Difficulties in the development of wide-angle zoom lenses
Zoom lenses have not been the topic of many previous NIKKOR The Thousand and One Nights tales. Of the previous 45 tales, and excluding two discussing COOLPIX products, zoom lenses have only been the topic of six tales. Part of the reason for this is that it is only in the past thirty or forty years that zoom lenses have become so common. This is only the second wide-angle zoom lens discussed in NIKKOR The Thousand and One Nights tales, the first being the Zoom Nikkor 28-45mm f/4.5 introduced by Haruo Sato in Tale 15. After examining the history of NIKKOR lenses, I was surprised to discover that the Ai Zoom Nikkor 25-50mm f/4, released in 1979, was a direct successor to the Zoom Nikkor 28-45mm f/4.5, released in 1975, and that not a single wide-angle NIKKOR zoom lens covering the 28mm focal length was released during the four-year period between these two lenses. It is certainly a new world today where the standard zoom lens starts at a focal length of 24mm, and 14-24mm f/2.8 fast, ultra wide-angle zoom lenses are available. 1979, the year in which the Ai Zoom Nikkor 25-50mm f/4 was released, was also the year before the Nikon F3 and Nikon EM were released. At that time, the 50-300mm (then considered an "ultra" high-power zoom lens) was available, and the Nikkor 13mm f/5.6, which reigned as the lens with the world's widest angle of view for many years (the topic of Tale 9), was already mentioned in brochures. This is an indication of just how difficult it was to design wide-angle zoom lenses.
The first difficulty worth mentioning is the number of complex computations required with the design of all zoom lenses, not just wide-angle zoom lenses. With prime lenses (those with a fixed focal length), simulation at only the specified focal length is necessary to evaluate and correct aberration and performance at close shooting distances. With zoom lenses, however, computational complexity is compounded several times because performance with changes in focal length must be considered in addition to performance with changes in shooting distance. Computers began to come into common use in the 1960s, and by the 1970s, calculation speeds had increased overwhelmingly. However, the calculation speed of mainframe computers at that time was nothing compared to the capabilities of the latest personal computers. Even the fastest mainframe computers of the 1970s were said to have a maximum speed of several MIPS*. As current high-speed computers are said to have processing capabilities of several tens of thousands of MIPS or more, 1970s mainframes required at least ten thousand times the amount of time required by modern computers to perform calculations. That means that calculations performed in one second by a modern computer took three hours with a 1970s computer. What an incredible difference! These are the conditions under which lens designers at the time were working.
*Million instructions per second (MIPS): One measure of a computer's central processing unit (CPU) performance. This is the number of millions of instructions a CPU can process in one second. Therefore, a 1-MIPS CPU is able to process 1,000,000 instructions per second. As MIPS was actually used only when comparing processing speeds for the same type of CPU, please forgive the very general and imprecise use here.
Despite the slow performance of 1970s computers compared to those of today, they did reduce computation time incredibly. In the 1940s, when logarithm tables and hand-operated calculating machines were used, it took approximately five minutes to calculate refraction for a single lens surface. The use of computers, however, increased calculation speeds by 1,000 to 10,000 times, resulting in qualitative changes in the methods used with lens design. This was the "automated design" method. Put simply, a matrix indicating the degree of change in each type of aberration according to minute changes in lens curvature and spacing between lenses was created, and a computer used this matrix to calculate the curvature and spacing that would reduce aberrations. In other words, an equation with lens curvature and spacing set as the variable parameters was formulated for each type of aberration to be compensated, and the computer calculated the simultaneous equations made up of these equations.
However, compared to the lens curvature and spacing that could change, there were a large number of aberration values that should be suppressed. Therefore, the equations were not calculated precisely. Lens design remained a process of trial and error; for example, the degree of aberration that needed to be suppressed sometimes increased, or improvements in performance resulted in irregular lens shapes. However, the use of even early computers was wonderful for designers because it allowed for high-speed processing of at least a part of the trial-and-error system that relied upon the designers' experience and intuition. The use of computers and data processing with the automated design significantly reduced the time required to perform the overwhelming number of calculations, as well as that required for trial-and-error processes, necessary in the design of zoom lenses.
Two-group zoom lens
There is another issue that makes designing wide-angle zoom lenses difficult. That is, finding the best zoom structure.
I refer you to Tale 4 about the Zoom-NIKKOR Auto 43-86mm f/3.5 released in 1963. This was a zoom lens constructed of three groups—a first group acting as a convex lens, a second group acting as a concave lens and a third group acting as a convex lens—giving the lens as a whole a triplet type structure. The triplet design is said to be simplest structure that allows the lens designer to overcome all aberrations, including chromatic aberration. It is a very basic lens design. The convex-concave-convex arrangement of lens groups provides for extremely simple and effective aberration compensation. Takashi Higuchi, designer of the Zoom-NIKKOR Auto 43-86mm f/3.5, surely designed the lens with a full awareness of the benefits of the triplet structure. It was this triplet design that made a level of performance sufficient for practical use possible with the 43-86mm zoom lens at a time when automated design was not yet possible. However, there is a drawback to this three-group design with which the first group is convex. That is, the wider the angle of view, the larger the front glass element must be. This led to the prediction among lens designers at the time that wider angles of view would be quite difficult to achieve.
So what about the two-group structure so common with today's wide-angle zoom lenses? With this structure, the first group is concave and the second group is convex, and focal length is adjusted by changing the distance between the two groups. At wide-angle positions the distance between the first lens group and the second lens group increases. With longer focal lengths and telephoto positions, the distance between the two groups decreases. Due to its retrofocus structure, this type of lens is considered advantageous for wider angles of view as well.
While this structure is quite simple, it is much more difficult to suppress variations in aberration levels at different focal lengths because the triplet design is not used for lens arrangement.
As mentioned in Tale 4, the Auto NIKKOR WIDE-ZOOM 35-80mm f/2.8-4, announced in 1961 but never released, was the world's first two-group structure zoom lens. So why did Higuchi adopt the three-group structure for the 43-86mm zoom lens while working on the ground-breaking invention of the two-group zoom structure? It seems likely that he did so in order to overcome the difficulties related to compensation for the varying levels of aberration that could potentially occur with the two-group structure. Even from design data for the 35-80mm zoom lens, we can clearly see that with zooming out from the telephoto position to wider focal lengths, significant variations in astigmatism and distortion occurred. Perhaps it was this low level of performance at wide-angle positions that prevented the lens from ever being released. The cause of this substandard performance was that position of elements in each group, and especially those in the first lens group, did not allow for flexible control of aberration generated with the first lens group. In a time before computers, and even with a trial-and-error process, it simply wasn't possible to perform the overwhelming number of calculations necessary to come up with the optimal structure for the first group.
Development of a wide-angle zoom lens with the first lens group acting as a concave lens, as well as two-group zoom lenses in general, would simply have to wait for the arrival of the computer.
The Ai Zoom Nikkor 25-50mm f/4
Now we can finally get to the subject of this tale. The Ai Zoom Nikkor 25-50mm f/4 was designed by Norio Mizutani. Working under Soichi Nakamura, introduced in Tale 15, Mizutani designed several zoom lenses. Design of the 25-50mm began around the time the 28-45mm lens was released, and was completed in February 1976. The lens was finally released in 1979 after performance and mass-production trials. After completing work on this lens, Mizutani stepped away from optical design and transferred to another department. Therefore, I have never met him. While it is likely that he was still working at Nikon when I started, I am disappointed that I never had the opportunity to speak with him.
Photo 1 shows the lens. As you can see from the cross-sectional diagram of the lens in Figure 1, the 25-50mm has a two-group structure consisting of four elements in the first group and seven elements in the second group.
The 28-45mm lens introduced in Tale 15 has a three-group zoom structure that allows for compensation of variations in aberrations that occur with zooming. However, the elegant structure of the first group in the 25-50mm lens enables a two-group structure that effectively suppresses variations in aberration. Let's compare the structure of the first lens group in these two lenses. The first lens group in the 28-45mm has a thin structure consisting of three elements in two groups, while the first group in the 25-50mm lens has a thick structure with four elements alternating between concave and convex (concave, convex, concave, convex). The thickness of the first group and the arrangement of the individual lens elements are important aspects of this lens design. By including a triplet structure in the first group (convex, concave, convex), flexible control of various types of aberration, including distortion, is possible, enabling effective suppression of variations in aberration that commonly occur with zooming with two-group zoom lens structures. This structure for the first group utilizes the minimum number of lens elements possible for a wide-angle zoom lens constructed only of spherical lenses. The same structure was also adopted for the Ai AF Zoom Nikkor 24-50mm f/3.5-4.5S that came later.
The second lens group also utilizes a triplet structure based on the Sonnar-type lens with four concave elements in the front and two at the rear for an asymmetrical structure. This allows for excellent compensation for spherical aberration and the barrel distortion common at wide-angle zoom lens. Though the structure was simplified somewhat, the same general idea was also applied to later lenses.
After nearly twenty years, the two-group wide-angle zoom design invented by Higuchi was finally completed.
Example photos were taken with the Nikon D700. Though designed more than thirty years ago, the Ai Zoom Nikkor 25-50mm f/4 is an excellent match for the modern D700 in every way, including design. As the lens is long and rather large, like a fast mid-telephoto lens, balance is good when mounted on a high-end camera. The lens is long because it uses an internal zoom mechanism with which group one lenses move independently inside the lens body without causing the end of the lens (72mm filter attachment size) to move.
Therefore, when the first group moves back with mid-telephoto to maximum telephoto zoom positions, the lens barrel itself acts as a hood. This, combined with quality coating effectively reduces ghost and enables clear rendering.
The first surprise is the lack of distortion. Even at the 25mm focal length, with which barrel distortion is most noticeable, the degree of distortion is comparable to that of a prime lens. It seems that Mizutani did not allow for any sacrifice in quality simply because this was a zoom lens. Rather he designed the lens based on the same standards used for prime lenses. Example 1 was captured with the Ai Zoom Nikkor 25-50mm f/4 at the 25mm focal length and an aperture setting of f/8, but the somewhat backlit buildings are still sharp and clear.
The lack of flare and excellent contrast at the wide-angle position are fantastic. As the aperture moves with the second lens group with zooming, in theory, more light should be admitted at wide-angle positions and less at telephoto positions. This lens, however, preserves the 1 : 4 aperture ratio by adjusting the diameter of the aperture according to the zoom position. Optically speaking, it's like using an f/2.8-4 lens. Mizutani took great pains to ensure sufficient aberration compensation. At the wide-angle zoom position, this effect similar to stopping down the aperture one stop significantly reduces flare and increases contrast.
Example 2 is a night landscape that was captured at the maximum wide-angle focal length of 25mm and an aperture setting of f/4. While coma and curvature of field at the extreme edges of the frame result in some blurring, the majority of the frame exhibits excellent contrast. While DX-format cameras are unable to capture the full 25mm angle of view, photos taken with such cameras, like the D300, most certainly exhibit sharp rendering throughout the entire frame with shooting at all aperture settings. Coma and curvature of field are significantly reduced at f/5.6, and nearly eliminated at f/8. As one might expect, illumination falloff in the four corners is noticeable at the maximum aperture and the wide-angle position, another reason to stop down the aperture to f/8 or f/11. Illumination falloff is also nearly eliminated, even at maximum aperture, by zooming in from 28mm to 35mm.
At the maximum telephoto position of 50mm, spherical aberration causes a slight amount of flare throughout the entire frame with shooting at maximum aperture. From design data as well, we can see that overcompensation for spherical aberration results in flare. Take a look at Example 3. This is a photo of the same scene captured for Example 2. It was also shot at f/4, but with the focal length changed to 50mm. Moving out from the center of the frame, we see that lights in the windows appear to bleed increasingly, and become distorted into elliptical shapes at frame peripheries due to sagittal coma. However, with actual use, the balance of this aberration seems quite good. That is because this bleeding of lights is rendered quite softly.
Example 4 is a close-up of a camellia captured at 50mm and the maximum aperture of f/4. Am I the only one that thinks the soft appearance of the flowers caused by a slight amount of bleed makes it look more alive? However, though the ring-shaped blur at center left in the frame is attractive, some may be disappointed by images of scenes like this due to the double-line blur that occurs in blurred portions behind the subject in focus.
As this double-line blur and flare are the result of the same cause, they are eliminated when the aperture is stopped down to f/5.6 for images with no flare except in the extreme corners.
After completing work on the Ai Zoom Nikkor 25-50mm f/4, Mizutani stepped away from optical design and transferred to another department, making this the last lens he designed. Despite his move, he must have wondered about the progress of development of the lens and was surely more pleased than anyone when it was released.
In 1981, Nikon substitued the Ai 25-50mm with a new Ai-S 25-50mm. The 25-50mm was very popular for the roughly ten years either the Ai or Ai-S version was sold until its successor, the Ai-AF Nikkor 24-50mm f/3.5-4.5S, was released in 1987. Zoom lens turnover is much faster than that of prime lenses. With such a long life, the Ai Zoom Nikkor 25-50mm f/4 was clearly a very popular lens. I have even received letters from photographers who still use the lens with digital-SLR cameras. The Ai Zoom Nikkor 25-50mm f/4 was truly a great lens.