increase in star saturation, and use Chris' approach to restoring this. 

Thanks again - much appreciated!

Nigel

[sharkmelley]

A noise ratio of 4:1 isn't bad.  If you increase the exposure time of your subs you will see a very marginal improvement at best.

Mark