Tak Epsilon Collimation

[MarkS]

Here's the latest episode in my ongoing battle with Tak collimation.  My final image taken at Kelling Heath was more or less spot-on.  But the two I took at DSC at the weekend were not.  Take a look at these diffraction spikes fom the central star (resampled by a factor of 2 for clarity):



Notice how the spikes are sharp at the bottom then get thicker (even splitting apart) as they go towards the top.  That ought to be telling me something (if I knew how to interpret it).

And here are the stars from the top left and top right of the image:



Notice how the diffraction spikes have now completely split into 2 (these are not double stars but are representative of the stars in that region of the picture).
Again that must be telling me something.

However the stars at the bottom of the image are pretty sharp with sensible diffraction spikes.

Last night, using the Cheshire to check collimation, I reckon it was very slightly out in the up down direction (relative to the camera orientation) so this is consistent with what I'm seeing in the image.  More experiments are required to confirm this.

Mark

[Rocket Pooch]

Mark,

If the stars are splitting this is where the focus is not spot on, it might not be collimation but the flatness of the camera on the scope.

I gues the way to find out is to see how flat the image is with CCD thingy.

Chris

[Mike]

Why not use CCD Inspector?

[Mac]

Taken off of the large image.

[RobertM]

Looks like the sensor is possibly tilted to the image plane.  I would have expected curvature of less than 10% with that flattener and the center of curvature should be in the middle of the image.

I'm not sure how accurately canon align their sensors but could that be part of the problem, after all we are talking very fine tolerences here !

Robert

[The Thing]

I'm not sure how accurately canon   Marks' align their sensors but could that be part of the problem, after all we are talking very fine tolerences here !

Robert

It's been well modded. The strap on peltier was maybe a bit too much for the internal metalwork. Maybe time to get another 350D Mark? Or you could experiment with those little washers that shim out the CCD after changing the filter glass.

[Rocket Pooch]

ok that explains it then, my ED80 with flattener is about 16% I think.

[MarkS]

I've run the original stacks through CCD Inspector (before the asinh range compression was applied).  Here are the results:

Veil Nebula from DSC:


Elelphant's Trunk from DSC:


The above two results are incredibly similar.

But compare with North American Nebula from Kelling Heath:


I think using CCD Inspector and a bit of collimation tweaking I should be able to get this almost spot on.  This guy http://pagesperso-orange.fr/remi.petitdemange/materiel/e180/e180-img2.htm has a curvature of 9.2% with the Tak Epsilon 180.

[MarkS]

Yesterday (Sunday), besides spending half a day at work (don't ask!) I removed the Tak's primary and secondary, cleaned the mirrors and completely collimated from scratch.  Did some tests last night before the clouds rolled in.  Now I'm back home from work today (don't ask!) I can show you the result. Here's what CCD Inspector shows:



Looks dire doesn't it?

Well, actually, no.   Following that image I rotated the whole focuser unit with attached camera through 180 degrees and took another image.  Here's the result from CCD Inspector:



It looks almost identical to the first.

So what is that telling me?  The "problem" clearly rotates with the focuser/camera.  So almost certainly it is a squiffy CCD in my Canon (probably from the earlier Peltier mod).  Either that or it's the Takahashi 2 piece adaptor ring that isn't quite square.

Anyway,  I'll soon the the proud owner of a £215 Canon EOS 400D off eBay (goodbye to amp glow!).  So I'll do lots of further testing with it BEFORE I rip its guts out.

Mark

[Rocket Pooch]

Hi Mark,

It sounds like you are where I am with my F4 8", I have this in bits as well, nothing was setup well, I even thing the hole for the focuser is not in the right placed, but hey hoo, gives us something to play with, I have 8 months to sort it before I will be using it in anger.  New springs are on order and the flocking paper is on in standby as is the ultra black flat paint.

So hey fast scope rule!  Well if you can get them all pointing in the right way.

I have to also say an F5 newt is so simple to colliamate compared to the F4, what the F2.8 is like I don't want to think.

Chris
 

[RobertM]

I can see more police coming ...

[Rocket Pooch]

My ED80 with a field flattener

[MarkS]

Thanks to Ian's micrometer I have determined that there are significant inaccuracies in the manufacture of Takahashi's T-ring adaptor.  The way it was assembled, the "left" side was 80 microns thicker than the "right". It actually comes in two pieces so I then took it apart and found both pieces had inaccuracies of up to 40 microns and, by bad luck, it was assembled in such a way that the inaccuracy was cumulative.

I'm stunned by the size of these errors given that the scope itself is such a high precision instrument - the focal depth of field at F2.8 is  +/-  9microns (using the Rayleigh criterion at the wavelength of green light).

Anyway, I've now reassembled the T-ring in such a configuration that the maximum error is now 20 microns.  This might be enough to cure the problem, otherwise I'll have to resort to packing it with tinfoil "shims".

Just need a clear night to do some further testing now ...

[Mac]

This might be enough to cure the problem, otherwise I'll have to resort to packing it with tinfoil "shims".

Sounds like its a job that needs to be done.

If you do it, then thats one less problems to worry about.

20microns when the depth of field is 9microns. hmmmmm.

I'd say its worth a try.

[Mike]

Surely the ring is not fit for purpose. Can you not send it back to teh manufacturer for replacement?