NBS 1963A resolution target
Years ago, I purchased an USAF 1951 resolution target from China, only to find it of such a poor quality to be essentially useless. This time, I decided to get a high-quality resolution target from a reputable source, of a type suitable for visual assessment like the USAF 1951.
I settled on an NBS 1963A positive target from Thorlabs in 5 x 5 cm substrate size (nominally 2" square, or 50.8 mm square). I like this target type better than the USAF 1951 is multiple respects. The NBS 1963A, as the name indicates, is more modern, and consists of L-shaped clusters of lines, five vertical and five horizontal, separated by spaces of the same thickness. The higher number of line pairs in each cluster, together with their increased length compared to USAF 1951, make it easier to visually assess whether the lines can be regarded as resolved. Dust, fibers and scratches on the target are less likely to make a reading impossible. Additionally, the use of NBS 1963A is made much easier by each cluster being directly labeled with its number of lp/mm.
This specific target is designed to be reproduced at a 1x magnification, as labeled along the bottom in Figure 1. It can be used at different magnifications, but the actual magnification must be factored into the calculation of on-sensor resolution in lp/mm.
The NBS 1963A also helps to compute the magnification of an image of the target, as R = L' / L, where R is the magnification, L' is the length of a specific black line on the image, and L the length of the corresponding line on the target. Specifications of the target include the length of the lines, in addition to their reciprocal spacing. Alternatively, the distance between any two convenient points on the image versus target can be used.
Thorlabs makes two models of 5 x 5 cm NBS 1963A positive targets, one ranging from 1 to 18 lp/mm (i.e. its smallest line pairs are 55.6 μm wide, too coarse for my use), and a second from 1 to 228 lp/mm, which at around 225€ is only 10% more expensive, but reaches down to line pairs only 4.39 μm wide. The latter is equivalent to group 7, element 6 of a USAF 1951, as well as finer than the 10 μm of ordinary micrometer scales for microscopy. Negative patterns of both specifications are also available. Edmund additionally sells much more expensive NBS 1963A targets that reach 512 lp/mm, as well as targets of the same type built onto opal glass.
To put things in perspective, with my Zeiss Stereomicroscope III and the Thorlabs NBS 1963A target, at 40x I can visually resolve 118 lp/mm without difficulty, 203 lp/mm about as often as not, and at 228 lp/mm I can see a hint of a pattern but cannot count the individual target lines. With unaided eyes and at a normal reading distance, I can resolve without difficulty 6.3 lp/mm, and just barely 7.1 lp/mm.
The NBS-1010A is comparable to the NBS 1963A, but generally used for microcopying and microfilming. The older NBS 1952 resolution target has longer lines than the NBS 1963A, but does not seem to be commonly used at present, and is difficult or impossible to find in high-resolution versions.
Mounting the NBS 1963A
For applications in photomacrography, it is essential that the resolution target be held perpendicular to the optical axis of the lens being tested. Given the extremely small DOF in photomacrography, any deviation from orthogonality will show as one or more corners of the target being out of focus, when focusing is done on the center of the target. The target must also be uniformly illuminated with diffuse light from the rear. This is best achieved by placing two diffusers, sufficiently spaced away from each other and from the target, along the illumination path. This works significantly better than a single diffuser, or two diffusers directly stacked on each other.
I did not trust myself building a sufficiently precise mount for the target, diffusers and other components. Therefore I purchased a minimum of mechanical components from the Thorlabs 60 mm cage system (I chose metric versions of the components whenever available) necessary to build a stable and precise holder for the various optical components. This target holder costs more than the target itself, but the use of lab-grade mechanical components avoids plenty of potential problems that could make the test results questionable.
In my photomacrography lab I use a vertical system based on a Zeiss measuring microscope stand. Therefore, the holder for the target also needs to have its optical axis vertical, with the test target at its top.
I tested different LED sources, then settled for the moment on a Godox LED64 mini-panel because of its continuously adjustable intensity, reasonably good CRI, lack of flickering, built-in male and female cold shoe mounts, as well as operation both via batteries and external DC supply. The LED64 simply rests sideways on the breadboard base of the filter holder. It is a little too large for this cage, but I have several of these LED mini-panels and prefer to standardize on re-usable parts shared by multiple setups and devices. On a few tests I flanked the LED source with an electronic flash (Godox AD 200) equipped with small spiral tube and a small snoot, especially at high magnification.
I decided on a horizontal orientation of the light source for multiple reasons, including:
A 45° mirror is needed within the base of the holder to reflect incoming light from the side and direct it upwards. The mirror is a 0.2 mm thick acrylic sheet, thin enough to cut with office scissors. The mirror size is 60 x 90 mm with cut-outs for the cage rods, in order to be held in place by the rods.
Two diffusers are further necessary to make the illumination of the target as homogeneous as possible. The first diffuser in the light path is a thin plastic sheet cut from a discarded box of screws and directly attached to the self-adhesive side of the mirror. A second diffuser is made from a 2 mm thick opal-white polycarbonate sheet, a material that absorbs slightly less light than ordinary opal glass or teflon sheet. Polycarbonate sheet can be slowly cut (to avoid melting) with a metal hacksaw and filed with an ordinary file. This diffuser sits in a second slot of the target holder plate (Thorlabs part LPP05/M), a few mm below the target.
The two diffusers in the cage are more than sufficient to evenly illuminate the target even when the light source is off-center with respect to the mirror. It is entirely feasible to simultaneously illuminate the mirror with both a small LED source and an electronic flash, each offset by a few cm to either side of the mid-line.
The test target is mounted in an LCP05/M mount for 2" filters, which has two filter slots separated by a space of a few mm. The optics are held in place by nylon-covered flexible steel blades.
A lens shade can be screwed into the LCP8S plate, or simply laid on this plate, to cut out stray light. There is no metric version of the LCP8S plate, but a 52 mm lens shade or stack of extension tubes screws here for about half a turn (you may need to try one or two before finding one that fits, since some lens shades and step-up adapters seem to be slightly out of specifications). This is enough to keep it in place. If you have Thorlabs SM2 extension tubes, they are made to screw into the LCP8S, but they are probably too costly for this specific use. As an alternative, a flat black cardboard sheet with a 2" round hole or a rolled-up black paper sheet can lie down atop the LCP8S, to block any stray illumination by the LED source from entering the lens. For testing low-contrast optics especially sensitive to off-axis illumination, the hole in the cardboard sheet can be made much smaller. Ambient illumination should also be lowered while shooting test images.
There is no practical way to prevent dust from settling on the target, other than blowing the latter clean before use with a rubber blower bulb, and closing the LCP8S, extension tubes or light shade with a lens cap when not in use. Storing the whole cage assemply in a clean plastic bag or tightly closed plastic box also helps to keep dust at bay. Dust on the pattern side of the target is most likely to be visible in test images, and severe dust may cause false MTF measurements.
I considered adding an XYZ micrometric stage between the breadboard and the cage, but the added precision is not necessary at the range of magnifications suitable for this target, and such a stage would add about 5 cm to the height of the setup. My camera fixture already provides precise fine focus, and sliding the cage assembly by hand on the base of the stand is already sufficiently precise.
At higher magnification, a modified microscope base and stage may be a better choice, allowing the use of a variety of condensers for transmitted illumination and quick swapping of resolution targets mounted on standard microscope slides.
Using an NBS 1963A for MTF measurements
In spite of this target type being designed for visual evaluation, an edge of its largest elements can be used for MTF measurements, as discussed here. The edges of these elements are very straight and sharp, and this allows the target, slightly inclined diagonally at an agle of about 5° counterclockwise, to be used for MTF measurements even at magnifications that substantially exceed the lp/mm resolution of its finest patterns.
This target is adequate for visually evaluating a high-resolution lens at 1x on a 20 Mpixel Micro 4/3 sensor (which has a theoretical resolution of 115 lp/mm at 50% MTF, and a Nyquist limit of 149 lp/mm), but it is not quite enough at 2x, and is definitely out of its useful range at 4x. However, my MTF resolution tests with this target prove the latter to be fully usable for this type of testing at 4x, and probably at higher magnification.
An NBS 1963A is a suitable replacement for the perhaps more common USAF 1951, and provides advantages over the latter in practical use. An NBS 1963A target with finest pattern of 228 lp/mm is adequate for testing a high-resolution lens at 1x on 20 Mpixel Micro 4/3, but not quite at 2x. It is however usable for MTF measurements up to at least 4x.