Camera Labs

Hi folks,

With a major upgrade to my telescope in progress during the summer, something that will also include purchase of a dedicated astronomical camera, I've been trying to decide what the optimal pixel size would be. In the process I came across the article CCD pixel size which I thought I'd share.

Given the usual limits of astronomical seeing it turns out that the 5.4µm x 5.4µm pixels of the Kodak KAF-8300 CCD are barely small enough for the best degree of oversampling at my expected focal length of 660mm though I could fix that with use of a 1.5x extender on exceptionally good nights so long as field of view considerations permit. Astronomical seeing, which probably won't be better than 2 arc seconds on more than a handful of nights each year from my location, means that for long duration exposures (longer than about 1/50th of a second) I don't need to worry too much about the telescope's Dawes' limit of about 0.9 arc seconds when choosing a CCD. That wouldn't be the case, of course, for planetary imaging using the lucky imaging technique but that's why web-cams, with their very small pixel sizes, are often chosen for those tasks.

The author also has some interesting thoughts to share about f-ratios for astrophotography which you can read at Sampling, Signal-to-Noise and the f-ratio Myth. I'm still trying to digest that one...

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

That last link proved a challenging read... I think the key point you need to mentally note before going there is that is done from an astronomer perspective, not photography perspective. And as my photographer's hat is a lot bigger than my astronomer's hat, it doesn't read easily.

The "f-ratio myth" presented, if I am reading it correctly, is that for the same basic optical device (telescope or whatever), using optical modifiers to increase the speed (reduce f number) would at the same time also reduce the focal length (like teleconverters). While you gain in speed, you lose spatial information for a given object and the two cancel each other out, resulting in no net gain. A similar argument applies on pixel sizes. While bigger ones get more light, you lose spatial information in the process.

What is NOT being discussed is the more intuitive photographer's scenario, in that if you have the same focal length, if you vary the f number lower it is definitely collecting more light. But in this case, the physical aperture will also need to be bigger. In astronomy, bigger aperture is usually better

_________________
Canon DSLRs: 7D, 5D2, 1D, 600D, 450D full spectrum, 300D IR mod
Lenses: EF 35/2, 50/1.8, 85/1.8, 135/2+SF, 28-80 V, 70-300L, 100-400L, TS-E 24/3.5L, MP-E 65, EF-S 15-85 IS
3rd party: Zeiss 2/50 makro, Samyang 8mm fisheye, Sigma 150 macro, 120-300 f/2.8 OS, Celestron 1325/13
Tinies: Sony HX9V.

Hi popo,

I agree with your analysis. Not in answer to it, but in continuation of the discussion in my initial post, I've read, from at least one source that I respect very highly, that f-ratio does matter when CCD imaging and, to be fair, when one has a fixed objective lens diameter and one CCD camera then varying the f-ratio by means of reducers or extenders does indeed "prove" that f-ratio matters, albeit by trading image resolution. But there's one instance where I believe the f-ratio myth argument is spot on and that's when specifying a new system as I hope I can make clear below.

I'm still window shopping for a new imaging system and I've settled on a refractor as I want portability and am too lazy to be possibly re-collimating every time I move the telescope on and off the mounting. After that the seeing gets murkier. For example, TeleVue do a very nice 127mm (objective diameter) f/5.2 'scope designed for imaging and I thought that a camera based on the KAF-8300 chip with 5.4μm pixels would be a natural complement to it. But for a little less money TEC do a 140mm f/7 'scope plus field flattener with the downside that should I use the same detector then my field of view drops by a factor of about 1.5. But if there were a similar CCD, in terms of total pixel count, but with 8μm pixels I'd be getting the same resolution in terms of arcseconds per pixel and, of course, getting about 21% more photons per pixel as a result of the larger objective diameter. Just to reiterate the point, provided the resolution, expressed in arcseconds per pixel, is the same then f-number is actually irrelevant when CCD imaging, particularly in reference to extended objects as well as stars. The only number that matters under those conditions is the diameter of the objective lens (or mirror).

To confuse matters even further for my own choice of imaging system there actually is a CCD with slightly larger 9μm pixels based on the KAF-6303E chip. Only 6.3MP compared to the 8.3MP of the KAF-8300 but, and it's a huge but, the KAF-6303E is significantly more sensitive, particularly at the all important H-alpha wavelength where I think I'll be spending much time. Net result is, if I've got my sums right, that the 140mm f/7 plus KAF-6303E combi detects Hα photons at over 1.5 times the rate (per pixel) of the 127mm f/5.2 plus KAF-8300 combo whilst achieving very similar arcsecond per pixel resolution. About half the gain is down to the larger objective with the balance being made up by the better QE.

The gotcha in this case is that, while the TEC 'scope may be a little cheaper, cameras based on the KAF-6303E are a lot more expensive than those based on the KAF-8300 so I'd have to up the budget. But I'd have to up the budget a lot more if I stuck with the 5.4μm pixels of the KAF-8300 and increased the objective diameter to count photons at the same rate while decreasing the f-number to maintain the same angular resolution per pixel as I'd achieve with the 140mm f/7 plus KAF-6303E combi.

All very confusing and I've still got a lot of thinking to do, especially as the TeleVue system has custom designed reducers and extenders, but it's only because I realised that f-ratio can be increased without a loss of "speed" provided that each CCD pixel individually collects light from the same area of sky that I even entertained the possibility that an f/7 system can compete with an f/5.2 system.

Phew, my head hurts, especially as I composed the above after a long day behind the wheel. Anyone care to prove me wrong...

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

I think I'll need to sit down with a hot cup of tea for this one. Not sure I'll be much help!

.
While Google Imaging this afternoon I came across this image captured with the TEC 140 and a camera based on the KAF-6303E, exactly the same configuration I was discussing yesterday:

It required 43 x 10 min subs at OIII, 30 x 10 min at SII and 50 x 10 min at Hα for a total of 20½ hours of imaging but patience had its virtue - I guess that's the deal when narrowband imaging.
Not sure I'll ever do so well but at least I have proof that that particular combination of equipment can produce spectacular results so any failures will be down to operator error rather than imaging at only f/7.

And talking of patience I've finally come off the fence and ordered a TEC 140 and will now have to wait about four or five months as I'm currently number 51 on the list. But at least I've followed up on my assertions above and put my money where my mouth is (or should that be keyboard? ). Time will tell if I've got it right.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Depends entirely on the type of image you wish to acquire, in other words the type of imaging you want to specialise in. The one you show is a nice enough false colour image of a small part of the Veil - and it took over 20 hours of good imaging time to get this small region (that is typically 5-6 good nights with excellent conditions, not common in the U.K.). Huge investment for getting just a small section of the nebula.

Now if you want to get the WHOLE of the Veil nebula, in true colour, at reasonable resolution, in less than a lifetime's imaging time - then you need a different setup (as discussed in my book).

You can then do something like this:



Note this is a downsized 1024 x 920 image so is not representative of the full resolution image. Processing as usual by Noel Carboni and includes narrowband (taken on the one-shot colour camera) as well as full RGB.

Greg

_________________
Nexstar 11 GPS, 2 x Sky 90, M25C, MaximDL, Photshop CS3, Noel Carboni's Photoshop actions, 7 foot Pulsar fibreglass dome, Canon 40D, 100mm macro lens, 28-200mm zoom lens, 17-55mm f#2.8 zoom lens, 100-400mm zoom lens, 1.4x converter, 2x converter.
http://www.newforestobservatory.com/
http://www.flickr.com/photos/12801949@N02/

Hi Greg,

Lovely image, as usual. I'm assuming that it was taken with the Tak 90 at f/4.5 with a resultant 3.97 arcseconds per pixel as that's the widest field of view you mention in your book. If I opt, as I hope to be able to do, for the KAF-6303E and its 9μm pixels then I'd be getting about 1.9 arcseconds per pixel. The pixel count is almost identical to that of the SXV-M25C so I'd have to mosaic 4, or more realistically 5 following the advice in your book, images to get the same sky coverage. If the seeing was good enough that mosaic should have twice the linear resolution, of course, but that is still potentially a lot of extra time at the telescope.

But, and you knew I was going to offer a but didn't you , the TEC 140 captures photons at about 2.4 times the rate of the Tak 90. Those photons are spread more thinly (1.9 arcseconds per pixel compared with 3.97 arcseconds per pixel) by a factor of about four which would imply that the Tak 90 imaging system would be capturing photons at each pixel at about 1.6 times the rate of my intended TEC 140 system. That would certainly seem to be a win for the Tak 90 when going really deep when the desired signal is only just creeping above the noise from the sensor though maybe there's a little compensation from the slightly higher QE of the KAF-6303E and, of course, the fact that it can use all of its pixels when narrowband imaging while the Bayer matrix of the Sony ICX453AQ chip is, when Hα imaging, only using one in four of its pixels

I think the problem with so many discussions about the "f-ratio myth" is that one never ends up comparing like with like. I believe I can argue quite successfully that if one put some mythical sensor with 18μm pixels and a 6MP count behind the TEC 140mm (objective diameter) f/7 'scope then it would, to a first approximation, outperform the Tak 90 plus a monochrome (non Bayer) equivalent to the SXV-M25C by that factor of 2.4 per pixel, as implied by the different objective lens diameters, as each pixel on the two systems is seeing the same area of sky. But 18μm pixels and a 6MP pixel count would not only be horrendously expensive in it's own right but it would need 79mm filters!!! But it would image at 2.4 times the rate of the Tak 90 at f/4.5 despite only being at f/7.

My aim in putting together my new system is to achieve about a 2 arcsecond per pixel resolution. Within the constraints of a reasonably high pixel count and 50mm filters that means employing a sensor with pixels of between 5.4μm and 9μm if I'm reading the various spec sheets correctly. That implies a focal length of between 557mm for 5.4μm pixels and 928mm for 9μm pixels. After that the choice would be to go for the largest aperture I can afford (at good quality). The only downside of my final choice of the TEC 140 f/7 with it's 980mm focal length is that, after discussion with Yuri at TEC, I know that it is not really compatible with focal length reducers but a 0.8x reducer isn't going to make such a huge difference anyway.

But hey, this is all theorising on my part and all the theory in the world can be disproved by practical experiment. Maybe I'll be eating my words (or should that be "eating my keyboard" ) in about six months time. I'm particularly interested in not only imaging the various emission lines (the usual suspects) but also imaging at the adjacent narrowband continuum wavelengths where I expect only stars to be visible thus gaining significantly from the 140mm aperture. The aim is then to combine that data in various ways during PP and hopefully achieve distinctiveness in the process. This will mean lots of time spent on relatively few objects so I'm going to have fun exploring the universe however it goes even if I can't go incredibly deep.

Bob.

P.S. I hope you'll forgive my rather laboured approach above and I certainly hope I've made no basic errors in logic or, indeed, "sums". I know that with your academic background this is is "Basic Optics 101" to you but I have to go a little more slowly. f/7 compared to your own f/4.5.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Hi Bob,

I think your numbers are pretty much spot on - you've done your homework and passed DS101 with top marks
The theory does break down in practice however. Look at Dietmar Hagar's work to see how (surprisingly) the one-shot colour fares against mono in narrowband imaging - I found the same result in practice. I will be using the new M26C camera on the mini-WASP (Sky 90s) and this will give 3 arc seconds per pixel, same field of view as the M25C, but 10Mp rather than 6Mp - so for what I'm doing it's ldeal.
The results you are expecting from your combo are indeed what you should get, and you have gone much further at this stage than I did (I bought the bits with half an idea and trusted to luck). The final bugbear in what your final image will look like is your local skyglow - and that opens up yet another can of worms
Good luck with what looks like will be a great imaging setup.
Oops - almost forgot - what are you going to use for the mount? All these numbers mean very little if the mount can't meet the rest of the system spec.

Greg

_________________
Nexstar 11 GPS, 2 x Sky 90, M25C, MaximDL, Photshop CS3, Noel Carboni's Photoshop actions, 7 foot Pulsar fibreglass dome, Canon 40D, 100mm macro lens, 28-200mm zoom lens, 17-55mm f#2.8 zoom lens, 100-400mm zoom lens, 1.4x converter, 2x converter.
http://www.newforestobservatory.com/
http://www.flickr.com/photos/12801949@N02/

(congrats on ordering your scope Bob, can't wait to see and hear how you get on with it!)

.
Thanks Gordon.

@ Greg

It's going off-topic for this thread but in answer to your question I have an ASA DDM60 PRO mount purchased earlier this year and sitting in the conservatory waiting for the pier to be constructed at a suitable spot which, unfortunately, will still have limited southerly viewing. It's a direct drive system with torque motors and no gears so no backlash or periodic error to speak of, the 0.02" encoder resolution is way better than I should need (even after factoring in the necessary oversampling for the control loop) and I've even heard of correctly set-up systems running unguided to sub arcsecond accuracy over 15 minutes or more. That's the theory awaiting verification (another one where I've put my money where my mouth is ) but the mount will, of course, accept a guider input if needed.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

.
Another stunning example of what can be achieved with 9μm pixels behind the f/7 TEC 140 and its field flattener (click the image to see how it was done):



The full size version here.

There are only two downsides.

1. The image needed 6 hours and 45 minutes to capture.
2. The camera used was the $10,000 FLI ML 16803-65

The camera is so expensive because it has a 4096 x 4096 x 9μm pixel array which is 36 x 36 mm in size (the KAF-6303E I've mentioned above has just 3072 x 2048 x 9μm pixels).

But on a wider point I do find it quite incredible that this sort of imagery is now within reach of the amateur for about the price of a luxury saloon. I reckon if you'd showed the guys
at, say, Mount Palomar something like this 50 years ago and offered that it was captured in someone's back yard they'd have been straight onto the 'phone calling the local lunatic asylum
on your behalf.

This is world class imagery and I'm sure I'll never get even close to it but, my, there's a powerful incentive to try one's best.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Third downside - Kodak chips are much noisier than Sony

Greg

_________________
Nexstar 11 GPS, 2 x Sky 90, M25C, MaximDL, Photshop CS3, Noel Carboni's Photoshop actions, 7 foot Pulsar fibreglass dome, Canon 40D, 100mm macro lens, 28-200mm zoom lens, 17-55mm f#2.8 zoom lens, 100-400mm zoom lens, 1.4x converter, 2x converter.
http://www.newforestobservatory.com/
http://www.flickr.com/photos/12801949@N02/

Hi Greg,

I know Starlight Express make much of that on their site but it is so difficult to compare like with like. Echoes of the f-ratio myth problem in that respect.

The Starlight Express page (here) for the Sony ICX285AL quotes a dark current of "less than 0.02 electrons/second @ + 10C ambient". That seems pretty remarkable and Apogee Imaging Systems quote (here) a figure of 0.01 eps (-15°C) which is definitely in line with the Starlight Express quote. Note the meagre full-well capacity of just 18,000 electrons though. For the KAF-16803 used above Apogee quote (here) a figure of 0.2 e-/pixel/sec (-25°C) though they do offer 0.03 eps in their low temperature housing. But at least the KAF-16803 has a full-well capacity of 100,000 electrons so subs can go longer without blowing the highlights. That would still seem to be a clear win for the Sony on the dark current stakes until one notices that the FLI spec (PDF) for their ML16803 with the KAF-16803 quotes a dark current of "<0.02 e-/pixel/sec. @ -30º C". So there are lies, damn lies and spec sheets it would seem

I've no idea how to prove or disprove these claims short of winning the lottery and testing samples from both Apogee and FLI. That's not going to happen but it seems to me that for subs of the order of 10 minutes one is likely to see a dark signal of the order of around 6 electrons from the Sony chip. But as the dark current will vary somewhat pixel to pixel maybe it would still be worth taking darks even with the Sony to get the absolute best signal to noise ratio? But hey, you've used the Sony chip to great effect so have you ever found a need to take darks with it when going deep?

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.

Hi Bob,

I've never taken a dark I also think Terry Platt (starlight) has a point when he says darks can actually introduce noise with the Sony chips and he does not recommend taking darks with them. Starlight are not partisan when it comes to imager chips - after all they sell both Sony and Kodak chip based imagers. I have seen darks from an H36 (Kodak mono imager) - not impressive.

Greg

_________________
Nexstar 11 GPS, 2 x Sky 90, M25C, MaximDL, Photshop CS3, Noel Carboni's Photoshop actions, 7 foot Pulsar fibreglass dome, Canon 40D, 100mm macro lens, 28-200mm zoom lens, 17-55mm f#2.8 zoom lens, 100-400mm zoom lens, 1.4x converter, 2x converter.
http://www.newforestobservatory.com/
http://www.flickr.com/photos/12801949@N02/

Hi Greg,

That certainly allows you to make best use of time at the telescope. As I'll be using a Kodak chip I won't be so fortunate but I have no choice in the matter as Sony doesn't make anything suitably large for the TEC 140 with its 980mm focal length. I've had confirmation from Yuri at TEC that he's received my deposit so now I just have to be patient until the fall. In the meantime I can get the pier built and check out the mounting.

Bob.

_________________
OM-D E-M5 + ED 12-50mm, H-PS14042E, H-F007014E, L-RS014150E, M.ZUIKO DIGITAL 45mm 1:1.8.
Gitzo GT1541T, Arca-Swiss Z1 DP ball-head.
Astrophotography: TEC 140 'scope, FLI ML16803 camera, ASA DDM60 Pro mount.