Back Focus Issue?

[PhotoGav]

I have just put my Celestron EdgeHD 8" scope with dedicated 0.7x Reducer on my mount again after some time using my other scopes and have noticed immediately that stars in the corners are slightly elongated, especially in the top left and right. Here's a 300s image taken early this morning - lots of moonlight, so a bit washed out, but the star shapes are very clear.

The question is: what is causing this elongation?

The only thing that has changed is a move from Baader to Astrodon filters. The optical train is as recommended: Focal Reducer > T-Adapter without extension > 55mm back focus to chip (made up of 50mm QSI 683 back focus and a 5mm T2 adapter). The QSI manual talks about -1mm back focus due to the thickness of the Astrodon filters, compared to -0.7mm for the slightly thinner Baader filters. Images with this set up and the Baader filters were never perfect, but were better in the corners than the above image. Should the QSI back focus measurement be 49mm taking the Astrodon reduction into account? This would suggest that I am 1mm short of the recommended back focus distance. Would 1mm really make that difference to the image edges? Or is something else at play here?

Help, please!

 

[tomato]

I am definitely no expert, but if it was a back focus dimension issue, wouldn't you be getting the elongation even in all four corners? I certainly saw that when I had the dimension out when using a field flattener on a refractor. It does look more pronounced in the top left and right corners to me.

Is that therefore indicating some tilt in the optical train?

[PhotoGav]

30 minutes ago, tomato said:

I am definitely no expert, but if it was a back focus dimension issue, wouldn't you be getting the elongation even in all four corners? I certainly saw that when I had the dimension out when using a field flattener on a refractor. It does look more pronounced in the top left and right corners to me.

Is that therefore indicating some tilt in the optical train?

Mmm, yes, you make some logical sense with what you say. I probably need to carry out some diagnostic tests... does this plan make sense?

Start with shot as above, then rotate whole train through 180° (i.e. do a meridian flip) and see if the same corners are affected. If it’s the same corners (i.e. bottom corners now elongated), then it would suggest that it is a distance thing. If the top corners are affected then it would suggest that the train is sagging. Then rotate camera through 90° and see where the elongation is. If the elongation moves with the camera rotation it is something in the camera end, if they stay in the same place (i.e. new corners as the camera is rotated) then it is to do with a sagging train.

The only thing against what you say is that if my chip is capturing a ‘central’ portion of the light cone, but is off centre and capturing the top portion, then perhaps the elongation is equal in all corners but not visible in the portion being imaged. In which case it could be a distance issue!

@steppenwolf - please join the discussion!!!!

[vlaiv]

I would say tilt.

Not sure how sound is meridian flip as tilt can be due to gravity. Let's say we have two types of tilt - loose one and firm one. Loose one will depend on scope position - so aiming straight up should show even stars in all corners. Firm one would not change with scope orientation (there is tilt but sensor is not affected by gravity and does not change tilt in respect to optical axis).

It could be de-center as well, sensor not being square on optical axis. There is "logical" way to assert this, and you can probably test it via flats. For logic application - compare sensor size to fully corrected circle of your setup - if it's just right with no room to spare - sensor of optical axis means one of the corners is outside corrected circle - and this is why one corner is most affected. This way is not very reliable - since spec for fully corrected circle are often misleading (inaccurate, or different meaning of word "corrected" is used ).

Other way to check de-center (and I'm not meaning lens de center - but sensor center not being spot on optical axis) is to examine flats - these should be center symmetrical. Don't know how reliable this method is but I would try it.

Do you have laser? You can put it instead of eyepiece and observe where it hits front lens - it should be in center (make some sort of paper circular mask with marked center to help you out). Mind you, laser should be well collimated, but you can check this by rotating laser - if spot stays in the same place, it's collimated properly.

There is small chance that tilt + de-center will put laser spot in exact center of the front lens (angles of tilt and de center distance need to match perfectly).

HTH

[vlaiv]

Ah, there is one more thing to check, since it's EdgeHD (I wrote above with refractor in mind) - collimation, make sure it's as good as it can be.

[PhotoGav]

Thanks for the thoughts @vlaiv  - I’m pretty sure that the flats show that the sensor is off centre to the optical centre, that is from what I remember the flats being like when I last used this setup last year. Collimation is a good point. I have noticed during focusing that the circles are not evenly concentric. They are not drastically off centre so I haven’t worried too much - I hate collimating the scope! - but perhaps it is time to accept that I must collimate...

[steppenwolf]

40 minutes ago, PhotoGav said:

@steppenwolf - please join the discussion!!!! 

@PhotoGav I think this is partly a tilt/collimation issue. The filters may indeed be playing a part (thicker filters require, IMHO, the camera to be moved further back from the reducer and here is where we open the old can of worms again but for what it is worth, this is how I understand the correction to be!)).

I would start by checking that collimation thoroughly as 'tilt' can come from many sources and in an SCT system, the optics are the first culprit! Once you have an even star shape in all four corners, you can tackle the elongation v spacing issue.

[PhotoGav]

Lovely, thank you for your input @steppenwolf and confirmation that I should spend some time getting collimation bang on before anything else. I will update as and when...

[wxsatuser]

1 hour ago, steppenwolf said:

@PhotoGav I think this is partly a tilt/collimation issue. The filters may indeed be playing a part (thicker filters require, IMHO, the camera to be moved further back from the reducer and here is where we open the old can of worms again but for what it is worth, this is how I understand the correction to be!)).

The filter will push the focal point behind the sensor plane

The confusion/can of worms is between cameras with integral filters wheels and ones without.

QSI for instance say the filter reduces the back focus of the camera by the filter correction.
They do not say subtract the correction from the the image train.

[PhotoGav]

17 minutes ago, wxsatuser said:

The filter will push the focal point behind the sensor plane

The confusion/can of worms is between cameras with integral filters wheels and ones without.

QSI for instance say the filter reduces the back focus of the camera by the filter correction.
They do not say subtract the correction from the the image train.

Good, I'm glad it's not just me that is easily confused by all this... So, I have a QSI 683 WSG8 (i.e. integrated filter wheel and OAG, quoted back focus distance of 50.17mm) and Astrodon filters. The QSI manual says -1mm for the filters. What does that actually mean? What do I do with that information, if anything?!

[wxsatuser]

Subtract the 1mm from the camera BF then add 1mm in front of the camera.

[PhotoGav]

Just now, wxsatuser said:

Subtract the 1mm from the camera BF then add 1mm in front of the camera.

Lovely, thank you.

I am going to collimate (hopefully tonight) and see what difference that makes, if any. I will then start fiddling with train length and camera orientation... Oh, don't you just love it when it doesn't work and you don't really know why?!!!? ?

Thankfully I sorted the problems I was having with autofocus this morning between 4am and 5am. Perfect clear skies... why couldn't that have been yesterday morning, then I would have seen a total lunar eclipse, rather than just cloud cloud cloud.... ?

[wxsatuser]

This is the link to QSI.

http://www.qsimaging.com/downloads/QSI-500-600-Series-Back-Focus.pdf

[PhotoGav]

8 minutes ago, wxsatuser said:

This is the link to QSI.

http://www.qsimaging.com/downloads/QSI-500-600-Series-Back-Focus.pdf

Yup, thank you, that’s the manual sheet I’ve been looking at. It’s good to hear confirmation that the backfocus of the camera is reduced by the amount specified, which then needs to be added in front of the camera. It’s not hugely clear on that sheet. Also, does 1mm of back focus make much difference to the image?

[steppenwolf]

Blimey, someone who agrees with me on increasing the distance between the camera and the reducer! ?

[vlaiv]

19 minutes ago, steppenwolf said:

Blimey, someone who agrees with me on increasing the distance between the camera and the reducer! ?

Should that not be decreasing distance ?

[vlaiv]

21 minutes ago, steppenwolf said:

Blimey, someone who agrees with me on increasing the distance between the camera and the reducer! ?

Ah, my mistake - camera and the reducer - indeed that one should be increased to compensate for "decreased" focal position.

[wxsatuser]

1 hour ago, PhotoGav said:

Yup, thank you, that’s the manual sheet I’ve been looking at. It’s good to hear confirmation that the backfocus of the camera is reduced by the amount specified, which then needs to be added in front of the camera. It’s not hugely clear on that sheet. Also, does 1mm of back focus make much difference to the image?

This is a good point, to be honest I try to get mine as near as possible others say there is some leeway.
Could be another can of worms but IMO I would want that spot on the chip or as near as possible.
If the images look ok that should be enough.

[tomato]

Just caught up with this after my lunchtime post. As you say you have successfully run this set up previously albeit with different filters, hence something has changed and it could certainly be the SCT collimation, so that must be a good place to start. My 1980’s Schmidt Newtonian used to go out of alignment after every session (I do not like large secondaries supported through a hole in a piece of plate glass) so that’s why I’m 100% refractors at the moment.

[wxsatuser]

For anyone interested here is a little diagram of what happens when a filter is introduced.

The black lines are the light path without the filter and the green show what the glass does to the light.
The diffraction shifts the focal point behind the sensor.

[Barry-Wilson]

20 minutes ago, wxsatuser said:

For anyone interested here is a little diagram of what happens when a filter is introduced.

The black lines are the light path without the filter and the green show what the glass does to the light.
The diffraction shifts the focal point behind the sensor.

? - very good illustration Mike.  I have a sketch just like this in my observatory notebook to remind me of the basic optics if I'm ever fumbling in the middle of the night when tired!

Gav - incorrect spacing can create all sorts of odd shaped stars and then when mixed up with collimation issues, I can see that you have a few jobs ahead of you so plentiful reason to disappear into the shed with a mug of tea and a biscuit (or two) for a quiet hour ?.

[vlaiv]

1 hour ago, wxsatuser said:

For anyone interested here is a little diagram of what happens when a filter is introduced.

The black lines are the light path without the filter and the green show what the glass does to the light.
The diffraction shifts the focal point behind the sensor.

This diagram explains nicely what happens when no flattener is in optical train - focus is shifted further away from objective.

Indeed, with flattener it is the same thing - distance between flattener and sensor needs to increase (physical distance). This can be explained to same result in to different ways:

1. Flattener needs to be in certain position in converging beam after main lens (depending on primary lens focal length) and flattener does not really care what happens to the light after it passes thru flattener. It just sits there doing its thing. Since focus position moves away from lens, so does too from flattener - spacing between camera and flattener increases.

2. In above diagram we see that "physical path" gets longer than equivalent "optical path". Flattener needs to be placed at certain "optical path distance" from sensor, which in this case equates to longer physical path - hence longer distance from flattener to sensor.

[PhotoGav]

Excellent stuff, thank you all for your input - quite a physics lesson, which I love!

Wise words Barry... I had a couple of hours in the shed. I am now frozen. Anyway, the results of my fiddling are nicely demonstrated in these two images:

Image 1 - Collimation. To me, that looked good enough to not even think about fiddling with it.

So what to do?

I tried imaging either side of the meridian, but always the same. I decided to try adding a thin spacer (not sure if it is 1mm, I think so) to the optical train. This made a difference. Hoorah! Progress. I had a pack of two thin spacers, so thought, what the heck, let's try the second spacer too. In it went and it appeared to make the corners a little better. It isn't theoretically correct to have two spacers, but it appeared to improve the image, so I stuck with the two spacers. I then slewed to M81 and shot a new 300s sub. Here it is:

It still isn't perfect, but it is certainly much better than the original attempt this morning. I think it is at a point that I am happy with star shapes and it should give me a decent image. Unless anyone thinks any differently!

As an aside, the conditions out there this evening are awful - high humidty and huge Moon, so it was never going to be an award winning image! Great conditions for a bit of tinkering in the shed though...!

[tomato]

 Considerable progress made, and in a single session, can’t be bad!?

Was really going to persevere tonight and get some NB subs, but clouds still not playing ball....but! Tomorrow night’s forecast is better?

[vlaiv]

30 minutes ago, PhotoGav said:

Image 1 - Collimation. To me, that looked good enough to not even think about fiddling with it.

Almost, but not quite

I also measured distance in pixels - or thickness of pattern and it's ~275px on right side and ~250px on left side. This is very big defocus pattern, so I can't tell how significant that difference is, if you want to make sure, maybe a take a series of smaller defocus patterns - starting with size of inner circle and doing 2 or 3 in decreasing diameter?