I thought this was a very informative post by Roland Christen of AstroPhysics fame:
Reflector - refractor, what is the difference?
Under real good seeing of course there is very little difference between reflector and refractor of same aperture size. Such seeing as we get in Chile at Las Campanas which is usually under 0.5 arc seconds. However, here in the middle of the country, we often have the jet steam above, so seeing is always compromised.
It is somewhat complicated, but I will try to make it very simple as a first order approximation. In a refractor the theoretical strength of the Airy Disc is 84% of the energy, with 16% in the diffraction rings surrounding the central Airy Disc. The first diffraction ring is the brightest, which makes the diameter of a star of long exposure to be approximately 2 times the theoretical resolution. When the image wiggles due to atmospheric instability it "paints" a star diameter somewhat larger than 2x the theoretical resolution, depending how much the star wiggles. Let's say that a 7" aperture can resolve 4.5/7 = 0.64 arc sec, so under really perfect seeing you might get 1.3 arc sec star disc, with maybe 0.9 arc sec FWHM value. The shape of the star brightness is a Gaussian distribution with a base diameter of 1.3 arc sec and a peak diameter of 0.64 arc sec. This would be the absolute limit for that aperture.
In a reflector, you always have some central obstructions. Large obstruction occur in fast instruments like the R-H F3.8 astrograph exceeding 50% by diameter, and perhaps 40% for an F8 Astrograph. The obstructions are necessary in order for the system to cover a wide field. You get similar obstructions in fast Newtonian astrographs, something that is unavoidable if you want to cover a large chip. So let's regard a 40% obstruction mirror system. The immediate effect is that the central Airy disc drops to around 60% of the energy with 40% remaining going to the surrounding diffraction rings. In fact, the second ring is now brighter than the first and has considerable energy. Thus the star image that gets "painted" in a time exposure is now some 3 times larger than the theoretical resolution limit versus 2 times for an unobstructed aperture. There is also considerable energy further out, so in terms of raw resolution you may see even in perfect seeing somewhat larger star sizes for the same aperture. When the seeing is not really good, all these rings begin to paint their own diameters and you get poor resolution.
Of course reflectors generally are larger in size, so the actual arc second resolution may still be larger than a much smaller refractor, especially when the seeing is excellent. In our case, the 12" F8 astrograph down in Chile experiences seeing of less than 0.5 arc seconds, and the best images I have recorded in tests have been on the order of 0.9 arc sec FWHM for a 10 second exposure. It might have been better but we are limited by the 9 micron pixel size of that setup.
Rolando
I think this is also why we see some star bloat with bright stars with some CCDs. Small wells mean the outer airy discs can saturate quickly given how much of the energy in some scopes goes to them. Its not a matter of pixel bleed as CCDs are usually anti blooming but merely the outer airy discs saturating the small pixels charge carrying capacity.
Interesting read Greg, thank you for sharing. I am still however uncertain about what all of that means in practice for those of us imaging with less than perfect seeing. Have I understood correctly, that for less than perfect seeing refractors could perform better for the given aperture, but since people own reflectors that are usually significantly larger in aperture, the difference is kind of nulled or could even be reversed?
As for star bloating, wouldn't small wells also mean smaller pixels leading to lower sensitivity? Perhaps I simplify things a bit but I think that stretching images makes pixels around bright stars brighter anyway thus possibly leading to bloated stars?
Having said that, I think what you are saying about stars bloating is very true when comparing images acquired with cameras with varying well depths but at the same resolution in arcseconds per pixel.
hi Greg.
In Australian conditions, the star shape and size for all but smaller refractors is determined almost entirely by the atmospheric seeing. With anything above about 6 inches aperture, you will never get anywhere near the optics diffraction pattern. ie the stars from my 250f4 and your CDK17 will be exactly the same shape and size as those from the AAT in typical conditions. It woud be nice to have seeing of 0.5 arcsec, but I tend to jump up and down if the seeing gets below 2 arcsec and I think that even the best Australian sites rarely get down to 1 arcsec.
I am not altogether sure what you mean by bloat, but if you mean that the stars look bigger, the two main reasons for this are: 1. that the pixels are small (eg the stars from your Trius will be 4x as big as those from your 16803) 2. you may be stretching the image to get to deep features (you see more of the star skirts as you go deeper - ie they look bigger).
If you are managing to saturate the stars excessively, then the simple answer is to expose for shorter subs - not just a bit shorter, but a lot. The optimum sub exposure varies almost with the square of the read noise - at the same aperture and pixel scale, you only need subs about 1/4 as long with your Trius as with your 16803. There is a tendency to overexpose with the new cameras, because that is how we had to do it with the older Kodak ones - those cameras have relatively high read noise, so you need long subs to overcome it - and then of course you need deep wells to handle the extra signal you get from a long sub. If appropriate subs are chosen, the results from both classes of camera should be identical.
Edit: as Slawomir said - I posted before reading his summary
Interesting read Greg, thank you for sharing. I am still however uncertain about what all of that means in practice for those of us imaging with less than perfect seeing. Have I understood correctly, that for less than perfect seeing refractors could perform better for the given aperture, but since people own reflectors that are usually significantly larger in aperture, the difference is kind of nulled or could even be reversed?
Yes that's how I read it. But the point of the article was in response to Roland making an earlier comment about how the refractor cuts through the seeing better. Most refractors are 175mm (in this case this was the refractor being discussed) or less and most reflectors are 200mm or larger.
As for star bloating, wouldn't small wells also mean smaller pixels leading to lower sensitivity? Perhaps I simplify things a bit but I think that stretching images makes pixels around bright stars brighter anyway thus possibly leading to bloated stars?
The Sony chips are higher sensitivity than the Kodak ones. Lower sensitivity due to small pixels is dependent on the scope aperture and focal length. If you go below .66 arc secs/pixel you could be said to be oversampling and lowering overall sensitivity by spreading the collected light over too many pixels.
Having said that, I think what you are saying about stars bloating is very true when comparing images acquired with cameras with varying well depths but at the same resolution in arcseconds per pixel.
Yes as a general rule I notice refractor images tending to have smaller star sizes (with exceptions) than reflectors although I am often amazed at how small stars look in some Newtonian images. Especially the ASA type scopes. I take it that is more from the excellent corrector they use more so than the Newtonian scope.
The bloated stars I am referring to are what Roland is saying here. With a reflector perhaps 50% of the energy is in the outer airy disc not the central bright dot of the star. Now if you have small wells and high QE like the KAF8300 and the Sony ICX694,814, 894 sensors then those outer airy discs are going to overexpose quicker than with a refractor where less of the energy is in the outer airy discs.
Quote:
Originally Posted by Shiraz
hi Greg.
In Australian conditions, the star shape and size for all but smaller refractors is determined almost entirely by the atmospheric seeing. With anything above about 6 inches aperture, you will never get anywhere near the optics diffraction pattern. ie the stars from my 250f4 and your CDK17 will be exactly the same shape and size as those from the AAT in typical conditions. It woud be nice to have seeing of 0.5 arcsec, but I tend to jump up and down if the seeing gets below 2 arcsec and I think that even the best Australian sites rarely get down to 1 arcsec.
I am not altogether sure what you mean by bloat, but if you mean that the stars look bigger, the two main reasons for this are: 1. that the pixels are small (eg the stars from your Trius will be 4x as big as those from your 16803) 2. you may be stretching the image to get to deep features (you see more of the star skirts as you go deeper - ie they look bigger).
If you are managing to saturate the stars excessively, then the simple answer is to expose for shorter subs - not just a bit shorter, but a lot. The optimum sub exposure varies almost with the square of the read noise - at the same aperture and pixel scale, you only need subs about 1/4 as long with your Trius as with your 16803. There is a tendency to overexpose with the new cameras, because that is how we had to do it with the older Kodak ones - those cameras have relatively high read noise, so you need long subs to overcome it - and then of course you need deep wells to handle the extra signal you get from a long sub. If appropriate subs are chosen, the results from both classes of camera should be identical.
Edit: as Slawomir said - I posted before reading his summary
Yes Ray you may be right there. I need to experiment with exposure length more. I am now settled on just using the 2 scopes - the AP RHA and the CDK with or without the .66X reducer. I see little difference between a 5 minute sub and a 10 minute sub (usually tighter stars with 5 minute subs due to less tracking errors) in terms of brightness, noise and detail. A bit more in the 10 min. I have been using 5 minute subs with the RHA and 10-20 minute subs for narrowband on the RHA (usually 10).
For example I imaged Corona Australis 5 minute subs and I noticed the main stars were overexposed. 5 minutes. Wow.
I am also thinking maybe 2 sets of subs like you do for M42. One for the bright areas and one for the general field and combine them in layers using Photoshop.
On the CDK with reducer 5 minute subs are usual and sometimes 10minute ones for fainter objects. That at 17 inches aperture and F4.4 with 77% QE so its pretty fast.
Interesting read Greg, thank you for sharing. I am still however uncertain about what all of that means in practice for those of us imaging with less than perfect seeing. Have I understood correctly, that for less than perfect seeing refractors could perform better for the given aperture, but since people own reflectors that are usually significantly larger in aperture, the difference is kind of nulled or could even be reversed?
As for star bloating, wouldn't small wells also mean smaller pixels leading to lower sensitivity? Perhaps I simplify things a bit but I think that stretching images makes pixels around bright stars brighter anyway thus possibly leading to bloated stars?
Having said that, I think what you are saying about stars bloating is very true when comparing images acquired with cameras with varying well depths but at the same resolution in arcseconds per pixel.
....I think the important thing for optical quality in a refractor for example is whether the optical resolution is "diffraction limited" rather than limited by glass imperfections etc....
Diffraction also occurs in human vision. The diffraction spread of light at the back of the retina is of comparable size to the spacing of the rods and cones which sense the light - pretty handy isnt it? Millions of years of evolution can do this. (there are several dozen examples of "sight" evolving independently in different species - must be something important in sensing light in your environment one would think??)
Having used many scopes the few that were way up there in their Strehl ratio and that were way beyond diffraction limited way out performed lesser figured scopes.
Diffraction as a barrier is only part of the picture. Light scatter, throughput, the extent of the spectrum that is in focus at the same point, anti reflection coatings reducing ghosting, increasing micro contrast etc all play a part. Even focusing a high end instrument is a lot easier as there is more of a snap to focus effect. Star sizes off axis, correction for the 6 optical aberrations all play a large part in making a high quality image.
I think diffraction limited has been pushed so hard so as to justify not being willing to or able to go the extra mile so they make out that there really is no need anyway. Its a marketing justification for an inferior optical system.
Greg.
I was quite disappointed when I looked through a telescope for the first time. Eons of evolution and only a few percent of Quantum Efficiency? That's one of the main reasons why so many of us in here utilise silicon retinas - CCD and/or CMOS, for converting light into electrical impulses and then the signal gets amplified by a computer screen so our limited rods and cones can detect it, ultimately allowing us to more fully admire the heavens
I see this first hand, very obvious and pronounced disc structure.... I sample at 0.44" with a 12.5" @f/8. When I run focus routines I can see the airy disc itself with the bright core where most of the energy resides and the outer rings. Sampling at this level tends to really clearly show issues such as tube currents etc. With smaller wells, the washy airy disc can quickly register a sizeable portion of the available dynamic range. It seems to further amplify the effects of the CCD micro-lens. The washy disc causes a "shimmer" over the lensing, very tiny diffraction spikes which further spread from the disc structure. - I have watched this live at 100+ fps using another sensor with similar sized pixels to my main 8300 sensor.
There are some cool advantages - I have revealed some pretty tiny structures within some targets but bright point sources of light tend to suffer greatly, i.e stars themselves and I am yet to find a way to suppress it. You can do short subs, but other parts of the target will suffer.
Oversampling excessively also results in significant loss of sensitivity and any forms of aberration present, either optical or environmental become extremely obvious.
Shorter sub composition for High dynamic range targets:
Core of the Sombrero (note the strange diffraction spikes.... thats a @!#$ spider building a web in my OTA that i didn't discover until the following morning)
It's a nice thing to have data present that would otherwise not be seen when sampling less, but I find 80% of my time is dealing with the adverse effects. Bigger camera should be here soon so I'll be looking forward to less fuzzballs...
When considering star size in photography -- especially long-exposure photography -- it is important to account for ultra-violet starlight, particularly between 300-400nm.
Light at these wavelengths is not absorbed by the Earth's atmosphere (in fact, the atmosphere does not completely absorb light between 280-300nm) and is more susceptible to atmospheric diffraction compared to light at longer wavelengths. This very fact alone will cause a star to become "bloated" over the course of a long exposure.
For most observers, this light is not visible to the human eye but is readily picked up by unfiltered monochrome CCDs and indeed color CCDs not equipped with a UV blocking window.
Choice of CCD sensor will almost certainly have an impact on star size when these wavelengths are not properly restricted. For example, a KAF16803 has an approximate QE between ~25% (350nm) and ~40% (400nm), whereas an ICX694 is much more sensitive to UV light, with sensitivity ranging from ~30% (350nm) to ~65% (at 400nm).
Having looked at many such graphs, both independent and vendor-supplied, I have found that QE sensitivity tests done by independent 3rd parties tend to give a much more realistic "real world conditions" assessment of CCD performance. A lot of this has to do with the fact that many camera vendors simply provide the default QE graph supplied by the CCD sensor manufacturer and if you read the fine-print which accompanies these sensor manufacturer QE graphs you will find that the test conditions under which they were generated not only vary between sensors, but also deviate wildly from the typical astrophotographical conditions in which these sensors are often used. Sensor manufacturer QE graphs also fail to account for other factors such as the coatings found on refractor lenses and camera windows.
This is where the CCD sensor comparisons provided by Point Grey are particularly reliable -- they use industry standard test conditions which are consistent right across all test articles and give you an accurate assessment of the performance of an actual camera, not just the sensor.
If star size is of concern in a photographic setting, I would definitely suggest at least investigating UV suppression. I have observed a noticeable decrease in star bloat after switching from a Baader filter with relatively low UV suppression to an Astronomik filter with high UV suppression. This is with an OSC camera with relatively poor UV blocking at the sensor window. Obviously in a narrowband setting the luminance channel would be the culprit.
But isn't this just an illusion? That the scope lacks the resolving power to actually see the disturbances in the atmosphere?
Don't get me wrong, I refractors...
In order to answer this properly, I think some evaluation of the optical coatings used on the refractor in question would be required. The number of elements used in the optical path (also resulting in more coatings between the light source and your eyeball) would have some impact also. Obviously there's a lot more going on there compared to having light reflect off a couple of mirrors.
But isn't this just an illusion? That the scope lacks the resolving power to actually see the disturbances in the atmosphere?
Don't get me wrong, I refractors...
\
I know what you are saying that a long focal length scope will show the disturbances better. But as the article from Roland shows, refractors have tighter stars. Also better contrast as the secondary mirror robs contrast.
So tighter stars and better contrast = the better view albeit at a wider field look.
Also less effect of thermal currents and boundary layers.
Though refractors and reflectors both have thermal issues, those of the refractor tend to be mild and short lived because the objective lens is exposed to the night air. The primary mirror of a Newtonian reflector, however, is shielded by a long tube that inhibits cooldown.