Scanner Nikkor 40 mm
This page discusses the lens of the Nikon Super Coolscan 4000 ED / 5000 ED / IV / V films scanners. This discussion is part of a small set of pages on scanners and scanner lenses.
The scanner Nikkor 40 mm is housed in a small metal cylindrical barrel with a deep machined groove offset from its center. A dab of white paint marks the end facing toward the scanner sensor. There are no calibration marks or other markings, no mounting threads, and no filter mounts.
The lens of the Coolscan 4000 (lenses used in the Coolscan 5000 and IV are supposedly identical) has a focal length of roughly 40 mm and a speed estimated between f/2 and f/2.2. It has no variable aperture. In the following discussion, I refer to this lens as scanner Nikkor 40 mm f/2. The optical scheme of this lens has been discussed at length by Marco Cavina.
I obtained my specimen of this lens already extracted from the scanner. Therefore, I don't have the scanner sensor. Based on a picture of the sensor provided by Robert O'Toole, the active region of the sensor is 32.0 mm long. Taking the standard width of a 24 by 36 mm frame as 24 mm yields 1.33x on-sensor magnification with le lens as used in the scanner, or a 0.75x magnification with the lens reversed.
Extracting the lens from the scanner
Instructions for the disassembly of the Coolscan 4000 and similar models are available from multiple Internet sources, for instance on shtengel.com. As usual in many of these scanners, the lens is placed horizontally at the bottom of the scanner chassis, behind a 45° mirror.
Mounting the lens
Both ends of the lens barrel fit snugly within C extension rings in my possession, and the end of the barrel sits against a flat ledge in the rings. This prevents any significant misalignment or decentering of the lens.
C rings from different manufacturers may have different sizes and tolerances. This includes not only the internal diameter of the tubes, but even the diameter and pitch of the threaded mount. A few of my C extension rings, purchased from different manufacturers from China on eBay, fail to screw into properly sized C mounts or on properly sized C lenses. Test the different pieces of setup together to verify their reciprocal compatibility, before epoxying extension rings to the lens barrel.
The C male thread at the end of the ring slightly decreases the working distance, but at the same time provides a shallow lens shade that helps to avoid direct illumination of the front element. At magnifications normal for this lens, there is plenty of working distance left. The C extension ring does not cause vignetting on Micro 4/3 sensors, but you need to test before epoxying the extension tubes in place, if you plan to use larger sensors.
A little epoxy keeps the C extension tubes in place. Rear caps for C lenses can be used to protect the front and rear elements when the lens is not in use.
A reversible way of mounting this lens could use a slightly longer C extension tube, modified by drilling and tapping three M3 holes in the sides of the tube. Thumbscrews or headless screws can be used to hold the lens in place.
Adaptors to mount C lenses on threaded mounts of larger diameter are not common, but at the time of writing there are adapters on eBay with a C female mount and a male M42 x 1 thread. These adapters do not have a flange to stop the adapter when screwed into a female M42 socket, but they can be locked in place, for example, with another M42 adapter. These flangeless adapters are actually made for mounting at the front of an M42 variable diaphragm. This type of diaphragm allows the cone of illumination of the lens to be restricted in order not to cause flare. It is not a substitute for the missing lens aperture, because it causes vignetting if closed too much.
The lens in theory
It makes sense to use this lens in its original orientation in photomacrography above 1x. 1x-1.5x is a reasonable magnification range for testing the lens in this orientation. For magnifications below 1x, the lens should be reversed, and in this case it is worth testing between 0.5x and 1x.
The scanner Nikkor 40 mm is designed to provide an image circle (in the original lens orientation) covering the length of the active area of the sensor (see above). Therefore, this lens, at least according to its specifications, does not provide a sufficient coverage on full frame, which requires an image circle of at least 43.3 mm.
The lens should cover Micro 4/3 (sensor diagonal 21.6 mm) and APS-C (28.8 mm sensor diagonal) at 1.33x. APS-C coverage at 1x may be marginal. I do not expect in to cover full-frame.
The specified lens resolution of 4,000 dpi applies to the subject side with the lens in its original orientation and at the magnification of 1.33x. Resolution should therefore be around 5,320 dpi on the image side, equivalent to 2,660 lppi (line pairs per inch) or 106.4 lp/mm (line pairs per mm).
A 20 Mpixel Micro 4/3 sensor (17.4 x 13 mm, e.g., the Olympus E-M1 Mark II) has a resolution of 3,026 lppi, or 119 lp/mm, so this sensor slightly outresolves the lens. Nonetheless, the difference is small. The lens resolution is instead no match for the sensor in enhanced resolution mode (50 Mpixel).
The Sony Alpha 7R II and its successors have full-frame (35.9 x 24 mm) 42.5 Mpixel sensors with a 3/2 aspect ratio. This translates to a resolution of 2,250 lppi, and the lens has therefore a distinct margin of advantage. It remains to be seen, however, whether this lens can be used on full frame because of its relatively small image circle.
Based on an image circle of 32 mm, the minimum necessary to cover the original scanner sensor, it does not seem feasible to use a focal length reducer (e.g., the Metabones Speed Booster series) to make this lens provide a lower magnification, even on a Micro 4/3 sensor. Doing so would most likely cause vignetting on this format.
It might be possible to use a focal length multiplier of good quality to cover a larger sensor. However, even the best 2x focal length multiplier would most likely degrade the image too much. A 1.4x teleconverter may be worth testing on Micro 4/3 and APS-C, but it is unsuitable on full frame.
Initial tests with a 1951 USAF target made in China showed that the target quality is not quite good enough for testing this lens. For this reason, I used for the test another resolution target, a small (about 20 by 20 mm) legacy glass target probably designed for optical alignment of equipment used in the semiconductor industry. Note that the pattern shown in the following figures is not a mm reticle (The reticle squares are smaller than 1 mm).
Results are excellent, with only a slight degradation near the corners. The left side of the target in the above picture is slightly out of focus because of a minute misalignment of the stage. When photographing flat subjects without focus stacking, the low DOF requires the subject to be rigorously perpendicular to the lens axis, and the whole setup must be aligned with a similar precision.
The slight degradation in the periphery at 1x is not a surprise, given that the image circle is only moderately larger than the Micro 4/3 sensor, and that this lens is being used at a lower magnification than the design optimal (1.33x, as used in the original scanner). The fact that there is already a slight degradation and darkening near the corners, on the other hand, suggests that this lens may show a more visible degradation in the corners of an APS-C sensor, and will probably display an unacceptable image quality in the corners of a full frame sensor.
The very sharp character outlines and line edges in the center suggest that the effective resolution in the center of the image circle is higher than the rated 4,000 dpi (the sensor provides a real resolution of over 6,000 dpi, and therefore outresolves the specifications of the lens). At this level of resolution, focusing must be very precise, and the precision provided by the maximum zoom on the LCD rear screen may be a limiting factor in the focusing precision. An external HD screen may be necessary to focus accurately. As mentioned above, subject and setup alignments are also critical.
There is no visible distortion on Micro 4/3 at 1x.
The magnification allowed by the helicoid shown in the above picture ranges from 0.75 x to 1.4 x. At 0.75x, there is a moderate but visible darkening in the periphery, as well as a fuzziness in the periphery caused mostly by curvature of field. This degradation is visible even in the reduced frame in the above figure. Image quality in the center remains instead excellent.
Image quality remains instead very good across the frame at 1.4x. In fact, there is no hint of darkening in the corners, and resolution in the corners is arguably better at 1.4x than at 1x. Without doubt, at this magnification there is less variation between center and corners than at 1x. This results fits the expectations, since this magnification is very close to the design optimum (1.33x).
Working distance at 1.4x is about 57 mm without the front C extension tube. Together with the narrow lens front, this makes it very comfortable to arrange a proper illumination around the subject.
Reversed on the same helicoid, the lens gives a magnification range between 0.56x and 1.2x. The lens design is therefore moderately asymmetric. At 0.56x, the same fuzziness outside the center and darkening in the corners are observed, as with the lens in the original orientation. This lens clearly cannot be used at magnifications significantly lower than 1x. At 1.2x, the peripheral regions are fuzzy to a higher extent than with the lens in original orientation at either 1x or 1.4x. There is therefore no good reason to reverse this lens. The extension ring I epoxied at the front of the lens remains useful as a lens shade and for covering the lens front with a C lens cap. It may also become useful to attach a cup-shaped diffuser around the front of the lens.
By adding another 35 mm with a stack of C extension tubes, the magnification becomes 2.2x. There is an evident loss of contrast because of reflections from the inner walls of the C tubes, which is entirely eliminated by flocking the interior of the tubes. The lens is still very good at this magnification, and the working distance is 42 mm at this magnification (about 47 mm without the C extension tube at the front).
With a total extension of about 185 mm on Micro 4/3, magnification becomes 3.3x, and working distance is about 40 mm (without the front C extension tube). The resolution pattern used in the preceding tests is too coarse, and in the above figure I used a test pattern from an IC photomask. This mostly transparent pattern places high demands on a lens in terms of contrast and dynamic range, and even after flocking all extension tubes (the above figure shows the results after flocking) contrast is somewhat low. An ordinary subject would give a much better contrast. Resolution is still good.
There is no axial chromatic aberration at any tested magnification. Transversal chromatic aberration is mostly absent, but I can see just a hint at 2.2x (far less than with legacy photomacrography lenses regarded as very good, e.g. the Leitz Photar 21 mm f/2). In other tests, there is a medium amount of transversal chromatic aberration (a few pixels, although gradual and without sharp borders), at 3.3x. This makes me recommend that this lens not be used above 3x.
Adding a variable aperture
I inserted a diaphragm mounted in an M49 extension ring between the scanner Nikkor 40 mm and the helicoid. The diaphragm does not have an aperture scale, but it proved easy to stop down in one-stop increments, based on the automatic exposure time. I started these tests by adjusting the illumination intensity to produce an exposure of 1/250s at 200 ISO with the diaphragm fully open, and gradually closed the diaphragm to produce exposures of 1/125s, 1/60s, 1/30s and (when possible) 1/15s.
With the setup shown in the above figure, the scanner Nikkor 40 mm provides a magnification between 1.97x (in practice, 2x) and 3x.
I fully expected to see vignetting, and in fact dark corners are evident at f/2.8 to f/5.6 at 1.3x (top row). It was not possible to close the diaphragm further. However, already at 2x there is no visible darkening in the corners at any aperture between f/2 and f/ 6.3. At 3.5x (bottom row) there is no darkening in the corners, either, but perhaps a faint hint of a central hotspot at f/8. The lens should not be stopped down beyond f/4 at this magnification, in any case. At this magnification, the diaphragm allows apertures ranging from f/2 to f/8. There is a moderate change of narrowest aperture with magnification, for reasons not completely clear to me but probably related to the non-optimal placement of the aperture with respect to the lens elements. There is also a noticeable change in color rendering of images shot at exposures slower than 1/125s, which is apparently caused by the camera, for unclear reasons. All images were shot with color balance set to flash.
It is therefore possible to use a variable aperture with this lens within the 2x-3x magnification range on a Micro 4/3 sensor.
Tests on full-frame and APS-C
The Scanner Nikkor 40 mm does not cover full-frame at 1.33x. On the other hand, center resolution is very high, because even the 42 Mpixel sensor used for this test has significantly larger pixels than a 20 Mpixel Micro 4/3 sensor (4.5 µm versus 3.4 µm).
At 2x, there is a slight improvement in the corners, while center image resolution remains much the same.
At 3x, the corners further improve, while center resolution still holds its own.
In APS-C crop mode (which simply discards the periphery of the image and records only an APS-C-sized central area of the sensor), the Alpha 7R II has a resolution of 18 Mpixel, and of course retains the same pixel size. The lens at 1.33x provides a much better corner image quality in this crop mode, compared with full frame. Unless complete sharpness in the extreme corners is a requirement, this lens can advantageously be used on APS-C format at this magnification.
The intrinsic difference in image quality between an 18 Mpixel and a 20 Mpixel sensor is not significant, and cropping the Micro 4/3 image to a 3:2 aspect ratio eliminates the difference in pixel count in any case. For maximum detail at a given magnification, a 20 Mpixel Micro 4/3 retains the advantage of its smaller pixels over its APS-C equivalent. These two sensor formats are therefore roughly equivalent in practice.
However, the Micro 4/3 sensor seems to slightly exceed the resolution capability of the lens, and therefore the resulting image sharpness is not as high as the sensor would be capable of recording. On this format, extreme corner sharpness is also better than on APS-C format. The difference in sharpness is small in practice, and it may be worth using a Micro 4/3 camera with this lens if the slightly smaller subject area than APS-C, at the same magnification, is an advantage.
In APS-C crop mode, at 2x and 3x the image quality is high and uniform across the frame.
The scanner Nikkor 40 mm f/2 is designed for an optimal magnification of 1.33x. It tests as excellent on Micro 4/3 at 1.4x, and still very good at 3x. It begins to show problems in the periphery at 1x, and bigger problems at 0.75x. There is no significant axial or transversal chromatic aberration up to 3x. The focal length gives excellent working distance without forcing an excessively long stack of tubes, and the setup is very portable. A variable aperture can be added at the rear of the lens, and is usable at 2x and higher.