Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

EAA camera decision - who can help?

beginner EAA SCT equipment
  • Please log in to reply
40 replies to this topic

#26 GazingOli

GazingOli

    Sputnik

  • -----
  • topic starter
  • Posts: 32
  • Joined: 13 Oct 2019
  • Loc: Stuttgart GERMANY

Posted 03 May 2020 - 12:35 AM

GrazingOli - with your new purchase you are right where I am - an F6 refractor (mine is 105/650), and here is my dilemma.  Using round numbers for seeing of 3: star size= 3*600/200= approx 9 microm, and we want pixel size to be 1/3 star size = 3 microm. The ASI294 camera doesn't meet this criteria (4.6microm).  The ASI 183 does at 2.4 and the 290 does at 2.9.  If seeing improves the pixel size drops, so I am thinking for 600mm fl we need as small a pixel size as possible (offered).

We are not exactly in the same situation

 

I am a bloody beginner, not quite sure about your math AND I also got a CPC800. The ASI294 was acutally recommended as a follow up for my ASI120 as causing not so much oversampling with the SCT. If I do not care about seeing conditions, using the following formula

 

Image Scale = Pixel Size / Focal Length * 206 (Ideal Range between 1 to 2) it looks not too bad for the BW100 scope

 

BW100 /w ASI120 fl600 -> 1.29 quite good
C8 f6.3 /w ASI120 fl1280 -> 0.60 oversampling and very small field

 

BW100  /w ASI294 fl600 -> 1.59 really good
C8 f6.3  /w ASI294 fl1280 0.75 -> better than ASI120 and reasonable field

 

Your math does not really make so much sense to me, because under seeing of 3 you take the whole thing times 3 and then you divide it by 3 at the end, which means that the result is the same with or without seeing. However I do not know where your math comes from so I cannot really validate it.

 

Thoughts:

What I saw, reading through the boards, is, that a lot of people seem to do quite well with oversampling, especially when using SCT or even MAK scopes with CMOS cameras. However small pixels with high resolution also means a lot of data to be handled by the hardware. So I would not want to go too high in order to use my Core i5 notebook for a while. And also it seems to be a bit of a trial and error affaire which camera works well with which scope under certain seeing conditions. Last but not least: there is not one single setup for all celestrial objects, as you said already. Planets and Moon will require a different setup than DSO.

 

My targets will be DSO, mainly galaxies and planetary nebulae. See what I can achieve with the ASI120, mainly with the f6 scope.


Edited by GazingOli, 03 May 2020 - 12:40 AM.


#27 GroupJ

GroupJ

    Sputnik

  • -----
  • Posts: 46
  • Joined: 02 Apr 2020

Posted 03 May 2020 - 04:02 AM

GazingOli - I'm not sure of my math either it comes from Astrophotography by Rob Kantelburg.  That reference says you want 3 to 4 pixels covering your telescopes star image.  I think it is the same equation you have with a scale = 1.  IF ideal image scale =1 to 2 for DSO, my conclusion is that undersampling is OK


Edited by GroupJ, 03 May 2020 - 04:05 AM.


#28 rnyboy

rnyboy

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 29 Oct 2019
  • Loc: Rochester, NY

Posted 03 May 2020 - 08:27 AM

Hi again GroupJ...

 

I think all three of the above images were 100x5s exposures saved as a fits file in SharpCap while livestacking.  Sensor gain was 450 on all the images and a dark frame was used during the livestacking.  (The below star fields where 3s exposures and dark field corrected)

 

As far as over/undersampling goes.  That depends on f-ratio and seeing conditions for the scope; my sensor spans a pretty useable range for sampling and at f/3.2 it's getting pushed away from optimal.  The best sampling for my setup and OK/good seeing conditions is in the f/10 to f/6.3 range for the scope.  Since I only have one camera it is what it is.  Everything is a compromise.  I traded speed for better sampling in this case.

 

No changes to scope, just a f/6.3 at the back of the scope and a cheap 0.5x in front of the sensor, so about f/3.2 in operation.  For some reason that I don't understand at the moment, when I plate solve an image I get a FOV pretty close to that expected for f/3.3 with my equipment (0.82ox0.46o), against a calculated FOV for my setup from astronomy.tools at 0.84ox0.48o, but the distances between each reducer and sensor face was too long based on reports I've read for the same combination.

 

The distances from the sensor face to the f/6.3 and 0.5x reducers in my setup were 105mm and 50mm, respectively.  These are very close to the separations for each reducer to sensor when used alone, not in combination, and from reports I've seen the distances for the f/6.3 and 0.5x to sensor face when in combination should be around 85mm and 35-ish mm, respectively.  Maybe the reason for the difference is that my scope is a 6SE and I think all the reports I saw were using an 8SE (I doubt if this is the issue here)?  Even the stars don't look particularly ridiculous near the corners of the full FOV images, so coma isn't too much of an issue but there is some vignetting showing up.  So I still have that bit of confusion to eventually sort out.

 

Here's a couple of full frame images of a star field that I took the other night...

 

(Center star (HD 126200, Magnitude ~6) was prefocused using SharpCap's focus assist at a faster exposure and lower gain than used for these images to greatly reduce the blow-out while focusing.  The vignetting isn't terribly apparent in these two images but it's relatively absent in the vertical dimension from the frame's center and becomes very apparent in the outer ~20% of the frame in the horizontal dimension.  Considering this setup has a cheap 0.5x reducer in the optics train I don't think the star distortions from coma are all that bad around the edges when looking at the other non-blown out stars around the image border.) 

 

  1st image with star in center of image...

 

1dlkrOm.png

 

2nd image of central star above in lower left corner of frame showing the worst case distortion from several images of this star in various locations of the frame.  Frankly, no corner appeared to be that much better or worse for distortion than this image...

 

tH6TLQ5.png


Edited by rnyboy, 03 May 2020 - 02:22 PM.


#29 Rickster

Rickster

    Apollo

  • *****
  • Posts: 1,177
  • Joined: 09 Jun 2008
  • Loc: NC Kansas Bortle 3 SQM 21.8+

Posted 03 May 2020 - 10:51 AM

GrazingOli - with your new purchase you are right where I am - an F6 refractor (mine is 105/650), and here is my dilemma.  Using round numbers for seeing of 3: star size= 3*600/200= approx 9 microm, and we want pixel size to be 1/3 star size = 3 microm. The ASI294 camera doesn't meet this criteria (4.6microm).  The ASI 183 does at 2.4 and the 290 does at 2.9.  If seeing improves the pixel size drops, so I am thinking for 600mm fl we need as small a pixel size as possible (offered).

 

I understand small pixel size means less sensitivity and for EAA I suppose that means longer sampling times for frames.  For me that would not be an issue as I'm not in a race - I'd rather wait a bit if I can resolve more detail in the assisted observation.

 

thoughts?

Seeing is measured in arc seconds not microns. 


  • DSO_Viewer and GazingOli like this

#30 Noah4x4

Noah4x4

    Gemini

  • *****
  • Posts: 3,430
  • Joined: 07 Apr 2016
  • Loc: Colchester UK

Posted 03 May 2020 - 02:00 PM

Ironically I found that guide to not be helpful. YMMV

I too found this confusing as it didn't reflect the current user interface. Sharpcap also offers different features for different makes of camera. This isn't a criticism of said abridged guide as I am sure it was brilliant when published. But Sharpcap has evolved further and it's become less relevant.

 

ASILive and Atiks Infinity/Dusk are far easier as they are dedicated to a specific camera manufacturer. But I can envisage folk using these to get started, and then graduating to the more challenging Sharpcap, SG Pro or Maxim DL later when they want to embrace more sophisticated stuff like Plate Solving. 



#31 rnyboy

rnyboy

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 29 Oct 2019
  • Loc: Rochester, NY

Posted 03 May 2020 - 03:11 PM

GroupJ...

 

You said:  "I understand small pixel size means less sensitivity and for EAA I suppose that means longer sampling times for frames.  For me that would not be an issue as I'm not in a race - I'd rather wait a bit if I can resolve more detail in the assisted observation."

 

You need to be careful in trading off speed for resolving detail.  The natural limit imposed by atmospheric affects is about 0.5 to 1 arcsec at the best.  Unless you plan on putting your scope in orbit there is a point where you don't gain any extra detail with more pixels and are in a permanent over-sampling condition.  The Reyleigh resolving limit for my 6SE is about 1 arcsec.  Apertures bigger than about 150mm (6 inches) don't gain you much of anything in extra detail if used here on earth.  As the aperture gets larger and larger they capture more and more photons and thus reduce the total exposure time required for the same quality of image with all other things besides aperture being equal.

 

Throw in other variables, like vibrations from whatever, clouds, how good the mount is, etc., you may regret going for more pixels and the reduced speed.


Edited by rnyboy, 03 May 2020 - 03:29 PM.

  • bsavoie likes this

#32 GroupJ

GroupJ

    Sputnik

  • -----
  • Posts: 46
  • Joined: 02 Apr 2020

Posted 03 May 2020 - 03:42 PM

rnyboy - I took a look at your photo album and I have to say your images look pretty good to me for this scope (C6E). Were all your DS images taken with the same setup? If I understand your setup (a photo of your rig would help) you essentially have two reducers in series to get to F3.3.  I am wondering if all that glass is smearing the image a bit?  It would be interesting to compare F6.2 with F3.3 for image quality.


Edited by GroupJ, 03 May 2020 - 03:43 PM.


#33 Rickster

Rickster

    Apollo

  • *****
  • Posts: 1,177
  • Joined: 09 Jun 2008
  • Loc: NC Kansas Bortle 3 SQM 21.8+

Posted 03 May 2020 - 03:44 PM

GroupJ...

 

You said:  "I understand small pixel size means less sensitivity and for EAA I suppose that means longer sampling times for frames.  For me that would not be an issue as I'm not in a race - I'd rather wait a bit if I can resolve more detail in the assisted observation."

 

You need to be careful in trading off speed for resolving detail.  The natural limit imposed by atmospheric affects is about 0.5 to 1 arcsec at the best.  Unless you plan on putting your scope in orbit there is a point where you don't gain any extra detail with more pixels and are in a permanent over-sampling condition.  The Reyleigh resolving limit for my 6SE is about 1 arcsec.  Apertures bigger than about 150mm (6 inches) don't gain you much of anything in extra details if used here on earth.  As the aperture gets larger and larger they capture more and more photons and thus reduce the total exposure time required for the same quality of image with all other things besides aperture being equal.

 

Throw in other variables, like vibrations from whatever, clouds, etc., you may regret going for more pixels and the reduced speed.  

Rnyboy, What you say is accurate for resolution.  What GroupJ said is accurate for resolving more detail.  So you are both right.  How about that!  :-)

 

Reyleigh Limit resolution is described in terms of being able to distinguish the light from two identical bright point sources, for example being able to separate double stars.  But DSO are typically not double stars.  For DSO, resolving more detail is most often determined by the signal to noise ratio.   SNR is improved by gathering more data, which can be achieved by either longer exposures or larger apertures (or both).


Edited by Rickster, 03 May 2020 - 03:45 PM.

  • bsavoie and elpajare like this

#34 Rickster

Rickster

    Apollo

  • *****
  • Posts: 1,177
  • Joined: 09 Jun 2008
  • Loc: NC Kansas Bortle 3 SQM 21.8+

Posted 03 May 2020 - 04:29 PM

And, it looks like I need to bring up one more common misconception, the idea that small pixels are less sensitive than large pixels.  This idea is usually supported by the logic that in a rainstorm, a small bucket catches less raindrops than a large bucket. Well, actually, it gets far more complicated than that.  https://www.flir.com...ra-sensitivity/

 

It gets so complicated and confusing that vendors, like ZWO, don't even publish sensitivity specs anymore. 


Edited by Rickster, 03 May 2020 - 04:30 PM.

  • Broglock likes this

#35 rnyboy

rnyboy

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 29 Oct 2019
  • Loc: Rochester, NY

Posted 03 May 2020 - 04:32 PM

No, not all the images were taken with the same setup.  Unfortunately I've been trying a lot of things over a short period of time and haven't been careful in recording everything.

 

Regarding your:  "It would be interesting to compare F6.2 with F3.3 for image quality."  That's a BIG "It depends" on what you mean by image quality.  Right off the bat the FOV is 2x different between the two options and it would take me 4x longer exposures to get to similar S/N at f/6.3.  On the other hand at f/6.3 I don't have that lower quality and smaller in dia. 1.25" 0.5x reducer in the optical stack.  So the f/6.3 images seem to be a bit sharper and the exposures are actually somewhat less than the theoretical 4x longer.  I think this is related to less scattering due to less glass in the optical stack and the f/6.3 reducer also being a coma corrector.

 

Right now I have the options of working at f/3.3, f/4, f/5, f/6.3, f/10, and f/20 (with 2x Barlow).  I still might have something of an f/2 option but at best I know if there is going to be a "reasonable" image it's only going to be in a small central region of the sensor, and this also assumes I don't run out of focuser travel; so mostly under-sampled images in the f/2 case.

 

I'm still enough of a noob at this that I typically don't have a single object I'm going after but more experimenting with what I can see and get with the particular f-ratio set up I start with for the night.  I'm sure when I go back to reimage some of the stuff I've already imaged I'll pick a more optimum condition based on what options are available to me once I get it all more sorted out.  My second attempt at the M104 (sombrero galaxy) is probably a case in point.  I think my second attempt at f/3.3 is actually a better quality image compared to my first attempt at f/5 because there isn't much detail to M104 to begin with and I had better S/N at f/3.3.  So the cropped image is brighter, larger, AND shows better definition of the sombrero's dark belt even though it covered a smaller area on the sensor.

 

The biggest determination of how I would set up for imaging an object would be its size, magnitude, and surface brightness.  Pretty much if the object is very dim then I'm forced to go with f/3.3 and having to live with it possibly being small and under-sampled in the final image because I'm limited in how long I can image with the SE mount.  For the brighter objects I have more options and have a chance at picking any available f-ratio option I have that allows the object to fill the frame nicely, or at least be as big as it can be in the frame at f/20 like for the planets.  One heck of a lot of the galaxies and nebula I've tried to look at are in the tens of arcsec to a few tens of arcmins in size with magnitudes of ~7 to ~9 and surface brightness values of ~12 to ~15.  With my 6SE that cheap 0.5x reducer is something of the sweet spot for exposures vs size of image filling the frame.  


Edited by rnyboy, 03 May 2020 - 04:57 PM.

  • Broglock likes this

#36 rnyboy

rnyboy

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 29 Oct 2019
  • Loc: Rochester, NY

Posted 03 May 2020 - 05:24 PM

Hi Rick,

 

When it comes to "sensitivity" with the current Sony CMOS sensors I do tend to think of pixels as being larger and smaller buckets with the sensitivity per unit photo-active area being pretty much the same.  To continue with the bucket analogy I think of the dynamic range as how much the bucket can hold.  So the trade off for the same total exposure time and sensor size is having dimmer (lower S/N) images with better resolution of details with smaller pixels or brighter (higher S/N) and lower resolution detail images with the larger pixels.  Unfortunately the buckets often overflow from things like stars.

 

I think the devil is in the details in regards to what types of noise are present and the magnitudes of those noise elements.  Which I think are more a secondary consideration when compared to pixel size.

 

With my 6SE I'm struggling more with speed than image resolution.  But both are significant issues.  The more limited FOV with my sensor is a smaller concern because there are tons of things that one can find to fit nicely in the frame if one could only get the desired S/N.

 

Your thoughts?



#37 Rickster

Rickster

    Apollo

  • *****
  • Posts: 1,177
  • Joined: 09 Jun 2008
  • Loc: NC Kansas Bortle 3 SQM 21.8+

Posted 03 May 2020 - 07:00 PM

Well, which holds more, four one gallon buckets, or one 4 gallon bucket?  Same, right?

 

Which collects more light, four 1 square foot holes or 1 four square foot hole? Same, right?

 

The old sensor rules and ways of thinking changed when they went from film to digital, and then changed again when they went from CCD to CMOS.  So much of what you hear and read is out of date and no longer accurate.  You really have to be skeptical, and sometimes a bit jaded. 

 

In the case of pixel size, the big change came with the low read noise that CMOS chips provide.  Now you can 2x2 bin (4 buckets) to make the equivalent of 1 large bucket, with very little read noise penalty.  For example, many have written that the 183 is not as sensitive and has a smaller full well than a 294.  Sure, it seems that way when you first try to use them, because the 183 has smaller pixels.  But a 183 can match the full well capacity and sensitivity of the 294 if you bin to make super pixels. 

 

Same thing for focal reduction.  It makes sense for film and for NV sensors, but with CMOS sensors, you aren't gaining any system speed by increasing focal reduction.  FR does increase the speed at the pixel by pixel level because it increases the flux at each pixel.  But if all you are after is speed (and not FOV) you can obtain the equivalent system speed by binning.

 

Keep in mind that we are talking about digital signals.  We can have infinite gain once the signal has been converted to digital.  Digital gain is just arithmetic multiplication.  You can use the software to make the image as bright as you want.  And you can combine pixels to increase display pixel brightness (sum binning) or combine pixels to increase display pixel SNR (average binning).   And do any number of other mathematical manipulations.  And when you display your image on a monitor, chances are the binning gets changed once again to match the monitor resolution. 

 

The bottom line is to make sure your camera has the smallest pixels that you think you will ever want.

 

Here is an example.  My main scope is a 16in f4.4 that I typically run at f5, 2000mm focal length.  It works well with 10micron pixels.  But, I typically have 4 scopes piggy backed on it, ranging from 264mm to 1000mm fl.  I sometimes switch cameras around.  So I bought a 183.  In the 16 in scope I typically bin at 4x4.  In the smallest scope, I don't bin (1x1).  And once in a blue moon, when conditions are extraordinary, I will bin at 1x1 and use 1 sec sub-exposures with the 16in scope to get the max resolution and smoothness that the sky has to offer.  And yes, it makes a difference.

 

Oh, and lets not get started on the way Sony rates their sensors.  That is a marketing rabbit hole.  tongue2.gif


Edited by Rickster, 03 May 2020 - 08:28 PM.

  • bsavoie, roelb and Broglock like this

#38 rnyboy

rnyboy

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 29 Oct 2019
  • Loc: Rochester, NY

Posted 04 May 2020 - 07:32 AM

Rick,

 

True on what you say about the 4 buckets but then you left out addressing which setup has better resolution, the four smaller buckets case if there's enough water to fill them appropriately.  The amount of water available is probably the biggest issue in all of these discussions anyway.  Hence the aperture is king law.

 

Aside from buckets, and sticking to CMOS to leave out as much as possible how the different types of noise enter into the thinking, isn't it generally correct to think of the sensor issue as making the tradeoffs between S/N, resolution, and exposure time?  It's like the old joke about cost, quality, or quantity... choose any two you want and then you have to live with the reality of where the third variable now sits.  Of the three for imaging everybody wants a decent S/N as something of a given, so that variable is the closest to being fixed and that just leaves trading off resolution with exposure times.

 

One's existing scope setup pretty much dictates how much wiggle room there is in trading off resolution and exposure.  If one's goal is more details than it's smaller pixels, but then forcing longer total exposure times, perhaps impossibly long if the wrong sensor is used in one's current imaging setup and then to get a "decent" image (S/N) with the same sensor will require moving up in aperture if one wants to keep about the same FOV, or going to lower f-ratio but with now a larger FOV, or a better mount.  Like you said, if one has a high resolution/small pixel sensor one can then bin in software.

 

Binning is an option, but again it's a tradeoff between resolution and exposure basically because the area that the object illuminates on the sensor doesn't change, so it effectively now spans fewer pixels.  I'm not sure, but the only reason I can see for binning is if viewing conditions are poor so having better resolution isn't helpful but shorter exposures could be.  Or maybe in my f/3.3 case, if an object is rather small in the FOV to remove the 0.5x reducer in order to toss out the glass and any artifacts associated with it, and operate at f/6.3 and bin.  That could perhaps give me a bit better quality image with the same effective pixel resolution and total exposure time all other things equal? 

 

You have a big aperture scope on a sturdy equatorial mount and I can see why your choice puts weight on small pixels and better resolution given the current crop of sensors available.  You're lucky in that both binning and likely dithering are options available to you.  Me, I have a small 6SE, and with that in mind I had to put more weight on larger pixels/lower resolution because I'm much more aperture and mount limited and I focused on setting up a jack-of-all-trades and master-of-none scope that can be used for a fair number of appropriately sized and moderately bright to bright DSOs and the solar system type bright things.  For the planets I'd love smaller pixels and more resolution and I may yet buy one of the cheaper higher megapixel/smaller pixel planetary cameras just for the smaller brighter things.  I'm even thinking the next bigger step maybe to a Hyperstar for the 6SE along with a higher megapixel camera.  Beyond that I think will require more major changes requiring a different scope and/or mount.


Edited by rnyboy, 04 May 2020 - 09:05 AM.

  • bsavoie, roelb and Broglock like this

#39 Rickster

Rickster

    Apollo

  • *****
  • Posts: 1,177
  • Joined: 09 Jun 2008
  • Loc: NC Kansas Bortle 3 SQM 21.8+

Posted 04 May 2020 - 03:23 PM

Rnyboy,

 

I think we are fundamentally in agreement. 

 

System speed, in terms of data throughput, is basically determined by aperture size minus transmission losses and minus photon to digital data losses.  It is not affected by focal reduction or binning because neither of these impact the number of photons converted per unit area of the sky.  Aperture is limited by physical size and budget constraints.  Today's CMOS cameras are near the theoretical max for conversion of photons.   And the transmission coefficient of today's optics is also near the theoretical max.  So, in today's world, the only way to substantially increase system speed is to increase aperture. 

 

Now, given that improving SNR is the top priority, data throughput becomes the top priority because it determines how fast SNR can be improved.  Doubling the data increases the SNR by the square root of 2, and so forth.  As you say, aperture is king.

 

Given that we all have to work with the aperture that we have, for now, it comes down to optimizing.  As you said, there are a number of trade offs to be considered.  So, what can be done?  Unfortunately, not a lot.  Once you have enough focal reduction that your signal swamps read noise within the period of your subexposure, further focal reduction doesn't improve SNR.  With today's low noise cameras, the swamping point is reached so quickly that it is rarely a factor (Unless you are aggressively filtering, such as when using a Ha filter.  Then more focal reduction is more better.)  Focal reduction will increase FOV, if that is desired.  But the FOV for a 6in SCT is limited by the baffle tube (primary mirror support tube) for reductions greater than about 0.63.  This is the reason that I maintain that the time proven f6.3 SCT reducer/corrector is optimum for the 6in SCT.  I see that Astrojedi is running f5 with a C9.5 these days.  So I presume that he has found that f5 is optimum for the larger baffle tube of the 9.5in SCT.  And Alphatripleplus is having success with 2 stacked f6.3 reducer/correctors when coupled with an Ha filter and the sensitive (yet small sensor/small pixel) 290 minimono.  Further, the guys using Newtons seem to agree that f5 is near optimum.  (Gotta luv an OTA that is the correct FR straight out of the box!)

 

So yes, I agree that it is all about finding the right trade off for your rig/skies/targets.  I have no doubt that your drive to learn, excel and experiment will get you to the right combination in record time.  I am curious to see where you will end up.


Edited by Rickster, 04 May 2020 - 03:23 PM.


#40 rnyboy

rnyboy

    Ranger 4

  • -----
  • Posts: 323
  • Joined: 29 Oct 2019
  • Loc: Rochester, NY

Posted 04 May 2020 - 03:37 PM

Where I'll end up is easy...      In the Poorhouse.  lol.gif


Edited by rnyboy, 04 May 2020 - 03:43 PM.

  • Rickster and GazingOli like this

#41 GazingOli

GazingOli

    Sputnik

  • -----
  • topic starter
  • Posts: 32
  • Joined: 13 Oct 2019
  • Loc: Stuttgart GERMANY

Posted 04 May 2020 - 05:05 PM

Anyway: very interesting discussion. Thank you all. Even if I did not yet understand all of it. I was not aware where my relatively simple question (I thought) could lead to. :)


  • Rickster likes this


CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics





Also tagged with one or more of these keywords: beginner, EAA, SCT, equipment



Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics