Sanity check: Atik 420 > 314L+ (long post!)

[codemonkey]

Hey guys,

After the problems I had with my 314e that ultimately led to me asking for a refund, I'd planned to re-evaluate the choice of CCD vs dSLR, which I've now done.

To my surprise, after articulating what was important to me, and assigning a weighting to each of those factors, then looking at what each option provided and the cost, it actually works out better "bang for buck" to spend more on a CCD. That's probably not news to you all, but honestly I was surprised.

Sine then, I'd planned on getting the Atik 314L+ (referred to as "314" from here on), which seems well regarded and is at the top end of my price range, so much so that I wouldn't be getting a filter wheel or filters for some time.

I've now done some reading on oversampling vs undersampling which has suggested that for my uses, the Atik 420 might actually be better, which was also very surprising to me.

First off, my primary interest is in shooting galaxies. I'm interested in photographing nebula also, but to a lesser extent. I'm shooting with a short tube refractor (600mm ED80) and I think I typically have pretty good seeing. As I understand it a good dark site will have seeing in the area of 1", and given that I'm 15mins away from the nearest streetlight in a semi-rural area I think that a safe bet would probably be seeing half that good (assumptions for the win!).

Following is the main stats I've found on the cameras / sensors:

FOV
420: 40.84' x 30.76'
314: 51.44' x 38.43'

Resolution (w/ 600mm scope)
420: 1.51"/px
314: 2.22"/px

Peak QE
420: ~ 65% @ 500nm
314: ~ 65% @ 500nm

H-alpha QE
420: ~ 30% @ ~ 650nm
314: ~ 55% @ ~ 650nm

Read noise
420: 4 e-
314: 4 e-

Pixel resolution
420: 1620 x 1220
314: 1392 x 1040

Pixel size
420: 4.4
314: 6.45

Ideal pixel size

According to this source (http://www.astro-imaging.com/Tutorial/MatchingCCD.html), the ideal pixel size can be calculated as star size / 3.3, where:

Star Size = (Seeing * Focal Length)/206.3

I think a pessmistic assumption here would be a seeing of 3. So this gives me:

Ideal pixel size = (((3 * 600) / 206.3) / 3.3) = 2.64. Of course, a better seeing value would require even smaller pixels!

So, based off all of this I'm reaching the following conclusions: the 314 would be significantly better than the 420 in terms of FOV and sensitivity if my primary interest was in photographing nebula. The downside would be significant undersampling and lower image resolution.

The 420 has better cooling, a smaller FOV, peak QE parity, better matches my OTA and will require significantly longer exposures for nebula. Combining later with a focal reducer should give it a better FOV for wide-field nebula shots and offset the lower H-alpha QE to some extent.

Based on all of the above, I'm really feeling that the cheaper 420 is actually a better choice for me. Does anyone have any advice that might contradict this? If not, I'll probably order a 420 tomorrow.

Cheers,
Lee

[Terry B]

You have made a good summary.
I have a few comments.
I live at 1100m altitude in dark rural site. Getting better than 2arcsec seeing is rare for me. It does occur but not often. 3 is the norm.
Galaxies are little objects and most are relatively dim. You are using a short focal length resulting in a small image size and compensating for this by using small pixels.
There is a reason that pro telescopes are big. They collect more actual photons and have a higher resolution. A larger diameter scope and larger pixels collects more photons for the same image scale.
The short focal length refractor is best at wide field imaging.
I think hunting for increased resolution and hence higher magnification with little pixels is not the best way to go. A bigger scope is needed.

[Merlin66]

I use the 314L for spectroscopy....
I'd be amazed if you get consistent seeing of 1 arc sec. A dark sky site doesn't mean good seeing.
IMHO I be assuming around 2 arc sec on the best of nights...
Both are good cameras....

[codemonkey]

Thanks Terry :) Good points!

I guess my major concern was the the impacts of such significant undersampling on the quality of images produced.

It seems to me that I'm already significantly undersampled with the 420 and the 314 might be taking it too far. Even if I switch to a 1.2m focal length I'd only be marginally oversampling with the 420 and still undersampled with the 314.



I think there's been some miscommunication here :-)

I wasn't thinking that I would get 1" here, I figured that if 1" was considered good seeing at a dark site that I should get about half as good. It might be that I was being overly optimistic though, it seems like 3 might be a more realistic number :-)

[Shiraz]

The following modelling results may possibly be of some use.

314
1.14 pixels/starFWHM in 2 arcsec seeing (heavily undersampled)
ideal broadband sub is 3 minutes under fairly dark sky

420
1.68 pixels/FWHM (undersampled)
ideal sub is about 7 minutes
takes more than twice as long as the 314 to get to a given SNR

As Terry says, a larger scope would be useful, but to image galaxies with the 80mm, the 420 seems to me to be by far the best choice - not much point in getting a galaxy image in half the time if it ends up too small to see.

Re seeing, "good" around here means about 2.5 arcsec. It occasionally drops below 2, but not often - and it can quite happily go over 5.

regards ray

edit: one other thing that you might consider is that, if you ever decide to upgrade scopes, the 314 would be suitable for a small SCT, RC or larger Newtonian, whereas the 420 is best with scopes up to about 1m fl and you would be limited to refractors or fast Newtonians.

[codemonkey]

Hmm, good points, Ray.

I have been thinking about one of those GSO RC8's in the future, but I've done absolutely no research into it and I'm not sure if my CG5 would handle one. If I did get such a scope, the 314 would be near perfect...

[codemonkey]

Well, in the end I bit the bullet and bought the 314L+ like I'd originally planned. I've seen some images shot with that camera on some really short tubes and the results didn't look bad undersampled or not. Seems like a smart move if I'm planning on getting something in the area of 1.6m later.

[Shiraz]

:thumbsup:
If you haven't already tried it, you can use dither while gathering images and then drizzle super-resolution stacking to recover much of the detail and image scale lost by undersampling.