Night-vision image-intensifier goggles have been a mainstay in the U.S. military's arsenal for several decades and are used extensively for a variety of ground, air, and maritime applications. Goggles are usually worn on helmet mounts or head straps, allowing the operator hands-free operation. The typical goggle structure combines three separable subcomponents: an imaging objective lens assembly; the intensifier tube assembly, which features an input photocathode integrated with a fiber-optic image inverter and an output phosphor screen; and an eyepiece assembly.
Although the real magic of the goggle lies within the details of the intensifier tube itself, the optical design of the objective lens group is by no means a trivial task. The combined requirements of a fast f/# and relatively wide field of view pose interesting problems for the aspiring designer. Vignetting, which is prevalent in all lightweight goggle objective designs, can offer an interesting degree of freedom.
Objective-lens specifications for military goggles typically permit a drop in relative illumination across the field of view to 40% of the nominal axial value. Consider a goggle objective with multiple fields of view. Each ray bundle for each position within the field of view has an associated pupil size. For optical-design purposes, we can estimate the relative illumination by comparing the area at the front aperture of the entrance pupil for the incoming ray bundles corresponding to each position within the field of view.
The size of the pupils for each field will be determined both by the vignetting values and by the aberrations of the pupil itself if the actual aperture stop surface is interior to the lens. Vignetting values as defined in most optical ray-trace software are of course related to the pupil sizes; generally, however, they are a little more simplistic in that they account for only the relative extent of the upper marginal rays within the aperture stop and thus do not account for pupil aberrations unless the stop is located in front of the powered elements.
Figure 1. An unvignetted patent lens is heavy and bulky. It also features mechanical interference in the three-element field lens.
Consider an intensifier objective-lens design derived from U.S. Patent #5,508,846, which is fairly unusual in the sense that the vignetting values are all zero across the field (see figure 1). This particular design projects a ±30° field of view onto a 25-mm-diameter image format. It operates at f/1.3. The aperture stop is located on the planar first surface of the cemented doublet; note that the rays for both the on- and off-axis bundles completely reach the edge of the stop. All the other apertures are free to "float" and are assigned by the computer such that nothing but the aperture stop limits the size of the ray bundles. The last element shown represents the presence of the flat cathode window that is actually part of the intensifier tube assembly. The glass weight of the assembly is estimated to be 54 g, which would be fairly heavy for a head-mounted system.
The relative illumination plot for the objective shows the value falls to about 73% at the edge of the field. This drop of illumination is entirely due to the optical projection of the aperture stop by the first two lenses, which gradually reduce the ray-bundle diameter as the field of view increases. Goggle objectives typically feature excellent resolution on-axis but a fairly rapid decay as the field angle increases. lightening the load
The ray trace of figure 1 is worth examination, as it reveals exactly where the ray bundles for both the on-axis field and the full off-axis field pass through each lens. Since the illumination is currently 73%, we now have an opportunity to reduce the size and weight of this objective by constricting lens apertures such that they will clip only the off-axis ray bundles, leaving the on-axis ray bundles with the maximum aperture as required.
It is clear that the apertures of lenses near the stop cannot be significantly reduced without impacting the on-axis ray bundle. Lenses near either the front or back end, however, are excellent candidates for aperture reduction since the off-axis bundles diverge from the centrally located on-axis bundle. Additionally, notice that the group of three field lenses, which features a positive bi-convex lens sandwiched between two negative lenses, experiences a slight mechanical interference at the edges. Reducing the aperture here will hopefully remove the interference problem as well as reduce volume and weight.
Ray-trace analysis shows the radial height of the upper marginal ray for the on-axis bundle at the first lens is 7.7 mm, yet the actual radius of this lens is currently extended to about 12.7 mm in order to pass the entire off-axis bundle. Therefore, we can start by reducing this aperture to some value near the on-axis ray bundle height. When we manually set the radius of the front lens to 8.0 mm, we obtain a revised result with a new estimated glass weight of about 40 g and a decrease in minimum relative illumination to about 54%. This is not enough to clear the interference of the back lens group. Fortunately sufficient relative illumination exists for another vignetting iteration.
Figure 2. Further vignetting addresses the issue of mechanical interference (top) while providing acceptable relative illumination (middle) and spatial-frequency performance (bottom).
Since the mechanical interference exists on both sides of the bi-convex positive element in the rear lens group, we can iteratively reduce its aperture until the relative illumination reaches the 40% specification. Reducing the aperture radii of both sides of this lens to 8 mm results in a design that meets that specification (see figure 2). Notice that the off-axis ray bundles no longer reach either edge of the stop surface: The lower rays are cropped by the front lens aperture, and the upper rays are cropped by the rear positive lens aperture. It is important to recall that the definition of the aperture stop is simply the location where the chief ray at the center of the bundle for each field of view crosses the optical axis. Even though vignetting limits the ray bundle pupil size in various places throughout the lens, the defined aperture stop location is unchanged on the planar surface of the cemented doublet.
The estimated glass weight for this configuration is down to about 35 g, and there is now no mechanical interference at any of the lens edges. The MTF of the vignetted design is shown in figure 2. There is, of course, no change to the on-axis response since no lens curvatures or thicknesses were changed. The off-axis MTF shows an improved tangential response but a degraded performance in the sagittal direction. This may at first seem counter-intuitive, since smaller pupil sizes generally suffer less aberration than larger ones, but in this case it turned out that the very rays we blocked turned out to be some of the best corrected portions of the full aperture. Thus, the result of the vignetting was to leave behind only the rays with the greatest contribution to the sagittal aberration. The impact is fairly minor, however, and can be improved by subsequent re-optimization of the lens curvatures. design degrees of freedom
The allowance for a drop in relative illumination across the field of view is vitally important to the image-intensifier objective-lens designer. In the example above, adjusting the vignetting alone reduced the weight by 35% and eliminated mechanical interference problems near the lens edges. Further optimization of the curvatures and thicknesses would certainly lead to further weight reductions since many of the edge thicknesses and diameter-to-thickness ratios are now much more generous than what is typically needed for manufacture at an optical shop.
Figure 3. Because software programs do not generally allow vignetting as a degree of freedom, they can sometimes produce designs with ray bundles that do not correspond to any limiting surface.
In design problems such as those for night-vision goggle objectives, the apertures of lens elements are nearly as important as their radii and thicknesses. Most optical-design software packages do not treat apertures as direct optimization variables; rather, they base them on calculations of certain ray positions and aiming commands. Thus, designing a system with controlled vignetting is often a more labor-intensive process than other approaches. The designer should be aware that the software may arbitrarily assign vignetting values to rays that do not correspond to any physical limiting surface (see figure 3). Simple examination of the ray-trace diagrams is usually sufficient to ensure that all vignetted ray bundles do indeed result from actual element or aperture edges.
Finally, it should be noted that the author chose the U.S. Patent example because its unusual lack of vignetting values made it easy to demonstrate the power of controlled vignetting in the design process. However, starting with an unvignetted, full-aperture design is often an extremely difficult approach. In most cases, the designer is probably better off starting with a few set apertures or vignetting values such that the initial form before optimization is closer in size and weight to the desired end product. Be warned, though, that optimization with vignetting and aperture parameters often requires extra vigilance and a thorough understanding of how the software performs its calculations so that the vignetting occurs at real edges. oe
John Hall is senior staff optical engineer at Optics 1 Inc., Manchester, NH.