Field flattener over or under correcting?

[Magnus_e]

Hi

My DIY observatory is soon done, and I must tackle a issue that I have been putting off.

From last winter I know that my FF is under (or most likely) over correcting.

The FF does not come with specs on how far to mount it from the sensor. All I got to go by is this product review https://www.firstlightoptics.com/reducersflatteners/skywatcher-field-flattener.html#product_tabs-3 stating it should be mounted 52mm from sensor.

So how to do that? Some things comes to mind. Is 52mm the distance between the FF and camera bothy, the distance between th FF and the camera sensor or th distance between the FF outermost optical lens and the sensor. Also, as I have astromoded my Canon the optical distance has changed in removing a filter. How to compensate for that?

The T ring i have is one of the thicker ones, and if the FF is over correcting, does that mean that I should get a thinner one. And perhaps this spacer kit? https://www.firstlightoptics.com/adapters/baader-t2-delrin-spacer-ring-set.html

To show the issue, here is a stretched master flat. I see the same pattern on any data I stretch with the current setup.

 

And a ~4h30m image with decent guiding. (Not much of an image, but I'm reprocessing it to learn some PI )

It's clearly visible that there is quite a lot of field curvation?

 

Any help / tips greatly appreciated

Magnus

[steppenwolf]

I would be surprised if the spacing requirement was 52mm. 55mm to 56mm would be more likely, this being the distance from the mounting face of the rear of the reducer/flattener to the sensor front face. This is made up by the camera lens mounting flange to sensor distance of nominally 44.0mm to 45.0mm (the Canon EOS is unusually short at 44.0mm) plus the depth of a standard 'T' to bayonet adaptor which is nominally 10mm -  45mm + 10mm = 55mm

[Magnus_e]

4 minutes ago, steppenwolf said:

I would be surprised if the spacing requirement was 52mm. 55mm to 56mm would be more likely, this being the distance from the mounting face of the rear of the reducer/flattener to the sensor front face. This is made up by the camera lens mounting flange to sensor distance of nominally 44.0mm to 45.0mm (the Canon EOS is unusually short at 44.0mm) plus the depth of a standard 'T' to bayonet adaptor which is nominally 10mm -  45mm + 10mm = 55mm

Thanks for the clarification!

The T-ring I have is the Omegon T2-Ring http://www.omegon.eu/t2-rings/omegon-t2-ring-canon-eos/p,2416#tab_bar_1_select

It's not stated the thickness of it, so I'll have to measure.

By looking at the flat. Would you say it's likely to be overcorrecting? It does look like there is some reversed vignetting?

[Magnus_e]

This is how a single 660sec stretched sub looks. The stars is pointing to the center from all edges I think?

Or is it guiding?

[steppenwolf]

I'm not sure that the flat is going to tell us anything other than the amount of vignetting there is in the light cone.

The image with the two galaxies shows some field curvature but nothing excessive:-

[Magnus_e]

Ok, thanks.

Not to much to worry about then (I somehow tought vignetting and field curvation was connected.)

Just have to take better flats to get all the dust bunnies!

What tool are you using to get those readings? It would be nice to test it against a older sub, before the FF.

[Magnus_e]

This is a 200sec sub before the FF...

[nandopg]

One method that I use to check the distance of flatteners is to measure with a scale the flattener's focal point.  To do this, turn on a lamp, put a piece of white paper underneath the flattner and suspend it until you have a point projected on the paper.  The distance from the lens and the paper is the optimum distance from the lens of the flattener and the sensor. With this distance in hands add all the spacers, OAG, filter width, etc...  Complete the distance measured to reach the optimum distance with spacers or if the optimum distance is lower than  the distance to sensor with everything add up, optimize the distance measured, by buying narrower adaptors slim OAG, etc...

This method works very well and sometime presents some surprises.

Fernando

[Magnus_e]

30 minutes ago, nandopg said:

One method that I use to check the distance of flatteners is to measure with a scale the flattener's focal point.  To do this, turn on a lamp, put a piece of white paper underneath the flattner and suspend it until you have a point projected on the paper.  The distance from the lens and the paper is the optimum distance from the lens of the flattener and the sensor. With this distance in hands add all the spacers, OAG, filter width, etc...  Complete the distance measured to reach the optimum distance with spacers or if the optimum distance is lower than  the distance to sensor with everything add up, optimize the distance measured, by buying narrower adaptors slim OAG, etc...

This method works very well and sometime presents some surprises.

Fernando

Thanks Fernando. This sounds like a top tip

An optical way to solve a optical problem!

[steppenwolf]

Quote

What tool are you using to get those readings? It would be nice to test it against a older sub, before the FF.

Here you are - please ignore the FWHM figures as I have no data on your sampling rate for either image! The two images have very different profiles so the fine detail can be ignored but clearly the field flattener is doing a good job!! The software I am using is CCD Inspector.

[steppenwolf]

Quote

This method works very well and sometime presents some surprises.

I'll bet it does too!

[Magnus_e]

2 minutes ago, steppenwolf said:

Here you are - please ignore the FWHM figures as I have no data on your sampling rate for either image! The two images have very different profiles so the fine detail can be ignored but clearly the field flattener is doing a good job!! The software I am using is CCD Inspector.

That's quite a remarkable difference

From 60% to 7.3% curvature with a (budget) FF @ £75

Happy with that upgrade!

I also see in the first measurement that tilt X is -0.0 and tilt Y is 0.1. I guess that does mean that I did not mess up the astromodding, ending up with a slanted sensor

[Uranium235]

After a bit of pixel peeping there are a couple of noteable things wrong:

1) There is elongation/distortion across the whole image.

2) The right side is far worse than the left - getting increasingly worse the further you progress across the image, which leads me to think that either an adaptor or fitting is not secure/sqaurely fitting, or the focuser is tilted slightly.

Also, Steve suggested that a spacing of 52mm is too short - I would concur with that. So I think you should increase it by 3mm then try again - this time checking your imaging train is solidly mounted. The easiest way to achieve the spacing will be to use a standard T2-EOS bayonet adaptor like this:

https://www.firstlightoptics.com/adapters/t-rings.html

[Magnus_e]

Hmm, about 2). The FF has the less favourable 'push fit'.

I have seen this in a thread 6months ago, or so. The DIY fix was using a lathe to remove the 'push fit', and make it a nice 90deg fit.

I would suspect this is what's making it mount at an angle.

If the distance is too short, then a couple of these delrin spacer would do the trick https://www.firstlightoptics.com/adapters/baader-t2-delrin-spacer-ring-set.html

[Oddsocks]

On 21/10/2016 at 16:59, Magnus_e said:

I see the same pattern on any data I stretch with the current setup.

 

On 21/10/2016 at 17:39, Magnus_e said:

(I somehow tought vignetting and field curvation was connected.)

Just have to take better flats to get all the dust bunnies!

To get back to your original question regarding over or under correction of vignetting, your flats are over correcting, leading to a darkening centre and brightening corners of the corrected image.

In this, vignetting correction is not dependent of flattener spacing or alignment, as long as the flattener spacing and alignment is the same for the light frame and the flat frame then applying a correctly exposed flat will calibrate out the vignetting field from the light frame no matter how far out the flattener spacing may be.

Looking at your sample flat there is clearly a multi-colour gradient present, green to the left and becoming pinker to the right. Application of a flat is not dependent on colour since the calibration procedure works with the undebayered monochrome image but this is only accurate as long as the flat is a uniform colour and yours show a multi-colour gradient so the issue to address is how do you take your flats?

These are DSLR images and I see from your signature that you use mostly small refractors and camera lenses so I suspect you may be using an iPad or laptop screen as the light source for your flats?

If this is the case you may be falling into the trap of using the laptop or iPad screen undiffused?

The light from a backlit TFT screen is polarised light and it interacts with the antialiasing filter and micro lenses on a CMOS DSLR sensor to produce a non uniform response at the sensor. To overcome this you need to add a simple paper diffuser in front of the screen. One or two sheets of plain white 80gsm to 90gsm printer paper placed over the source screen will diffuse and depolarise the light. Make a master flat by taking at least ten flat frames and moving/rotating the laptop/iPad/paper etc a little between each exposure. When you examine the master flat debayered it should not have a multi-colour gradient, just a single-colour density gradient.

If the multi-colour gradient remains then something else is going on and a set of pre-dawn or after-sunset sky flats should produce a perfect master flat but you do need to be quick when taking them, try to take twenty or so sky flats in not more than ten minutes to avoid introducing linear sky gradients into the flats.

William.

[Magnus_e]

HI William, and thanks for the clarification to the vignetting vs curvation.

The flat I posted is probably not the best representation of color. I have been taking sky flats, but I allways seem under or over expose So the flat posted is stretched and clipped to show the vignetting. I have the same vignetting on all data I stretch a lot so in that sense it's a good representation.

My master flat however is very flat in it's natural form, but too bright to be a good flat? Many of my other master flats have ended up almost black and useless, so I'm going the right direction.

Unstretched iso 800 master flat, calibrated with master dark and bias in PI.

[Oddsocks]

Examining the flat from the last post the pixel values are too low.

When the flat is used for calibrating the image only the difference between the brightest and darkest pixel values in the flat are used to generate the correction that is applied to the image. It doesn't matter if the darkest pixel value in the flat is 500 and the brightest 1400 or say 5000 for the darkest and 5900 for the brightest, the end result would be the same as the difference between the darkest and brightest remains 900 and the relationship between the darkest and brightest pixels in the flat is linear, it should not change overall as the exposure time is changed because usually the exposure time for flats is too short for dark current to become a problem. But it is important that the darkest pixels in the flat frame (ignoring cut-off corners due to image circle clipping) are above the black level (noise floor) of the camera and the brightest pixels do not reach the saturation point.

The problem with your flat is the minimum pixel values in the red channel are the same as the black value, normally you would measure a dark frame to determine the black value but you have a handy sensor dark spot on the right side of the frame and measuring that with the cursor in PI gives a black reading of R:0.12, G:0.15 and B:0.27. Measuring the dark bands in the right and left sides of the image gives R:0.12, G:0.20 and B:0.74. Measuring in the centre of the image we only measure R:0.13, G:0.16 and B:0.29, all of which are too low causing the noise floor to become part of the flat and resulting in over correction when the flat is applied to your images.

When creating a flat aim for the average pixel value in the centre of the frame to be around half the saturation value of the detector, so if you take a flat and set an exposure time to saturate the image, measure the centre of the raw debayered flat image with the pixel cursor in PI to know what the maximum pixel value possible is for your camera, (it should be ~1.000 in P.I.) then divide by two and adjust the exposure time until measuring the flat frame with the pixel cursor shows the image centre is close to that 50% value. To confirm if the flat is correctly exposed it is best to now take a dark and record the black level pixel value for each colour channel anywhere in the image. Now load the flat and measure in the darkest corners (but avoiding any image circle cut-off), the pixel values for each of the RGB channels must read higher in the vignetted areas than the dark value by at least 0.1 

Adjusting the stretch of your posted image the thing that strikes is the odd dark shading bands along the sides of the image and the vertical noise right through the image and this is apparent because the image is underexposed and has been stretched right down to the noise floor of the sensor.

Below is a comparison of a single uncalibrated sky flat taken with my Nikon and your posted Canon image showing the pixel values across both images, the pixel values in the Nikon NEF frame, measured with the cursor in PI, are those you should be aiming for in your Canon flats. The image is not an exact measurement though, it is only to demonstrate what you should look for since to be a true comparison you would need to compare a single raw Canon flat with a single Raw Nikon flat but I hope you get the idea.

 

 

 

William.