20

Canon EF 70-200mm f2.8L IS II USM Lens Construction Diagram

Lens manufacturers usually share diagrams like this for their products. However, they conspicuously omit specifications that would allow one to completely analyze the optical system, like the power of each element and the exact spacings between groups.

But presumably an optical engineer could disassemble any lens sample to extract that information, element by element.

So, does the industry maintain any pretense of "secrets" or proprietary advantage in the optical design of lenses, and if so, what exactly is considered sensitive and why?

Previously I assumed that lens competition was done entirely on glass quality, element grind/polish quality, assembly and packaging (i.e., housing quality, sealing, and tolerances), and quality of mechanisms like focus motors and zoom throw. I thought the only trade secrets, if any, would be in production techniques to economically produce a known design to a particular tolerance. Are there, in fact, any design secrets in the lens industry?

28

In general there are no secrets to lens design. Everything important, all breakthroughs, etc., are shared publicly or semi-privately with modification through patents, conferences, papers, etc. There are "temporary secrets" where something is closely guarded until it is published: For example, we have completed the final piece of the puzzle in free-form design without limitations imposed w.r.t. a form of symmetry, but it is not information we have shared yet since it is part of a Ph.D., and IP protection is very important.

When it comes time to publish, everything is laid out in patents, for example the Nikon 24 mm f/1.4G patent lays out the entire optical prescription for all to see. It is completely correct, except for the aspheric coefficients on at least the front-most asphere being fudged. The design is extremely sensitive to that asphere, so there is no way to brute force optimize your way to the correct coefficients easily from the ones given. The second asphere may also be fudged. The first one is what allows the lens to have such a wide angle of view at all while the second one is to enhance performance. With the former being incorrect the design fails altogether while the latter simply causes bad performance.

Zeiss also used to publish the Zeiss Index of Photographic Lenses through about 1950 that detailed each patent and lens in their production.

Different manufactures favor different design choices, though I cannot say for certain I know why. Zeiss tends to favor more "elegant" design solutions with more "fundamental" element shapes and design forms. Leica frequently uses the uncommon Merte surface (see page 40). In general Leica also pares things down to as few elements as possible with rather extreme designs in terms of surface curvatures.

Canon and Nikon frequently use what boils down to a double gauss master lens with a wide angle attachment in front of it to design high-speed wide-angles like the 24 mm/1.4G, 24 mm/1.4L.

Things like how to speed up autofocus (get the smallest element(s) you can to move for focusing) are also wide-spread knowledge. Perhaps the best kept secret is to pay the price for Code V (several thousand USD/month/user) in lieu of Zemax (about $35,000 per user) for its vastly superior tools for tolerancing a design and prepping it for manufacture.

  • To clarify the Nikon patent example: Did they intentionally obfuscate the exact specification of the front aspheric in the patent? And even so, isn't it a matter of taking a production sample's lens and putting it in front of a lensometer to get those coefficients, should someone wish to reproduce it once it's off patent? Or is it not so easy to measure aspherics? – feetwet Jul 3 '15 at 19:13
  • 2
    @feetwet a lensometer does not measure the required specifications - they are for eyeglass lenses. Nikon certainly obfuscated the aspheric coefficients to protect their intellectual patent. A 3D profilometer is needed to re-generate the aspheric coefficients, and this provides a limited degree of precision. When you reach the stage of doing curve fitting to that profile there may also be multiple solutions with that profile. An interferometer could also be used, but the first asphere in that lens has a very large departure - too great for an interferometer to see. – Brandon Dube Jul 3 '15 at 19:42
  • 2
    An enormous number of them have no good reverse-engineering tool. The glasses are particularly difficult - even if you know the index at one wavelength and the V number you do not know for certain what material it is. Additionally, any asphere will be remarkably difficult to replicate without the original equation. Measuring the airspaces and the radii of curvature is also very difficult. – Brandon Dube Jul 3 '15 at 20:28
  • 7
    I don't doubt you at all, but that obfuscation seems to be a terrible violation of the spirit of the patent system, which is essentially an agreement to reveal secrets in exchange for commercial protection. If the revealed secrets are actually half-truths, that protection shouldn't be given — the patent should be invalidated. – mattdm Jul 4 '15 at 3:17
  • 1
    @mattdm: Patents must include information that a typical person skilled in the art would find both novel and useful. It is not necessary that a patent include all such information known to the inventor. If the patent includes a design which would work better than any other design that didn't rely upon proprietary information, such a design would qualify as "novel and useful information" even if such improvements were nowhere near as great as others known to the inventor. – supercat Jul 4 '15 at 20:34
0

Of course you could always reverse engineer the lens housing completely, first place it in an X-ray machine at different angles to determine the spacing, possible movements, builtin mechanisms etc. Then take it apart physically to inspect the lenses, you can then determine the shape of each lens and its properties. You can also inspect other objects. If you really want to take it to the extreme you can determine the composition of each object using elemental analysis.

  • It is nigh impossible to precisely determine optical materials from a sample without many pieces. They must be broken to ascertain exactly what they are. Additionally, the coatings are impossible to precisely re-formulate. – Brandon Dube Jul 3 '15 at 20:44
  • 1
    This answer does not really answer the question about if there are any secrets in lens design. Can you elaborate the answer? – Hugo Jul 3 '15 at 21:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.