Best Gear for small DSOs

[Tommohawk]

Hi All

I'm still very much in the early stages of learning the DSO imaging curve, but I think my main interest will be detail in smaller DSOs. Currently I've only used my SW200P with Canon 550D and have got some very pleasing results.

So the question is what next? Ideally I would go to CCD/Astro CMOS - I'm looking at maybe the ASI1600 when it comes out - but for now probably stuck with the Canon. On a good day my tracking/guiding is pretty good so I think I can get away with a longer EFL.

Which would be better - a good x2 or x3 barlow, or an SCT with reducer? Do I need a larger aperture to achieve good resolution? 

I do realise this might re-ignite the discussion about F values and exposure times but hey, its a cloudy night.

Thanks for any thoughts

[johnrt]

There is much to consider when thinking about a DSO set up, you need to bear in mind that adding a x2 or x3 barlow is going to also have that same multiplication effect on the f/ratio of the rig and therefore lead to some pretty long exposure times, which is why they are not generally used in DSO imaging. An f/6 scope being turned in to an f/18 scope will be an impossible and frustrating set up for imaging dso's.

You are going to be looking at a decent aperture to allow a manageable exposure time, remember the longer smaller the dso's you're after the longer the focal length required and the more demands that places on your guiding.

You also need to consider your pixel scale that the ota and camera combination will produce. Anything much below 1"/pixel is going to be beyond the seeing we can normally achieve here in the UK.

Are you planning to keep the HEQ5 for this rig too?

I would have a look at this useful tool and look at some fov's that different set ups will provide http://www.12dstring.me.uk/fovcalc.php it will also report the f/ratio and pixel scale.

[ollypenrice]

9 hours ago, johnrt said:

There is much to consider when thinking about a DSO set up, you need to bear in mind that adding a x2 or x3 barlow is going to also have that same multiplication effect on the f/ratio of the rig and therefore lead to some pretty long exposure times, which is why they are not generally used in DSO imaging. An f/6 scope being turned in to an f/18 scope will be an impossible and frustrating set up for imaging dso's.

You are going to be looking at a decent aperture to allow a manageable exposure time, remember the longer smaller the dso's you're after the longer the focal length required and the more demands that places on your guiding.

You also need to consider your pixel scale that the ota and camera combination will produce. Anything much below 1"/pixel is going to be beyond the seeing we can normally achieve here in the UK.

Are you planning to keep the HEQ5 for this rig too?

I would have a look at this useful tool and look at some fov's that different set ups will provide http://www.12dstring.me.uk/fovcalc.php it will also report the f/ratio and pixel scale.

Agreed. (By the way, John, you always strike me as the most affable chap so I'm surpirised to see the infamous Fritz Zwicky as your avatar! Apparently one of his colleagues suggested that the 'Zwicky' become the standard unit of abrasiveness. It was then objected that no known substance could ever be measured in a unit larger than the MicroZwicky...)

Back on track I'd have thought you needed to consider the following:

Pixel scale. I'd be thinking about 1 to 2.5"PP. Your average guiding error needs to be about half the pixel scale to be on the safe side.

Field of view. This can be modelled in various planetaria.

To avoid the F ratio discussion, once you have the FOV pinned you simply need as much aperture as possible. Some small DSOs are bright but most are not so plenty of aperture is what you really need.

Olly

[johnrt]

1 hour ago, ollypenrice said:

Agreed. (By the way, John, you always strike me as the most affable chap so I'm surpirised to see the infamous Fritz Zwicky as your avatar! Apparently one of his colleagues suggested that the 'Zwicky' become the standard unit of abrasiveness. It was then objected that no known substance could ever be measured in a unit larger than the MicroZwicky...)

Back on track I'd have thought you needed to consider the following:

Pixel scale. I'd be thinking about 1 to 2.5"PP. Your average guiding error needs to be about half the pixel scale to be on the safe side.

Field of view. This can be modelled in various planetaria.

To avoid the F ratio discussion, once you have the FOV pinned you simply need as much aperture as possible. Some small DSOs are bright but most are not so plenty of aperture is what you really need.

Olly

I recently read about Mr Zwicky and his dunkle Materie, and of course his personality was described along with his spherical quote. It made me laugh so much I had to have him as my avatar pic. You can rest assured I'm about the least Zwicky type of character you could meet, and don't worry I'm not planning to turn over a new leaf to be more like him

However, I do have the unusual enjoyment of working with two Zwickys, so abrasive they could fall out with themselves if you left them alone for half an hour, and in over 3 years I think I'm the only person to never have a cross word with either of them!

 

Back on topic, how about a nice 8' RC, a carbon one would keep the weight down if your keeping the HEQ5.

 

John

 

[DaveS]

Or possibly an 8" Edge HD and 0.7 reducer? 1400-ish mm FL.

[Tommohawk]

Hi all and ta for thoughts - very helpful. Working today so slow to respond - sorry! 

As usual I have probably muddled a few issues together. Let me ask a few questions if i might:

1. If I have an F5 scope, and am imaging a small target - say only 10% of chip - if I use a x2 barlow, will I need longer exposure and if so by how much? It seems to me that the increase in image size will spread the light over more pixels, so one would think this must need longer exposure. 

2. What would be better - an F5 newt like the SW200P with a x2 barlow (and CC), or an F10 200mm RC? Would the losses in the barlow exceed the losses from the larger CO in the RC?

3. If I'm only interested in detail on small DSOs, I would probably want an EFL of around 2500mm or even more. Leaving aside guiding issues  (but thanks for the comments above on that - very useful) what the best way about this? Having a larger aperture will obviously reduce the F ratio and minimise the exposure time, but what aperture would I need to ensure I don't lose resolution? I suspect the answer to that lies in the information already above but I'm afraid I cant extract it! 

4. Which would give better results for DSOs - an 8" RC with EFL 1624mm (like the TS optics) or a (non-edge type)  9.25 SCT with a F6.3 corrector, EFL ~1500mm? It happens these two scopes are similar prices, and the SCT would double as a planetary scope.

Sorry thats a lot a questions!

[johnrt]

Example 1  You are adding a x2 barlow to your ota, which effectively doubles the focal length of the scope, while the aperture remains the same. This changes the f/ratio from f/5 to f/10, a change of 2 full f stops, so to achieve the same amount of light capture per pixel you will need to expose the image for 4 times as long. The equivalent of 5 minutes exposure becomes a 20 minute one with the barlow.

Example 2 - On paper they would be the same, but I would expect the RC to produce superior results and illuminate a DSLR chip much more evenly in reality.

3 - for a focal length of 2500mm you would be wanting to aim for something at around f/8 at the very slowest for DSO imaging, it is possible to image with slower OTA's, but for someone starting out I wouldn't be recommending it at all, especially not in the UK. For a 2.5 meter focal length scope to get close to an f/8 ratio you're looking at a 300mm (12 inch) aperture scope. That same 12 inch set up will give you a 0.35"/pixel sampling rate with your 550D, well below the possible seeing here in the UK. I would not want to put something that big on an HEQ5 either!

4 - I'm an RC man so I'd pick the RC all day long But even that will give you a sampling rate of around 0.55"/pixel with your DSLR, so you are way over sampled still and wasting all that focal length, I'd be looking to reduce it down and speed it up with a 0.67 reducer.

[Tommohawk]

John - thanks for a very detailed and orderly answer - most helpful. I need to spend some quality time with the whole under/over sampling "/pixel thing!

 

[Tommohawk]

Right, I've done some homework and am slowly getting to grips with this.... I think!

A couple of thoughts.

Given that 1" per pixel is about the going rate, whats the point in, say, a 12" RC  with an EFL of 2432mm, when this would imply a pixel size of about 12 microns?

Presumably, if using such a chip, the pixel output would be beefier maybe?

In reality if using a 12" RC with eg 5 micron pixels, would mean somewhat (undersampled*) oversampled. I can't help wonder if this gives some real advantage, because the turbulent and variable nature of the atmospheric disturbance means that there are moments of much greater possible resolution and also interpolative effects as the image shifts transversely.

Another oddity it that when considering a DSLR used at 50mm, the ideal pixel size would be around 0.25 microns. In reality that means massive (oversampling*) undersampling - so what's all that about, as they say?

Any thoughts?

PS. Ahhh... just engaged brain and realised my comments above re moments of clear seeing and image shift wouldnt really apply to long exposures - only for video. And that explains the 0.25" per pixel recommendation (on Starizona) for planetary work. Another piece of the puzzle!

[derrickf]

Tom,

at 1 arc-second/pixel you will be able to critically sample stars with a FWHM of 3.3 arc-seconds - see Stan Moore et al for the maths behind this. IMHO 1 second/pixel is undersampling typical UK seeing - typical FWHM at my unremarkable) site are 2.2arcseconds FWHM with the best nights at around 1.8 arcseconds.

You can see an image here (if you click and view it full size you will see the full native resolution image) taken with a 12" RC at F/8 and Atik 383 - plate-scale 0.46 seconds/pixel. This was on a night of typical seeing and the stars in the composite image have an FWHM of 2.3 arc-seconds (individual frames are a little lower but the integration process always results in a slight blurring due to the necessary re-sampling during the registration process).

At 0.46 seconds/pixel I am moderately oversampled and I plan to try a reducer to increase the plate scale to 0.6 seconds/pixel for comparison. One advantage of slight oversampling is that images are less susceptible to artifact generation with aggressive processing techniques like iterative deconvolution (although this does need high SNR data)

I cannot over-emphasise the need for an excellent mount and flexure free imaging train when imaging at high resolution.

 

HTH

 

Derrick

[Tommohawk]

Thanks Derrick - yes agreed mount needs to be up to it, and my HEQ5 is pretty much on the limit with guidescope as it stands. 

I get good guiding generally but pretty horrible close to equator. 

Also your post drew my attention to an error in my previous post - thanks for that!

Nice M13 image too!

Any thoughts on the DSLR 50mm lens undersampling issue?

[derrickf]

Hi Tom,

A 50mm lens on a DSLR does undersample stars but, typically, imagers using this configuration (and others using short focal length refractors etc)  are more interested in a wide FOV to enable them to show large scale diffuse structures so the undersampling is not an issue. So for example, an image centred on M13 with a DSLR/50mm setup would show M13 as a small object surrounded by lots of open space and the odd background galaxy but no amount of processing will enable you to discern individual stars in the central part of M13. Dawes limit defines the maximum theoretical  point source resolution you can achieve but you can detect smaller features than the Dawes Limit in diffuse objects.

Basically, as others have said, you choose your FOV and platescale to show what you want then try to find an optical setup to achieve it.

Good luck getting your head around this - just when you think you've understood it based on one set of parameters you'll find some other variables that seem to contradict what you think you had just learned. It doesn't help that people, often unthinkingly, trot out erroneous information they have heard third hand as absolute truth ... 2 arcseconds per pixel as the ideal platescale, F-Ratio myth etc. etc.

 

Derrick

[Tommohawk]

4 hours ago, derrickf said:

 

Good luck getting your head around this - just when you think you've understood it based on one set of parameters you'll find some other variables that seem to contradict what you think you had just learned. It doesn't help that people, often unthinkingly, trot out erroneous information they have heard third hand as absolute truth ... 2 arcseconds per pixel as the ideal platescale, F-Ratio myth etc. etc.

 

Derrick

You're so right! I was even reading a thread yesterday which was questioning what the F ratio myth actually is... the F ratio myth myth maybe!

[ollypenrice]

On 22/4/2016 at 15:05, Tommohawk said:

Hi all and ta for thoughts - very helpful. Working today so slow to respond - sorry! 

 

1. If I have an F5 scope, and am imaging a small target - say only 10% of chip - if I use a x2 barlow, will I need longer exposure and if so by how much? It seems to me that the increase in image size will spread the light over more pixels, so one would think this must need longer exposure. 

 

F ratio myth territory. Hold onto your hats.

1) With or without Barlow you have exactly the same number of photons from the object hitting your chip. (Well, OK, the Barlow will lose a few but hey-ho.)

2) How many pixels do they land on? (More from the long FL. But, if you are above the read noise does it matter?) Because...

3) You are not going to look at the picture on your camera, you are going to process it and look at it on your PC. If you resample the barlowed image downwards, or the un-barlowed image upwards, to give the same object image size on screen in the final image, how much difference will there be? Not enough to get excited about. The difference will bear absolutely no relation whatever to the F rule of exposure time going as the square of the F ratio.

Olly

[michael.h.f.wilkinson]

34 minutes ago, ollypenrice said:

F ratio myth territory. Hold onto your hats.

1) With or without Barlow you have exactly the same number of photons from the object hitting your chip. (Well, OK, the Barlow will lose a few but hey-ho.)

2) How many pixels do they land on? (More from the long FL. But, if you are above the read noise does it matter?) Because...

3) You are not going to look at the picture on your camera, you are going to process it and look at it on your PC. If you resample the barlowed image downwards, or the un-barlowed image upwards, to give the same object image size on screen in the final image, how much difference will there be? Not enough to get excited about. The difference will bear absolutely no relation whatever to the F rule of exposure time going as the square of the F ratio.

Olly

Unless the read and dark noise are large enough to cause trouble. The signal is spread over more pixels, and more pixels means more thermal electrons. Darks can only correct for the mean dark noise, not the variance.

[Astroboy239]

Small dso we use rasa astrograph and a c14 with Barlow 

 

Varad

[PhotoGav]

This is an interesting thread and I hope it is helping you get closer to some decisions. I thought I would add a few points from my real world experience of an HEQ5 mount and a long focal length scope. I use an EdgeHD 800.

Exposure times - the scope is f10 unreduced and it does require long exposure times, 1200s Luminance seems to be my standard for galaxies and 1800s for narrowband. This is perfectly achievable, but does require great guiding, which I can only achieve with an off axis guider. So, I highly recommend that you factor in an OAG to your designs and appreciate that, for half decent results, you will need to allow plenty of imaging time for a single target.

Focal length - 2032mm unreduced, 1422mm reduced. Superb! I would love a refractor that gives me that focal length, but am yet to find one that I can afford or can use on an HEQ5.

Would I recommend the EdgeHD 800? Yes and no! It is a lovely scope and is certainly the best value long focal length scope I have found so far. It also neatly doubles up extremely well for planetary imaging, especially with a 2.5x Powermate and fast frame rate planetary camera. As for DSO work - it performs pretty well, but so far, my images have suffered from a slight softness and don't have that refractor cutting sharpness. This might be a problem with the way I am using the scope and is simply poor focusing - I am working on this and hope to find improvements. I chose the scope over an RC as I wanted to avoid diffraction spikes and that it certainly does!

I am yet to fully comprehend over / under sampling, so if anyone cares to explain in simple terms that would be extremely helpful!

Good luck in your search and I think you should definitely drop the idea of a Barlow for DSO imaging!

[Tommohawk]

OK, got a busy day today so will get back later. Got some more questions!!

[theastrodragon]

Rather than a Barlow, why not go for a ccd with small pixels that you can normally use binned. Then rather than use a barlow, don't bin.

In my case, my SX814 would give me 0.67"/pixel, which is way too small for normal UK seeing, so I use it binned at 1.3"/pixel for most things.

The only time I use it unbinned is for planetaries, where they are small and usually high surface brightness, and look rather tiny when binned.

 

[ollypenrice]

19 hours ago, michael.h.f.wilkinson said:

Unless the read and dark noise are large enough to cause trouble. The signal is spread over more pixels, and more pixels means more thermal electrons. Darks can only correct for the mean dark noise, not the variance.

Agreed.

Olly

[Tommohawk]

Right, back on the case. First thanks for all the useful comments - I suspect there will be others out there that will find this useful too.

Couple of quick things. As I said probably my main interest will be detail in smaller DSOs. I dont intend to print any images, so dont need massive resolution - images will be viewed on my laptop, or maybe a bigger HD type screen (1920x1080 pixels I think). So with my Canon 550D DSLR, (5184 x 3456 pixels) or the ASI1600 (4656x3520) I can crop the image and just use the centre - much as I do now for many images. That effectively pretty much triples the image scale with no loss of resolution. Please correct me if this isnt so.

Also, if I go to the ASI1600, the pixels are 3.8 microns rather than 4.3 for the Canon - so the screen image will effectively be about 11% larger. (And the field correspondingly smaller of course)

So, moving on...

On 4/23/2016 at 19:29, ollypenrice said:

F ratio myth territory. Hold onto your hats.

1) With or without Barlow you have exactly the same number of photons from the object hitting your chip. (Well, OK, the Barlow will lose a few but hey-ho.)

2) How many pixels do they land on? (More from the long FL. But, if you are above the read noise does it matter?) Because...

3) You are not going to look at the picture on your camera, you are going to process it and look at it on your PC. If you resample the barlowed image downwards, or the un-barlowed image upwards, to give the same object image size on screen in the final image, how much difference will there be? Not enough to get excited about. The difference will bear absolutely no relation whatever to the F rule of exposure time going as the square of the F ratio.

Olly

I'm not sure I quite get this!

Are you saying that x2 barlowing an image, and then displaying it smaller - the same size as the unbarlowed image - would give exactly the same appearance? 

That sounds logical - presumably the "brightness per sq. cm" (or whatever the correct term is) would be the same in either case. But if I'm using a barlow specifically to make the image larger (without losing resolution) I guess I wouldnt then be displaying it smaller?

Put another way, the barlowed image, if using the same exposure as the unbarlowed image, would only have a quarter the "brightness per sq. cm" - or maybe not?

Michael - I take the point about read noise.

Gav - thanks for you real world experience. I'm sort of leaning toward a C9.25 (non-EDGEthough ) with reducer so interested in your pints about image sofness.

Re undersampling etc I found some good info here.

Astrodragon - yes, thats sounds good. The C9.25 with reducer would give 0.53"/pixel unbinned or 1.06 if x2 binned, and would give 0.33"/pixel when not reduced for planetary/lunar work, or 0.16"/pixel if x2 barlowed

 

[Chris Murray]

im at the same point, looking to get closer and get more detail, i also wanted to try rbg, narrow band, so i just got qhy,s minicam,f  tiny field of view and over sampling, with my 10in quattro,  but to get ideal sampling i would need a reducer so brings me full circle,   clouds and moon are not helping getting any results, cheers 

[ollypenrice]

8 hours ago, Tommohawk said:

Right, back on the case. First thanks for all the useful comments - I suspect there will be others out there that will find this useful too.

Couple of quick things. As I said probably my main interest will be detail in smaller DSOs. I dont intend to print any images, so dont need massive resolution - images will be viewed on my laptop, or maybe a bigger HD type screen (1920x1080 pixels I think). So with my Canon 550D DSLR, (5184 x 3456 pixels) or the ASI1600 (4656x3520) I can crop the image and just use the centre - much as I do now for many images. That effectively pretty much triples the image scale with no loss of resolution. Please correct me if this isnt so.

Also, if I go to the ASI1600, the pixels are 3.8 microns rather than 4.3 for the Canon - so the screen image will effectively be about 11% larger. (And the field correspondingly smaller of course)

So, moving on...

I'm not sure I quite get this!

Are you saying that x2 barlowing an image, and then displaying it smaller - the same size as the unbarlowed image - would give exactly the same appearance? 

That sounds logical - presumably the "brightness per sq. cm" (or whatever the correct term is) would be the same in either case. But if I'm using a barlow specifically to make the image larger (without losing resolution) I guess I wouldnt then be displaying it smaller?

Put another way, the barlowed image, if using the same exposure as the unbarlowed image, would only have a quarter the "brightness per sq. cm" - or maybe not?

Michael - I take the point about read noise.

Gav - thanks for you real world experience. I'm sort of leaning toward a C9.25 (non-EDGEthough ) with reducer so interested in your pints about image sofness.

Re undersampling etc I found some good info here.

Astrodragon - yes, thats sounds good. The C9.25 with reducer would give 0.53"/pixel unbinned or 1.06 if x2 binned, and would give 0.33"/pixel when not reduced for planetary/lunar work.

 

If you want more detail by using the barlow then you'll need to increase the exp time in accordance with the F ratio rule, so 4x the exp time for a 2x Barlow. You are, after all, asking the system to give you more information.

Olly

[Tommohawk]

1 hour ago, ollypenrice said:

If you want more detail by using the barlow then you'll need to increase the exp time in accordance with the F ratio rule, so 4x the exp time for a 2x Barlow. You are, after all, asking the system to give you more information.

Olly

Yes that makes perfect sense - I thought you were suggesting otherwise in your previous post. In practice I probably wont use a barlow - maybe just the C9.25 with reducer.

BTW Olly, can I ask what kit would you use to image smaller DSOs? Assume a CCD/astro CMOS like the ASI1600 and a budget of say approx £1500 for the scope. Actually it would be interesting to know what you would use even without the budget constraint.* 

Also assume EQ6 mount.

[PhotoGav]

Thank you for the sampling link Tom. It is starting to make a lot more sense now and by all accounts oversampling is the right side to be on whereas undersampling is a definite no-no.

It seems as though the classic 9.25 is the current favourite in this race. I will watch with interest, especially for your first light image.

Good luck.