Flat frame light box - test for even illumination

[Jammy]

I'm making a flat frame light box for use with my C9.25 SCT using the traditional method of light behind opaque perspex sheets, and set within a foam board box.

Large EL panels just feel too expensive at this moment in time.

In order to make sure I've spaced the perspex correctly before I glue everything together, is there a way of checking how even the illumination of my flat frames are?

I've seen on forums where people have posted flat frames with numbers, contours etc to show how even the illumination is, but I've no idea where these things are done.  Do I need specialist software to do this?

Thanks for any help

[symmetal]

Hi Jamie,

With the scope still focused at infinity as normal and using your usual capture software, point the scope at the centre of the light box and take an image at whatever exposure gives you a reading of around 30000 ADU at the centre of the image. Ignore any readings away from the centre.  An average of a block of pixels would be more useful, some software gives values of a 21 x 21 block or similar under the mouse cursor when moved over the image. Then keeping the exposure the same, take an image with the scope pointing at a corner of the light box and see if you again get the same reading at the image centre. Repeat for the other corners. If the readings are all the same then the light box is evenly illuminated.  

The flat frames you've seen with numbers and contours superimposed are from software like CCD Inspector which show the light fall-off characteristics of the imaging train. However to get these you need a flat frame from a source which you know is evenly illuminated. In your case, at the moment, you don't, so you wouldn't know if the fall-off characteristics were from just the scope, or a combination of the scope and flats box.

Alan

Edited February 8, 2019 by symmetal

[NGC 1502]

6 hours ago, Jammy said:

In order to make sure I've spaced the perspex correctly before I glue everything together

 

This could be a useless comment because I’m not at all familiar with what you’re constructing.  But instead of gluing, could you use removable screws instead, enabling easy dismantling for modifications ?   Or is there some reason that you must use glue  ?

Ed.

[adyj1]

4 hours ago, symmetal said:

take an image at whatever exposure gives you a reading of around 30000 ADU at the centre of the image.

I'm using a cheap light panel and would like to do this - how can the ADU be checked? 

Thanks, 

Ady 

[Oddsocks]

Hello Jamie.

As described above in Alans reply, @symmetal, the contour maps are produced by specialist software however this would not be of much use to you in this case because they map the entire optical chain of the telescope as well as the light source and with a large aperture OTA such as the C9.25 you won’t be able to move the entire OTA over the face of the light box to determine the evenness of light distribution without overlapping the edge of the light box and producing an inaccurate result.

Happily there is a simple solution using your ZWO ASI 1600 camera by converting it into a temporary photometer.

Remove the camera from the telescope and if not already fitted then attach the standard camera 2” nosepiece.

Using a piece of aluminium kitchen foil, temporarily wrap the foil over the nosepiece so that it lies flat across the nosepiece aperture. Make a 1mm diameter pin hole in the foil in the centre of the nosepiece aperture. Carefully slide the foil off the nosepiece then cut a couple of pieces of white printer paper into small disks or squares for use as a diffuser/attenuator, approx 1cm x 1cm, shape doesn’t really matter, and glue or tape the paper disks (or squares) across the pinhole on the inside of the foil as a double thickness. Reattach the foil to the nosepiece and secure with tape or rubber bands. Reinforce the foil with a couple of strips of transparent sticky tape across the face ensuring that the pin hole is roughly centred to a piece of the tape and not lying across the edge of the tape (to avoid diffraction effects).

The camera can now be used as a crude but accurate and effective photometer.

Place the camera nosepiece in close, direct contact with the surface diffuser of the flat panel, somewhere near the centre of the panel, ensure the flat panel has been switched on for at least fifteen minutes to allow it to stabilise, darken the room to avoid stray light affecting the measurements and then take an image with the camera.

Adjust the camera exposure time to achieve an average ADU for the entire image of approximately half max value (saturation) then using the same exposure time simply move the camera across the face of the flat panel in a rough grid pattern, 1cm steps would be sufficient, make a note of the camera average ADU reading at each sample grid point as you move the camera over the flat panel and then plot the resulting readings.

What you have constructed could be described as a 'spot' photometer and it's sample area is very small, dictated by the size of the pinhole in the foil.

Measurements made with a spot photometer can be adversely affected by small defects in the surface of the flat panel diffuser which, when used as a light source for the telescope would not be noticeable as the light source is so far out of focus. You could enlarge the size of the pinhole and add additional paper attenuators to the inside of the foil to increase the sample area size however you would achieve better results by converting the 'spot' photometer into an 'area' photometer by moving the surface of the foil/camera away from the face of the flat panel diffuser.

To do this simply measure the length of the camera nosepiece and cut a cardboard tube approximately 1cm longer from the inside of a roll of toilet paper or kitchen paper (or simply 'roll your own') and slide it over the nosepiece so that the foil face and pinhole is now sitting inside the tube to a depth of 1cm. Now when you hold the camera against the flat panel the pinhole being further from the panel will be sampling a larger area, depending on the geometry of the camera probably ~5mm to ~10mm diameter, which will even out the small imperfections in the flat panel diffuser and give you a more meaningful measurement.

When used as an area photometer it would be good practice to blacken the face of the aluminium foil that looks towards the flat panel to prevent over illumination due to internal reflections between the face of the flat panel diffuser and the foil, it probably won't make a lot of difference when used for this task as the inaccuracies due to the reflections will be the same across all the measurements however if you wanted to improve the accuracy of the photometer you could blacken the foil, or place a blackened cardboard aperture mask across the face of the foil, use a different material for the foil such as thick opaque black plastic etc, just depends what materials you have to hand.

Your capture program should be able to display image stats like max/min and average ADU pixel values and using the camera as a photometer in this way the average ADU would be sufficient. If your capture program does not display image stats then there are several free image processsing programs and FITS viewers available that can show this information.

A well made EL panel would typically show a variation of less than ~1% average ADU across it’s illuminated surface, a LED draughting panel rather more, up to 5% is not unusual, even for a very well made panel. If your flat panel shows a variation of less than 5% ADU across the field then it should be adequate.

HTH.

William.

 

 

Edited February 8, 2019 by Oddsocks
Added text for area photometer

[symmetal]

4 hours ago, Oddsocks said:

 and with a large aperture OTA such as the C9.25 you won’t be able to move the entire OTA over the face of the light box to determine the evenness of light distribution without overlapping the edge of the light box and producing an inaccurate result.

Oops! you're right there Oddsocks. I was initially thinking of focused spot readings when I wrote it but added the infinity focus as a normal flat procedure. Of course this will integrate the whole exposed area which isn't much use. It was 3 am when I wrote it which is my excuse. ?

How about fitting an aperture mask for the front of the scope with a small offset hole which could then cover the box corners without spilling over the edge.

Or, you could set the light box at the end of the garden and focus on the box corners in turn. 

Or, you could follow Oddsocks detailed and elegant solution. ?

5 hours ago, adyj1 said:

I'm using a cheap light panel and would like to do this - how can the ADU be checked? 

Your capture software normally allows you to move the mouse over the image and give a readout of the ADU value under the mouse. Or looking at the histogram display will give a good indication of the ADU value spread. The histogram reads from 0 to 65535 from left to right (for a 16 bit display, as most are). Also the image statistics display will give you max, min and average ADU values of the whole image.

Alan

[Oddsocks]

7 minutes ago, symmetal said:

ops! you're right there Oddsocks. I was initially thinking of focused spot readings when I wrote it but added the infinity focus as a normal flat procedure. Of course this will integrate the whole exposed area which isn't much use. It was 3 am when I wrote it which is my excuse. ?

How about fitting an aperture mask for the front of the scope with a small offset hole which could then cover the box corners without spilling over the edge.

Or, you could set the light box at the end of the garden and focus on the box corners in turn. 

Or, you could follow Oddsocks detailed and elegant solution. ?

Thank you for the kind comments Alan.

I think you can probably tell I spent too much of my youth watching Blue Peter!  ?

 

[Davey-T]

The white opaque plastic in my 10"SCT light box rests on the front of the scope, seems to work OK, I've seen similar ones.

Illuminated from the rear with four dimmable halogen bulbs

Dave

 

[Alien 13]

If you just wanted to check the unevenness of illumination as a percentage you could just take readings from various points on the panel using a Light Dependent Resistor and a multimeter set to K ohms, if you wanted you could calibrate the above with a known light source.

Alan

[Jammy]

 

@symmetal @Oddsocks

Many many thanks for the detailed solutions.  You've excelled yourselves ?

I'm going to construct the box this weekend.  Fingers crossed it will all work as planned.  I've drawn a CAD model of what it will hopefully look like.  Now to put the theory into practice!

I'll give feedback when I'm done, but only positive ?

[Jammy]

3 hours ago, Davey-T said:

The white opaque plastic in my 10"SCT light box rests on the front of the scope, seems to work OK, I've seen similar ones.

Illuminated from the rear with four dimmable halogen bulbs

Dave

Looks good Dave.  I'm hoping to make mine compatible with my Evostar ED80 too by allowing a sleeve to fit within the opening for the C9.25

I'll post a pic when I'm done ?

[Jammy]

12 hours ago, NGC 1502 said:

 

This could be a useless comment because I’m not at all familiar with what you’re constructing.  But instead of gluing, could you use removable screws instead, enabling easy dismantling for modifications ?   Or is there some reason that you must use glue  ?

Ed.

I'm going to make something like this photo (image stolen from Richard McDonald's website).  Foam board doesn't take screws very well (if at all), but can be jointed with a hot glue gun.  It's the distances between the inner baffles that I'm trying to space correctly before I fix it all together.

[Davey-T]

8 minutes ago, Jammy said:

I'm going to make something like this photo (image stolen from Richard McDonald's website).  Foam board doesn't take screws very well (if at all), but can be jointed with a hot glue gun.  It's the distances between the inner baffles that I'm trying to space correctly before I fix it all together.

Can't see the point of having that gap it must introduce shadows, and it's better to have the illumination source behind the corrector plate with baffles dispersing it evenly around the box interior.

A dimmer is indispensable.

Dave

Edited February 8, 2019 by Davey-T

[Horwig]

May I add my twopence worth to this thread.

I've built  a couple of light boxes for flats, here's the thread on the one I use on my refractor.

To measure the evenness of illumination, I used a stills camera in a darkened room, not filling the frame with the lightbox, so as not to get affected by vignetting on the lens. Then stretch the contrast on the image in photoshop, a good visual way of doing it, but not as accurate as Oddsocks method above.

Just my opinions?, naturally there are more ways than one of skinning this feline.

 

Huw

[IanL]

In a truly dark room, or at night in situ if you have no nearby light sources and little light pollution to leak in, fit the box to the scope focused to infinity. Take a flat, rotate the box 90 degrees, take a flat, and rotate/expose twice more.

Calibrate the first flat with the second (put the first flat in your software as if it was a light frame and calibrate with the second) Then stretch the result as much as possible. If the flat box is evenly illuminated, being the only variable, you should see no gradients or vignetting as the second flat will have corrected just the uneven illumination from the optical train and uneven pixel response.

Then flatten the original (unflattened) first with the third and review, etc.

You may also want to take median ADUs of 25x25 pixel areas on the unstretched flattened flats.

Any gradients mean the box is not evenly illuminated. Using the median measurements you can calculate the percentage difference across the field and decide if it is acceptable.

Edited February 10, 2019 by IanL

[Jammy]

Thanks for all the replies guys.

I managed to make the flat frames light box at the weekend.  I used @Oddsocks method for testing the light distribution, which really helped spacing the perspex sheets.  I also needed to add some inner foam board to diffuse the light because of a slight light drop off at the very corners of the box.  I probably wouldn't have spotted this without the photometer, to the naked eye it did look very even.  I also did a stretch in photoshop as described by @Horwig.  That idea came to me after I'd glued it all together, and before I read your reply!

I've sized the front circle large enough to fit over the front of the C9.25, and I've also made an insert that will allow me to use it with an Evostar 80ED.

I'm now waiting for a dimmer switch to arrive that I found on eBay, then I'll be ready to start taking flats.

First time I've used foam board, surprised just how strong it is ?

[symmetal]

Very neat job Jammy 

Alan

[Oddsocks]

Glad that the DIY ‘photometer’ method worked out for you, looks like a good build!

I’m not a great fan of dimmer controls for LED panels, most of the models at the lower price end of the market use PWM (Pulse Width Modulation) to vary the light output and at very low light levels used in flat boxes this means the LEDs are only lit for very short periods in each clock cycle, in effect they become high frequency strobe lights.

PWM controllers are ok for tungsten filament / Halogen lamps because the filament does not react quickly enough to the strobing for it to affect the quality of the captured flat frame but with LEDs they do switch on/off very quickly and although your eye does not see this strobing the camera can.

LED strobing can manifest itself as horizontal, vertical or diagonal bars of various thickness in the flat frame.

It can not be predicted which cameras and sensors will be affected, it depends on things such as whether the camera has a mechanical shutter or uses a global or rolling electronic shutter, exposure time, frequency of the PWM brightness control and it’s pulse-to-space ratio etc.

For LED lights in a flats source a linear brightness controller is best but these are more expensive, generally bulkier and heavier, inefficient and run quite hot.

To avoid strobing artefacts run the LEDs at full brightness, preferably with no brightness control at all, and use several layers of neutral density photographic gel filter to dim the output of the flats source.

Linked below is a UK source of neutral density gel filters, 1 sheet, 1.22m x 0.53m @ £7.00 would be enough to give you three or four thicknesses of filtration, each layer of 1.2 density reducing the light by 4 stops.

https://www.sblite.co.uk/neutral-density/265-409-rosco-299-12-neutral-density.html#/23-sheet_roll-sheet

These gel filters are surface coated polyester and though tough the surface coating can be scratched so its best to place them behind a sheet of clear acrylic if you intend to place the light box over the OTA aperture so that there is some contact beween the output diffuser and the front of the telescope.

Technical specs for the linked gel filters supplied by sblite of sheffield are found at the manufacturers web site, Rosco, (e-colour+ series) here:

https://emea.rosco.com/en/product/e-colour

William.

[symmetal]

I use Rosco filters on my Huion LED flats panel just like Oddsocks though I use 1 or 2 sheets of 0.9 ND (3 stops) depending on the scope. 

Alan

[Jammy]

Thanks guys, I'll take a look.

You've been really helpful ?

[Graeme1858]

Hello

I put a cardboard and duct tape flat frames light box together recently. After reading this very informative thread I can see that my effort is definately a mark 1.

Foam board is the way to go!

Regards

Graeme

[Jammy]

On 20/02/2020 at 21:34, Graeme1858 said:

Hello

I put a cardboard and duct tape flat frames light box together recently. After reading this very informative thread I can see that my effort is definately a mark 1.

Foam board is the way to go!

Regards

Graeme

Foam board is surprisingly strong.  Still using my light box and getting good flats.  They sell foam board and hot glue guns in 'The Range' if you have one near you.