DSO Photography for Dummies - Telescope and Sensor

Introduction | Deep Sky Photos | Applications | Final Words | Links | Appendix: Collection of Rules of Thumb

On this page I want to investigate how telescope and camera sensor can be tuned to each other for deep sky images "the easy way".

Notes:

For the Hurried...

The quality of the fit of a camera sensor with a given pixel size to a given telescope focal length can be judged on the basis of its image scale (rule of thumb):

The image scale should lie between the guiding values of 1 and 2 (often, a value of 1.5 is mentioned). If you want to consider the seeing, halve the FWHM value ["] for the local seeing:

and use this value or these values as a guide.

The rules of thumb that are presented and derived on this page can be found in Appendix: Collection of Rules of Thumb.

 

Introduction

Questions...

There are a large number of astronomy cameras from different manufacturers on the market. A distinguishing feature is the size of the cells of the camera sensor, also called pixel size. Hobby astronomers, who want to get into astrophotography or EAA (Electronically Augmented Astronomy) or even buy another astronomy camera, are thus faced with the question of what pixel size the sensor of such a camera should have in order to fit the focal length of their telescope or telescopes optimally ("optimal adaptation"). Conversely, for a given camera, i.e. with a given pixel size of the sensor, the question arises what focal length your telescope should have so that it fits it optimally. This raises a number of questions: Why does pixel size matter? What does "optimal fit" mean in this context? And how do you find it? I would like to answer these questions in the following!

Answers

Digitization...

Unfortunately the answer to these questions is not easy and has to be somewhat "theoretical". First of all, we have to realize that using a digital camera on a telescope is a process in which an analog signal, the optical telescope image, is converted into a digital one, namely the image produced by the camera sensor. Ideally, this conversion, also called digitization, should be lossless, so that in the digital version even fine or, more so, the finest details of the original are preserved. For example, if you digitize music for a CD, the aim is to transfer all audible frequencies, i.e. all frequencies between 20 and 20,000 Hertz. But how do you achieve (as far as possible) loss-free digitization and how does this look like in concrete case of a telescope with a digital camera attached?

The Digitization of Spatial Signals (Images)

While when digitizing temporal signals the analog signal is measured (sampled) in rapid temporal succession, spatial signals are measured (sampled) "side by side", that is, spatially distributed and often temporally in parallel. In digital photography, where two spatial dimensions are to be captured, this "spatial juxtaposition" is realized by rectangular sensors, which are built up from a matrix of tiniest light-sensitive cells, called pixels. Here, too, the aim is to preserve the finest details, that is, to prevent objects and spatial structures that are as small as possible from disappearing. In the case of a telescope, these are the smallest stars that a telescope can show. The size of these "smallest stars" is determined by the resolving power (resolution) of the telescope, which depends on the aperture of the telescope. So these "smallest stars" have to be obtained when imaging with a digital camera!

And now to the Initial Question, the Question of the Pixel Size!

A camera connected to a telescope captures the optical image produced by the telescope with a sensor consisting of a rectangle of tiny sensor cells, the "pixels". And, as we know from digital photography, the number of pixels that a camera sensor has is important - and for a given sensor size, this number also determines the size of the pixels, which we usually care little about. This is, however, different in astrophotography; here the size of the pixels plays a role, and precisely in the question of how to achieve the best lossless digitalization possible. Our somewhat "imprecise" initial question, namely, what size the pixels should have in order to achieve an "optimal adaptation" of the telescope and camera sensor, can now be reworded as: What size should the pixels of the camera sensor have so that the optical signal can be digitized without losses so that that even the finest stars that the telescope can show are preserved.

The Answer, theoretical and general at first...

This question is first of all answered in general terms by the Nyquist theorem: It states that the "sampling rate" must be at least twice as high as the highest frequency to be transmitted. For CDs, therefore, 44 kHz is chosen in order to transmit 20 kHz safely. In the case of spatial signals (we speak of so-called "spatial frequencies", more difficult for laypersons to imagine...), the "receiving grid" of sensor cells must be at least twice as fine as the finest details of the original image, which should still be preserved.

And now practical!

For astronomy cameras, this means that the smallest imageable stars must fall on at least two pixels for them to be imaged "optimally" (if they fall on three pixels, the stars become even rounder...). The finest stars that a telescope can show correspond in size to its resolving power, so a pixel must be half the size or less than the resolving power of the telescope used. So we basically got the answer to the question asked at the beginning! What is still missing are formulas to calculate the optimal pixel size, because the resolving power is given in arcseconds and the pixel size in micrometers. I have found such and other formulas on the Internet and would like to present them in short form below. More detailed formulas and derivations as well as reasons for certain factors and values can be found on page Telescope and Sensor.

Even more practical: The air turbulence (Seeing)!

In astronomical practice, there is unfortunately still a complication! The air tends to be restless and turbulent, in English we speak of "seeing" (I will use this term in the following), and this enlarges the star images to some degree. In practice, this does not have an effect on short time exposures (moon, sun, planets), but it does have an effect on photos with longer exposure times, like deep sky photos. For these photos, the telescope resolution is therefore not important, but the larger seeing value (as a FWHM value), which in principle is a measure of the size of a "bloated star". This case can be treated with the formulas mentioned above by entering the desired FWHM value into the formulas instead of the resolution (see below).

Why "optimal pixel size"? Types of sampling

The formulas for pixel size on the Internet typically refer to an "optimal pixel size", and I have used this term as well. In fact, the Nyquist theorem has only an upper limit on pixel size, and therefore the pixels might be as small as you like. So there must be practical reasons for the upper limit to be the optimum and therefore also the lower limit, although in certain cases, such as the image scale formulas listed below, you may want to aim for a range around the optimum.

To the upper limit first! If a star falls on less than two pixels, the digitized image becomes coarser than the original. In "technical jargon", this is referred to as "undersampling". The Nyquist theorem helps us to avoid this! Now to the lower limit! Basically, the larger the pixels of a sensor are, the more light-sensitive it is (and the pixels themselves as well). Small pixels therefore lead to a lower sensitivity, and therefore the pixels should be as large as possible to keep exposure times short. They are, as we learned above, when a star falls on exactly two pixels. The range around this optimum is also called "good sampling". However, smaller pixels are not only less sensitive to light, but in the case of astronomy, where we are dealing with weak signals, the smaller the pixels, the more the signals, i.e. stars, spread over increasingly more pixels. This further weakens an already weak signal. On the other hand, the more pixels an object is distributed over, the more details appear (provided that these details can be reproduced). Therefore, in applications where there is enough light available, such as in lunar, solar, and planetary photography, this approach, called "oversampling", is used in practice. For this purpose, formulas have been developed which calculate an optimal compromise between details and exposure time (see page Telescope and Sensor).

Note: On this page, I will only consider the case of deep sky photography; the case of lunar, solar and planetary photography is also covered on the page Telescope and Sensor.

Outlook

In the following I will introduce some simple formulas for deep sky photography, for which there are often also "rules of thumb", which simplify the calculations. The formulas for pixel size and telescope focal length are a direct application of the approach just described. For the other formulas I did not find any derivations, but they are also based on the basic principle described here. More detailed formulas and derivations as well as reasons for certain factors and values are not to be found on this page, but on page Telescope and Sensor.

 

Deep Sky Photos

In the following I will present some simple formulas for the deep sky photography; often there are "rules of thumb" for them, which make things easier in practice:

  1. If you are looking for a suitable camera for deep-sky photography, you will use the formulas for pixel size and telescope focal length where you can also consider the influence of the seeing.
  2. If a camera is already at hand, you will want to determine the image scale for different telescopes in your own equipment, where there is also the possibility to take the seeing into account.
  3. And finally, the recommended focal length range of a telescope can be determined for a sensor with the help of the image scale (with and without seeing influence).

(1) Pixel Size

Depending on the Resolution

For the optimum pixel size or telescope focal length, the following formulas have been developed, in which the resolving power of the telescope after Rayleigh is indirectly a determining factor:

This formula is typically not used for deep sky photos and only presented here for reference (they are used in a table further down).

Depending on Seeing

For DSO images, the influence of seeing is usually taken into account when fitting a camera sensor to a telescope. Instead of the resolution, the local seeing is entered as an FWHM value (in arcseconds) in the formula (rule of thumb) for the pixel size (or telescope focal length):

Example

Depending on the Size of the Airy Disk

The diameter of the Airy disk, which is the effective aperture diameter of an optical system, determines its resolving power. Two points can be separated reliably according to the Rayleigh criterion if the maxima of their images are separated by at least the radius of the Airy disk. The diameter also indicates the minimum size with which stars are imaged in the telescope.

The diameter D (length, angular size) of the Airy disk is calculated according to the following "rules of thumb" (for the exact formulas see the appendix):

Often only the rounded value "277" is used. In angular measure, the Airy disk is twice as large as the Rayleigh resolving power (on which it is based), because the resolving power refers to the radius, while the Airy disk is typically used with the diameter. And it is four times as large as the "optimal pixel size". Since the Airy disk should fall on two pixels, the "ideal pixel size" is half the size of this one, whereas the "optimal pixel size", which also covers two pixels, is half the resolving power and thus, a quarter of the size of the Airy disk.

When observing DSO, the Airy disk may be larger than the current seeing values, measured as FWHM values (in seconds). In such a case, the larger value, i.e. the size of the Airy disk, has to be used. For a comparison with the FWHM value, the size of the Airy disk in seconds is needed, for determining the pixel size, its size in µm. The latter has to be halved for arriving an the sensor's pixel size, because the Airy disk size refers to two pixels.

Example (Vaonis Vespera)

(2) Image Scale

For "Good Sampling

The image scale of a camera sensor with a given pixel size at a given telescope focal length is used to judge the quality of the fit of a camera sensor/telescope combination. It is calculated as (Rule of thumb):

Often only the rounded value of "206" is being used.

For the deep-sky photos, the rule of thumb for "good sampling" is to aim for an image scale of about 1 to 2 seconds per pixel*. Values for the image scale above 2 are called "undersampling", values below 1 are called "oversampling".

Example (TLAPO1027)

*) Other specifications that I found are: 1.25, 1.5, 1.5-2, 1-2.5 and even 0.7-3. Reasons for these values are usually not given, but obviously they are based on typical values for the seeing (in Central Europe). More about this below!

Depending on the Seeing

In order to take the seeing into account, one simply halves the seeing value (FWHM) in practice and uses this as the desired image scale value. Thus, the image scale calculated according to formula 3 is not checked according to whether it lies between the "ideal" values of 1 and 2, but rather whether it is close to the image scale value determined by the FWHM value. More on this below!

To determine the pixel size of a sensor at a given telescope focal length, the formula for the image scale has to be be converted; the same applies to the telescope focal length at a given pixel size:

Example (TLAPO1027)

Astronomy.tool "Tweak"

In order to achieve "round" stars, the authors of the Astronomy.tools Website propose to sample with the 3-fold frequency of the analog signal. First of all, they assign FWHM value ranges to the different seeing conditions, and by dividing the values by 3 (for the lower value) or 2 (for the higher value) they arrive at "recommended" value ranges for the image scale (which they call "pixel size"...). This leads to the following table, in which I also included the standard procedure of "halving":

  Seeing
Image Scale
 
Seeing Conditions FWHM- Value Astronomy.
tools
Rule of Halving Remarks
  From To From To From To  
Exceptional good seeing 0.5" 1" 0.17 0.5 0.25 0.5  
Good seeing 1" 2" 0.33 1 0.5 1  
OK seeing 2" 4" 0.67 2 1 2 Mean value = 3" for Central Europe => 1.5
Poor seeing 4" 5" 1.33 2.5 2 2.5  
Very poor seeing 5" 6" 1.67 3 2.5 3  

Using an online calculator on the Astronomy.tools Website, you can calculate the image scale for your configuration (it calculates according to the rule of thumb given above) and relate it to the values of the local seeing. So you do not check whether this value lies between 1 and 2, but whether it lies within the limits given by the local seeing conditions.

Example

Where Do the Recommendations for the Value of the Image Scale Come from?

As already mentioned, Internet sources usually do not provide any justification for the "ideal" image scale values given. My suspicion that they are based on typical values for seeing in Central Europe seems to be confirmed by the table above.

The often mentioned value range of 1-2 for the scale of reproduction corresponds to "OK Seeing", the also often mentioned value of 1.5 corresponds to the "average seeing" of 3", which H.J. Strauch states for Central Europe. Other values or value ranges seem to be merely "variations" of this.

(3) Recommended Focal Length Range

With the help of the recommendation that the image scale should be between 1 and 2, one can also determine the focal length range recommended for a sensor and thus check if one's own telescopes are in a suitable focal length range. For the sake of simplicity, I use here the rule of thumb for the image scale, which I reform accordingly:

To determine the focal length range, I now insert the values "2" and "1" into the formula one after the other:

If you want to include seeing (see Astronomy.tools), just enter the corresponding values for the image scale (upper and lower limit, e.g. 0.67 and 2 for "OK Seeing") into the formula.

Example

 

Applications

In the following, I present tables with calculation results based on the above formulas for my and some other telescopes and for camera sensors that are relevant for me. At the end of this section, I try to check the suitability of three sensor sizes for my telescopes using a reduced table.

Calculations for My and Other Telescopes and Some Sensor Sizes

I calculated the following table using an Excel spreadsheet based on the formulas presented here.

Optimum Pixel Size

The optimum pixel size is calculated either by using the Rayleigh resolution or the seeing according to the halving rule.

Telescope
Resolution ["]
Optimum Pixel Size [µm]
Examples
Focal Length [mm] (mm) 
Aperture [mm]
f
After
Rayleigh
Via
Resolution
Via Seeing (FWHM)
2"
3"
4"
5"
Stellina 400 80 5 1.73 1.68 1.94 2.91 3.88 4.85
APO 80/480 480 80 6 1.73 2.01 2.33 3.49 4.65 5.82
Heritage 100P  400 100 4 1.38 1.34 1.94 2.91 3.88 4.85
TLAPO1027 714 102 7 1.36 2.35 3.46 5.19 6.92 8.65
PS 72/432 432 72 6 1.92 2.01 2.09 3.14 4.19 5.24
eVscope, Newton 114/450 450 114 4 1.21 1.32 2.18 3.27 4.36 5.45
Newton 114/500 500 114 4.4 1.21 1.47 2.42 3.64 4.85 6.06
Heritage P130 650 130 5 1.06 1.68 3.15 4.73 6.30 7.88
6" Newton. Explorer 150PDS  750 150 5 0.92 1.68 3.64 5.45 7.27 9.09
6" Newton  900 150 6 0.92 2.01 4.36 6.54 8.73 10.91
6" Newton 1200 150 6 0.92 2.68 5.82 8.73 11.64 14.54
8" Newton. GSD 680  1200 200 6 0.69 2.01 5.82 8.73 11.64 14.54
Skymax-102 1300 102 12.7 1.36 4.28 6.30 9.45 12.61 15.76
Skymax-127 1500 127 11.8 1.09 3.96 7.27 10.91 14.54 18.18
Skymax-127R 750 127 5.9 1.09 1.98 3.64 5.45 7.27 9.09
Celestron C8 2032 203 10 0.68 3.36 9.85 14.78 19.70 24.63
Celestron C8R 1280 203 6.3 0.68 2.12 6.21 9.31 12.41 15.51
Celestron C8R2 1016 203 5 0.68 1.68 4.93 7.39 9.85 12.31
Celestron C14 3500 350 10 0.40 3.35 16.97 25.45 33.94 42.42

Image Scale

The rule of thumb was used for the image scale, because the exact formula delivers the same numerical values. The more exact value of "206.265" was used instead of "206".

   Telescope
Image Scale ["/Pixel]
Examples
Focal Length [mm] (mm) 
Aperture [mm]
f
Pixel Size [µm]
2.4
3.75
6.45
Stellina 400 80 5 1.24 1.93 3.33
APO 80/480 480 80 6 1.03 1.61 2.77
Heritage 100P  400 100 4 1.24 1.93 3.33
TLAPO1027 714 102 7 0.69 1.08 1.86
PS 72/432 432 72 6 1.15 1.79 3.08
eVscope, Newton 114/450 450 114 4 1.10 1.72 2.96
Newton 114/500 500 114 4.4 0.99 1.55 2.66
Heritage P130 650 130 5 0.76 1.19 2.05
6" Newton. Explorer 150PDS  750 150 5 0.66 1.03 1.77
6" Newton  900 150 6 0.55 0.86 1.48
6" Newton 1200 150 6 0.41 0.64 1.11
8" Newton. GSD 680  1200 200 6 0.41 0.64 1.11
Skymax-102 1300 102 12.7 0.38 0.59 1.02
Skymax-127 1500 127 11.8 0.33 0.52 0.89
Skymax-127R 750 127 5.9 0.66 1.03 1.77
Celestron C8 2032 203 10 0.24 0.38 0.65
Celestron C8R 1280 203 6.3 0.39 0.60 1.04
Celestron C8R2 1016 203 5 0.49 0.76 1.31
Celestron C14 3500 350 10 0.14 0.22 0.38

Image Scale Depending on the Seeing (Repetition)

An image scale taken from the above table can now be compared either with the "good sampling" value range of 1-2 (or the value 1.5) or with the following image scale values, which are provided for different seeing conditions (I repeat the table for convenience):

  Seeing
Image Scale
 
Seeing Conditions FWHM- Value Astronomy.
tools
Rule of Halving Remarks
  From To From To From To  
Exceptional good seeing 0.5" 1" 0.17 0.5 0.25 0.5  
Good seeing 1" 2" 0.33 1 0.5 1  
OK seeing 2" 4" 0.67 2 1 2 Mean value = 3" for Central Europe => 1.5
Poor seeing 4" 5" 1.33 2.5 2 2.5  
Very poor seeing 5" 6" 1.67 3 2.5 3  

Focal Length Ranges for Different Sensor Sizes

The focal lengths were calculated for an image scale of 2 and 1.

Camera/Telescope Pixel Size Focal Length (2) Focal Length (1) Range Opt. Pixel Site for Native Tubes Fits...
µm mm mm mm µm
Stellina 2.4 247.5 495.0 250-500 1.68 PS72/432, APO 80/480
eVscope, ASI120, ASI224 3.75 386.7 773.5 400-800 1.32 TLAPO1027, SM127R, PS72/432, APO 80/480
Atik Infinity 6.45 665.2 1330.4 650-1300 n.a. TLAPO1027, C8R, C8R2, SM127R

Instead, the focal lengths can also be calculated for the border values of the respective seeing conditions.

Application to My Telescopes

In the following, I reduced the above table to my telescopes and tried to apply the "rules". At the columns for the optimum pixel size, I highlight the 3" column (and slightly the 2" and 4" columns); at the image scale columns, I backlight suitable cells in color and highlight the sensor of my Atik Infinity camera.

Telescope   
Resolution ["]
Optimum Pixel Size [µm]
Image Scale ["/Pixel]
My Telescopes
Foc. Len. [mm] (mm) 
Aperture [mm]
f
After
Rayleigh
Via
Resolution
Via Seeing (FWHM)
For Pixel Size [µm]
2"
3"
4"
5"
 
2.4
3.75
6.45
TLAPO1027 714 102 7 1.36 2.35 3.46 5.19 6.92 8.65   0.69 1.08 1.86
PS 72/432 432 72 6 1.92 2.01 2.09 3.14 4.19 5.24   1.15 1.79 3.08
eVscope 450 114 4 1.21 1.32 2.18 3.27 4.36 5.45   1.10 1.72 2.96
Skymax-127 1500 127 11.8 1.09 3.96 7.27 10.91 14.54 18.18   0.33 0.52 0.89
Skymax-127R 750 127 5.9 1.09 1.98 3.64 5.45 7.27 9.09   0.66 1.03 1.77
Celestron C8 2032 203 10 0.68 3.36 9.85 14.78 19.70 24.63   0.24 0.38 0.65
Celestron C8R 1280 203 6.3 0.68 2.12 6.21 9.31 12.41 15.51   0.39 0.60 1.04
Celestron C8R2 1016 203 5 0.68 1.68 4.93 7.39 9.85 12.31   0.49 0.76 1.31

The cells in the image scale columns are highlighted in cyan if they are in the range between 1 and 2 (halving rule); cells with values between 0.66 and 1 are highlighted in blue ("Tweak" according to Astronomy.tools). Text colors are meant to highlight relevant data.

I conclude from this:

The combination of both reducers also fits the C8, as I found out.

In short: on the TLAPO1027 the ASI224 (better with good seeing) and the Atik Infinity, on the PS 72/432 the ASI 224, on the Skymax-127 and C8 the Atik Infinity, but actually only with focal length reducers (on the 127R the ASI224 would also work).

 

Final Words

I hope that I present everything reasonably correct and understandable, so that other hobby astronomers can apply the formulas to their own equipment. If you want it easy, just read "fir the hurried"...

 

Links

 


Appendix: Collection of Rules of Thumb

The rules of thumb presented here use a light wavelength of 550 nm.

Resolving Power

Pixel Size and Telescope Focal Length (Based on Resolving Power)

Pixel Size and Telescope Focal Length (Based on Seeing)

Airy Disk

Image Scale

 

An den Anfang   Homepage  

gerd (at) waloszek (dot) de

About me
made by walodesign on a mac!
17.10.2020