Last summer, Eric (CNer Longbond) and I became acquainted while trying to solve some NV adapter problems and we have been corresponding ever since, mainly concerning H-a filtration with NV. Last fall, he started using an unmounted, 24mm diameter filter inside his Mod 3C with C-mount lenses. Thus began a study of how REAR mounted filters, those which are mounted behind the optical system, effect the image that is ultimately presented at the NVD ocular.
Front mounted H-a filtering (also called first surface) is quite rare when used with a telescope. It is limited by the number of lenses or optics that can accommodate 1.25” and 2” filters without causing significant mechanical vignetting (a smaller filter covering a larger aperture). The Envis lens, and shorter focal length C-mount or camera lenses and just a few astro lenses like the Askar 180, can accommodate front mounted 2” filters. Custom filters in larger sizes become exorbitantly expensive.
We know that H-a images we see or photograph through our NVDs are sometimes effected by band shift, the optical aberration which causes a loss of filter performance (H-a contrast) at the EoF. Band shift is caused when light rays passing through an H-a filter encounter steeper angles of incidence as they pass through steeper angles of glass at the edge of a lens. The faster the optical focal ratio, the steeper the angle of incidence, resulting in the filtering process being pushed out of phase at the EoF.
We also know that band shift severity is effected by the pass band width of the filter and the focal ratio of the optical system. But a question arose in our communications… how does filter placement in the optical train effect band shift and filter performance? Eric and I both wanted to know how filter performance was effected when the filter was placed after the optics, but in front of the NVD, in what we call a rear-mounted position… at what point do rear mounted filters become most effective in use. We started testing in January.
Test #1: This is a comparison photo using a 50mm Nikon camera lens at f:1.4 with a 7nm filter of the Rosette Nebula from my light polluted, red-zone home. The image on the left was taken with the filter rear-mounted (behind the lens) while the image on the right was taken with the filter front-mounted (in front of the lens). This photo presents the same filter in both mounting positions with exactly the same camera settings of ISO 100, 1s exp, 10s average in NightCap. On axis, the front mounted filter is significantly superior to a rear mounted filter, as was expected.
The same thing happened with the 3.5nm filter, taken the same night at ISO 250, 1s exp, 10s ave. Again, the filter was rear-mounted for the left photo; the right photo was front mounted. It is obvious that front-mounted filters provide better H-a contrast when the subject is on-axis.
Now look at both comparison photos. The front-mounted filter images (on the right in both comparisons) look correct, with the 3.5nm filter showing more H-a contrast than the 7nm. But the rear-mounted images (on the left) show a tiny bit better contrast with the 7nm filter. The 3.5nm filter performed worse than the 7nm filter when rear-mounted with this optical system.
Test #2: In the next photo, the 7nm filter was rear-mounted on the 50mm lens at f:1.4 and shows the Rosette centered and at the very edge of field, with no significant loss of contrast from band shift, ISO 250, 1s/10s average. So although H-a contrast is lower with a rear-mounted filter, there is little or no loss of the H-a subject near the EoF from band shift. When I performed this test with the 3.5nm rear mounted, results were the same; no EoF drop-off in H-a contrast/brightness.
Test #3: The next night I repeated the first two tests but used a 300mm Nikon at f:2.8. The difference was less EoF darkening (alternately referred to as “spotlight effect,” or “vignetting”) using the slower optic. But using the 3.5nm filter and the slower optic with the rear-mounted 3.5nm filter, did show the H-a object better, with more contrast. This photo compares the contrast between the 3.5nm (on the left) and the 7nm filter (right) at f:2.8:
Using the 50mm lens again, I took two photos of the NAN/Gamma Cygni complex using a rear-mounted 3.5nm filter, the first image at f:1.4, ISO 500, 1s exp, 15s average, and the second at f:2.8, ISO 800 to compensate for the slower focal ratio, to show the significant improvement in H-a contrast using the slower focal ratio with the same rear-mounted filter:
The results of these tests show that both front and rear-mounted narrowband H-a filters present disadvantages. Front-mounted (first surface) filters show H-a with greater contrast on-axis, but create significant EoF darkening with a substantial loss of contrast when the H-a subject approaches the EoF, where H-a filtering is pushed out of phase with fast optics. Rear-mounted filters reveal H-a all the way to the EoF, but reveal the H-a subject with less overall contrast. Concerning this issue, Eric wrote: “With front-mounted filters, band shift is EXPLICITLY localized because the angle of incidence is dependent on the object’s location relative to the optical axis. However, rear-mounted filters are affected by the angle of incidence from the light cone. Even an object in the center of the field gets MOST of its light from the steep part of the light cone. If the edge of the light cone becomes too steep, you lose s/n.” The effects of band shift are still present with rear mounted filters, but instead of a high contrast image at the center FoV, the filter spreads the effects of EoF contrast loss across the entire FoV.
Since most of our NV filtering places the filter behind the optics, you may find that a narrower band filter performs no better than a wider pass band filter, because the optical system may be TOO FAST. Generally, the shape of the light cone in faster optics creates limits for the application of H-a filters. The wider the pass band, the less concern there is about phase shift; the narrower the filter pass band, the more you may need to pay attention to the optical focal ratio. In addition, the relative aperture of the optical system, as compared to the sensor size, seems to have a bearing on filter performance.
Thankfully, most telescope systems are slower and handle rear mounted H-a filtering very well with little or no EoF H-a drop-off. I am content with the performance of my Newt at f:2.8, with a 7nm filter. When I use the Newt at f:4, I may choose the 3.5nm filter. But when using a faster optic in prime focus, like an f:1.4 or f:2 lens, a narrower 3.5nm filter performs no better (or even worse) than a 7nm filter.
This study did not involve afocal testing with a rear-mounted filter, as I do not have a 24mm filter to use inside the C-mount of my Mod 3C. But my limited experience with afocal has revealed that significant EoF darkening is usually present when a narrow band filter (such as a 3.5nm) is mounted directly to my TV 55/67 eyepiece in my 8" Newt at f:2.8.
Understanding how filter placement in the optical system effects performance may help those who are considering the purchase of a narrower band (~3nm) filter. If the optical system is too fast, a 3nm filter may not perform well. On the other hand, if the optical system is too slow, increased scintillation/noise will result. Between f:2.8 and f:5.6 seems to be a sweet spot for these very narrow filters when used rear-mounted. When used front-mounted, the fastest optical systems do provide best H-a contrast on-axis, with the caveat that a bigger portion of the FoV will be pushed out of phase than when using a 5-7nm filter. .
Much thanks to Eric for his help with this study and in editing this material. This effort was a true collaboration.
Edited by GeezerGazer, 10 June 2021 - 07:06 PM.