Jeffery Richards:
John, I can't believe I'm going to say this , but I have to disagree here. Most (all?) filters have the anti-reflection coating on one side and that side needs to face the sensor to help prevent/minimize internal reflections causing issues such as this. Antlia clearly states this on their website.
Jeff,
I am totally good with disagreement! Heck, I'm not always right and it's a good opportunity to learn something. However, in this case, I'm going to do the calculation to show you why the folks at Antila are saying that and why it makes little practice sense--in spite of the widespread notion that filter orientation is critical.
Let's look at the amount of light that gets dumped into the stray reflections. Here we are going to only consider thin film filters but the calculation holds true for absorption filters as well. In the first simple case shown below, I've demonstrated how you compute the total amount of light in the stray reflections assuming a perfect AR coating on the filter. If you think a little about this, the conclusion that the stay light will be the same no matter how the filter is oriented becomes obvious. The AR coated surface simply has no effect at all.
So, let's be more realistic and use a filter with an imperfect AR coating on the surface opposite to the thin film stack that forms the filter. Here, I've picked different AR values for each surface just to make sure that it doesn't seem like I'm doing anything tricky--and to make it easier to follow the numbers. We again assume that there are no absorption losses so that the reflectivity and the transmission always add to 100%. You should check my work to make sure that I didn't miss anything but if you add them all up in either case, you get exactly the same result of a stray beam that contains 0.244% of the incident beam.
This means that as far as the
amount of stray light is concerned, it doesn't matter which way the filters are oriented. (This is really a consequence of the principle of conservation of energy.) However, one of the things that we might want to worry about is the overall irradiance of the stray light incident on the sensor and in that case, it might make a VERY small difference which way you orient the filter--particularly with a faster optical system using thick filters. When the filter side is closer to the sensor, you can see that it's contribution to the total stray light is a little bigger and since it's closer, the resulting defocused stray beam will be a little bit smaller, which will produce a slightly higher irradiance on the sensor than if the filter is reversed. My sense is that for systems slower than around F/4, with 3 mm thick filters and more than ~10 mm between the filter and sensor, this is likely to be a VERY minor effect and if you can actually see the stray, the real problem will be with the AR coatings; not with the way the filter is oriented. Flipping the filter is unlikely to reduce the irradiance enough to make the problem go away. With really fast systems, it's better to use very thin filters both to reduce this problem but also to minimize the amount of SA that gets introduced. Spacing the filter further away from the sensor will also help to spread the stray over a larger area as well. So, for
most practical purposes, the orientation of the filter
will not matter.
John
PS. Before anyone jumps all over me, I should mention that I've only considered the first bounce for the stray light. Of course you can look at 2nd, 3rd, and so on more bounces to add up the total stray light but the numbers get really small, really fast so that's a pointless exercise.
, but I have to disagree here. Most (all?) filters have the anti-reflection coating on one side and that side needs to face the sensor to help prevent/minimize internal reflections causing issues such as this. Antlia clearly states this on their website. 