Light Box

There are several methods for making flat fields including sky flats and 'dome' flats, some people just use a T-shirt over the end of their telescope. These methods can be tricky to get consistent results so many people favour building a light box. With a light box you are guaranteed a consistent light source that is completely under your control, but why bother? One object of image processing is to tease out faint details by various algorithms that 'stretch' the histogram. When the field is unevenly illuminated and not flat corrected the stretch exaggerates the uneven illumination and is impossible to get rid of later. If the image has been calibrated with a flat field first you can be sure you will only enhance the object features and leave the rest of the field evenly illuminated. Flats also correct for dust 'donuts' on the sensor or filters. 

This light box is version 3 and finally I'm completely happy with the design. I worked from similar designs found on the Internet and in the AIP4WIN handbook. The basic idea is that the light source illuminates a white surface, the light scatters in all directions from this surface, some of it back down through a piece of diffuser 'milk plastic' down the optical path of the telescope. The scattering method produces a much more even illumination than direct illumination even through two layers of milk plastic. I studied the paper by Simon Tulloch of the RGO before building my original version. This paper describes the construction of an evenly illuminated source for CCD testing but the variables to consider in order to achieve even illumination across the full diameter of the field are the same for a light box.

My original version was cylindrical in shape and used thick card to form the cylinder with mains powered low wattage lamps. This version turned out to be surprisingly heavy which became a problem when the telescope was at a low angle. So I wanted the new design to be as lightweight as possible and use battery power instead to avoid dangling cables.

Most of the designs I examined used 'foam core board' for construction. For my 14" reflector the amount of board I needed became quite expensive and so I opted to use a cheap pack of polystyrene ceiling tiles instead. The polystyrene tiles are extremely friable and so I papered over the outside edges and corners with some brown paper and PVA glue. Once dried this becomes very strong, to further protect the surface I painted over the whole of the outside of the box with gloss paint with the inside painted with white matt emulsion. The tiles were glued together with PVA glue and the back end of the box left just as a tight fit to allow access to the light source. The milk plastic sheet was sandwiched in place between some pieces of tile and the edges sealed with translucent silicone sealant to help hold it in place. The sheet was recessed by about 5cm and a hole cut in the outer polystyrene tile so that the box would hold itself in place on the end of the tube.

I initially used four torch bulbs powered by a 9V PP3 battery. This worked reasonably well but used quite a lot of battery power. The main problem though was the wavelength of the light. Incandescent lamps emit much of their light toward the red end of the spectrum so flats taken through blue filters needed excessively long exposures, up to 30s in some cases. I changed the torch bulbs for white 'jumbo' LED's. These give a much better spread of light across the spectrum and consume much less power as well. I wired in a 10K Ohm variable resistor with the LED's to adjust the light level when using 1x1 or 2x2 binned exposures. Each LED was placed in the four corners of the box adjacent to the milk plastic. Small white card screens were glued across the diagonal so that no direct light entered the milk plastic screen.


I typically find I can get good flats at 25-30K ADU (around half full-well capacity) on my Artemis 285 with 2-5s exposures. If I am using the 13nm Ha filter the exposures need to be around 10-15s to get the right exposure. I'll often take the flats at the end of an imaging session with the telescope tube moved to an upright position. I have noticed though that the flats aren't as good when taken this way, they probably have a slight mis-match to the image because of flexure in the optical train as the tube is moved. This causes more gradients on the final calibrated image. The best results are when the light box is used with the telescope left in the same position as used for the imaging session.