Photographic zooms

June 19, 2929 - 195 zoom lenses contributed by Bill Claff

Bill Claff has contributed 195 models from his website Amazingly, these aren’t isolated models that only exist in lens design code. Instead, Bill has correlated the disclosures in lens design patents to the lens schematics in lens manufacturers’ product literature. This correlation opens a wealth of possibilities, linking these design models to publicly-available data.

US patent application 20210003832 offers a good example of such correlation. Bill has shown that example 2 from this patent is an excellent match to the lens schematic shown in product literature for the Canon RF100-500mm F4.5-7.1 L IS USM. The figure below shows my side-by-side comparison between the Sigma product literature and my Zemax model built from the patent disclosure. Visit Bill’s Optical Bench Hub to see the two designs superimposed on one another; the match is excellent.

All glasses had an excellent match to real glass types, generally to Ohara. The only exception to this excellent matching is the back element of the first doublet after the stop. The patent application calls for this lens to be Nd = 2.0509, Vd = 26.94, which is a very high-index glass. The closest real glass I could find was Ohara’s S-LAH79, with Nd = 2.0033, Vt = 28.27328. Such a substitution gives both poor performance and a poor match to focal length. I see no description of this special material in the patent application. Canon’s patent application didn’t indicate that there was anything special about this element.

Matching Canon’s MTF curves to Zemax calculations was straightforward using the apertures shown in Canon’s product schematic. In the plots below, the top row is Canon’s MTF data and the bottom row is my Zemax output.

Ken Rockwell has a thorough product review of this lens. He thinks very highly of it - particularly its flexibility as a macro. I didn’t model this performance because the patent application didn’t describe the lens motions for macro. For zoom alone, eight spacings change, so I didn’t feel like I had enough information to find the focus adjustments.

Ken seems unimpressed with the bokeh in this lens, describing it as only “neutral to good.” Good bokeh is obtained from a round, blurry exit pupil. Examining the beam footprints on the last lens element gives an idea about the roundness of the exit pupil, which looks pretty good to me. Because the spherical aberration is well-controlled, the exit pupil wouldn’t be very blurry, degrading bokeh.

Ken sees no distortion at 100mm, and “very little” pincushion distortion at longer focal lengths. Zemax agrees, calculating <0.05% distortion at 100mm and +1.8% at 225 and 500mm. These values agree quite well with the Photoshop correction factors that Ken found.

Ken characterizes vignetting by taking photographs of a uniform grey screen. ImageJ offers a nice tool for quantifying this image, showing 32% falloff at full aperture at f=100mm. Zemax calculates a 34% falloff, matching Ken’s measurement exactly.

Left image is copyright Ken Rockwell. Used with permission.

Ken's photograph on the left is of a uniform screen; it's intended to characterize falloff. The center graph is a diagonal trace of this photo, quantifying the falloff. The plot on the right is a Zemax vignetting plot of the lens model.

Ken looks for lateral color and finds none when operating under typical conditions, in JPG with the camera’s Chromatic Aberration Correction on. Deliberately trying to find the lens’s underlying performance, Ken is able to find “slightest bit of green-magenta lateral fringing at 100mm, none at 200mm, and just the slightest bit of magenta-green lateral fringing at 500mm.” This assessment matches Zemax calculations shown below, which are radial ray fan curves at 70% field, for f=100, 225, and 500mm. At the mid-zoom, any spectral splitting is smaller than the overall ray aberrations. At the ends of the zoom range, a bit of splitting is evident, and the two zoom extremes have different sign.

Axial ray aberration curves for the Zemax model at each of the configurations in the patent application.

Lens Rentals offers a very different hands-on description of photography lenses. They publish “teardowns,” in which they take lenses apart to see how they work. These teardowns are Gold for anyone wanting to learn about the engineering of photographic lenses.

Roger Cicala’s description of the teardown of this Canon lens is an unusually interesting read, for several reasons. First, it’s a particularly complex lens. Roger beautifully documents the bearings, sensors, and mounting. Even better, he describes why he finds them particularly high-quality. Furthermore, this teardown is special because if involves a plot - the mystery of the broken lens.

LensRentals describes that several of their early-release lenses of this model were returned with a broken lens element in the interior of the assembly. Roger describes his teardown as a mission to understand this breakage.

The broken lens is the negative element immediately behind the stop. This lens looks a little thin, with an aspect ratio of 25:1, which may contribute to its breakage. Other factors in its breakage might include that It also moves with focus and isn’t clamped when powered off.

I was curious if the lens material might have played a part in its breakage -some optical materials are particilarly susceptible to thermal shock. According to the patent, it’s made with glass nd = 1.497, Vd = 81.54. These values match exactly with Ohara glass S-FPL51. This glass is unusually susceptible to thermal gradients because it has an unusually high coefficient of thermal expansion (CTE) and an unusually low thermal conductivity.

The thermal-induced stress in a glass is described in Schott’s technical document, “TIE-32 Thermal loads on optical glass” For a given thermal load, I *think* the stress from a thermal gradient is proportional to a E /L /(1-m), where a is the coefficient of thermal expansion, E is the Young’s modulus, L is the thermal conductivity, and m is the Poisson’s ratio. By this measure, S-FPL51 is way out of the ordinary. It has 3X the thermally-induced stress as well-behaved glasses such as S-BSL7. S-FPL51 is the second-worst glass in the Ohara catalog for this measure.

Maybe this thermal characteristic played a role in the breakage observed at LensRentals.

Comparison of glass properties contributing to thermal stress.

May 1, 2021 - 11 Zoom lenses posed from Iain Neill’s recent paper

In his recent paper, Iain Neill divides the development of zoom lenses into 1) early fundamental technologies, 2) technological improvements that enabled zoom lenses to be alternatives to fixed focal length lenses, 3) technologically driven improvements brining zooms to parity with primes for some applications, and 4) potential optical design technology and manufacture for the future. These developments are richly curated, with 44 references and 20 illustrative figures.

I was able to generate Zemax models for 11 of these figures.

- Neil's Figure 16 (US Pat 7,012,759) is a special design, a compound zoom, with 300X zoom ratio. Its total track s about 1100mm, and uses 41 glass elements, with only 2 aspheric surfaces and only 3 moving zoom groups. It has 19 glass types. The mass of the glass alone is 14kg, and the total axial thickness of the glass is 347mm. Absorption of the glass increases at short wavelengths; relative to the maximum transmission, bulk glass transmission is 80% at 488nm and 70% at 460nm. A little Googling shows that lenses like this were actually built, with broadcast sports the intended application.

- Other good models are from Neil’s Figure 5 (US Pat 4,632,498, an afocal telescope for 8-12um band) , Figure 15 (US Pat 6,122,111, a cine zoom lens from the late 1990s that made use of linear bearings to support achieve excellent performance across 3.5X.) , and Figure 21 (US Pat 7,855,838, a hybrid zoom lens system incorporating a liquid cell.

-- Five of these models (Neil’s Figures 1, 2, 4, 6, & 7, US Pats 756779, 696788, 2566485, 2847907, and 2663223) are primarily of historical interest. They illustrate the limited performance that was achievable with the limited technology at the time.

- Two of the models (Neil’s figure 8 & 9, which are US Pat. 9,250,422 B2 and SPIE vol 2539) are intended as illustrations only; they aren’t fully-functional models. The references had no prescriptions, but I was able to get partially functional models from scanning the figures. Such simplified models are often useful as starting points for new designs.

I was unable to generate Zemax models for several of Neil’s figures. Several examples (Figures 10, 11, 13, and 19) don’t have prescriptions in the references or subreferences, and the figures aren’t clear enough to scan. Three examples (Figure 3, 17, and 18) have prescriptions in their references, but the prescriptions contain errors that I wasn’t able to find. Please contact me if you’ve had better luck with these models.

November 10, 2019 - 2 zoom lenses posted from Handbook of Optics

February 8, 2017 - 59 zoom lens designs posted, including 5 by Obama

Akihiko Obama, that is - he's a lens designer at Nikon.

The patent literature is full of zoom lenses. In January 2017 alone, at least 17 zoom lens patents were issued. The posted files are a sampling of some of the subcategories I've been interested in, including compact zooms for DSC cameras, high quality zooms for full-frame photography, and lenses for television cameras with large zoom ratios.

April 8, 2015 - 55 zoom lens designs posted

Today I posted 56 zoom lens files. All are for folded zooms, which provide an interesting study. As I described in a paper at the 2014 International Optical Design Conference, this class of optics is interesting be they are of a reasonable complexity that the designs are likely to be instructive, they offer a small range of design constraints because of the standardization of CMOS imager sizes and the mechanical constraints imposed on DSC cameras, and there is enough recent activity in awarded patents that a representative sample of contemporary designs can be collected.

Some of the interesting conclusions from the IODC paper include:

- An analysis of the most-common glass types for this application.

The most-frequently disclosed glass type is n=1.855, V = 24, which corresponds to glasses such as FDS90 and N-SF57; its prevalence suggests the adoption of high index glasses despite their coloration and workability. The next most-widely disclosed glass type is n= 1.505, V = 82, which corresponds to glasses such as MC-FCD1-M20; its prevalence suggests widespread use of glass molding, which is consistent with the high volumes in this application. The most-common glass types are listed in the table below:

n V # Glass types

1.855 24 148 FDS90, N-SF57

1.505 82 59 MC-FCD1-M20, FCD1, N-PK52A

1.535 56 54 E48R

1.885 41 52 TAFD30, N-LASF31A

1.925 21 50 E-FDS1, M-FDS1

1.495 70 42 J-FK5, FK5

1.735 55 40 LAK18, H-LaK5

- Even for the designs with the most-similar paraxial properties, and, probably, the most-similar design constraints, the resultant designs are quite different.

US Pat# 7,193,786 and US Pat# 7,417,800 have almost identical in terms of f/#, image size, focal length, zoom ratio, length, # of zoom groups, # of elements, and # of aspheres; however, the designers arrived at quite different solutions for what was apparently the same set of design constraints. Differences between the two designs include: 1) The third group in the ‘786 design, is a doublet, while the third group in the design disclosed in the ‘800 design, is a singlet. 2) The front element in the ‘786 design is meniscus, while the front surface of the ‘800 design is nearly planar. This difference is significant in a functional sense in that the nearly-planar front surface might offer some attractive packaging options. 3) The bi-aspheric element in the ‘800 design is nearly at the stop, while the bi-aspheric element in the ‘786 design is at the far side of the group from the stop. 4) Although the total number of elements is the same, the number of elements in groups two and three aren’t the same for the two designs. 5) Group 2 in the ‘800 design includes a particularly low-index material.

I found it interesting that these differences weren’t reflected in the patent claims. The key differentiator in the patents is the power across the various zoom groups.

Summary of designs


Design files