A "new" method of collimating

[Dennis]

Hello,

I have just been reading the manual for CCDSoft which is the camera control and image processing software provided by SBIG for their ccd cameras, and I found an interesting function called “3-D Surface Plot” under the “Research menu”. This command displays a three-dimensional graph of an image.

So, I grabbed a few old images and cropped a single star to generate a 3-D surface plot for that single star, then I viewed the plot from above and lo and behold, I think it tells a story about collimation. The images I chose were made with exposures of between 60 and 90 secs with my ST7, which may be too long? Anyhow, here are the results for:

Celestron C9.25 at F6.3 (using reducer/corrector)
Celestron C9.25 at F10
Vixen 102mm f9 refractor

It shows the Vixen refractor is nicely collimated whereas collimation appears off with the C9.25. Next time I am out (=when Brisbane gets some clear skies) I may try collimating using a star with say, 5 sec exposures.

Cheers

Dennis

[janoskiss (Steve H)]

I would not make too much of it yet. Do the grid squares represent pixels? Is the distortion of the grid due to perspective? Are we looking straight down on the surface plot?

You can take an image of a star in photoshop or similar image editor, blow it up, and then posterise for a similar representation.

[davidpretorius]

i like to pretend i understand this! naaagh,i don't!

[asimov (John)]

This can be done in Astraimage too. I've yet to work out what I'm seeing though..

[Dennis]

Unfortunately, it appears that there is no further information in the CCDSoft help files, or in the manual. I am assuming that that the function, when invoked, plots a contour map of the number of photons per charge well (pixel) and then allocates a colour accordingly.

When I isolate a single star in the Vixen 4” refractor image, and magnify it (see new image uploaded in this post), we can see the x & y cursor position as well as the photon count of the pixel at that location, as displayed in the status bar. As the mouse cursor is moved across the magnified star, these values change. In the example uploaded, the coordinates are x=009, y=012 and the number of photons =10,532 at that location.

• The image scale with the ED102S refractor at f9 is 2.02 arc sec per pixel, which would be good seeing.
• The image scale with the C9.25 at F10 is 0.79 arc sec per pixel, which is extremely rare.
• The image scale with the C9.25 with the F6.3 Reducer/Corrector is 1.25 arc sec per pixel which is excellent seeing.

The example images uploaded in the original post show stars with exposures for over 60 secs, therefore atmospheric smearing, drive errors, etc will have affected the contour plots. However, for shorter exposures of say 3 to 5 seconds, I’m sure we would be capturing the actual intensity plots of the diffraction rings; hence they would provide us with a view of the state of collimation of the optical system.

Until the next clear night…….

Cheers

Dennis

[Astroman (Andrew Wall)]

Is that a planet orbiting that star?

Hmm not sure how you would go about collimating using this method, it seems a little long timed. Compared to say a visual star test or laser. Wouldnt your mount have to be taken into account aswell? I dont think there is any mount available that could track this well to perform such a test over that period of time exposure. Maybe shorter say 5s on a brighter star maybe? but then you may get blow out which would give you inaccurate results too. Then again, I may be barking up the wrong tripod here.

[Dennis]

Originally Posted by davidpretorius

i like to pretend i understand this! naaagh,i don't!

Hi Dave

Pardon my rudeness for not explaining the images more clearly, I was kinda just thinking out aloud at the time.

I once read a nice explanation of a ccd chip, where the writer likened the chip to a paddock (in Tasmania) filled with an array (rows and columns) of buckets. When it rains, each bucket will collect a number of rain drops; some more, some less.

This is analogous to the individual photo sites on the ccd chip that collect photons rather than rain drops. So, the first image with lots of multi coloured spikes is a 3-D representation of how many photons have been collected by the entire ccd chip at each photo site.

The number of photons counted at each photo site is then converted to an image on the computer screen, made up of pixels. Each pixel on the screen represents one photo site on the chip. In my case, with the SBIG ST7 ccd camera, the chip has 765 x 510 photo sites (also known as charge wells) that collect photons and so when the image is drawn on the screen, you get an image of 765 x 510 pixels.

So, what the CCDSoft camera control software has done is to paint a picture of the 765 x 510 image captured by the ccd chip in a 3-D colour map. I think my ccd camera can hold a range of 0 to over 65,000 drops of water in any one of its buckets. Because the screen cannot display 65,000 colours, the software has said something like;

“I’ll display 0 to 1000 drops of water as red”
“I’ll display 1001 to 5000 drops of water as green”
“I’ll display 5001 to 10000 drops of water as blue”
“I’ll display 10001 to 15000 drops of water as yellow”
“I’ll display 150001 to 20000 drops of water as pink (the colour of your tutu!)”

…etc., right up to the largest count which in the 3-D graph is brown or a dirty green colour.

For the other three images, I cropped a single star from the full ccd image and ran the 3-D plot on the individual star alone. I then rotated the plot, as best as I could, as if I was looking down on the star from above, a bit like looking at the contours of a hill or mountain on a map.

The purpose of this? Bored with the jets stream; bored with the clouds and rain, so I thought I would experiment and see if I could use these contours as a representation of the collimation of my scope. Nice, concentric rings would indicate good collimation. Off set rings would indicate not so well collimated.

Hope I haven’t confused you any more!

Cheers

Dennis

[Merlin66 (Ken)]

Dennis, you're on the right track.
For optimum definition on this "type" of method you need to be able to freeze the atmospheric seeing, so this normally means a very short exposure say 1/100 sec, obviously then we need to use brighter stars. Also the image HAS to be exactly focussed, so there's a two stage approach
- find the minimum image size ( diam) ie in focus
- measure the light distribution ( accuracy of optics etc)

There's a freeware package called Reduc which can be used to assist ( it's really for double star measuring.... but a very similar issue).