Parameters of Lens Specification and the Properties of Lenses

by Mike Johnston

The Sunday Morning Photographer — a weekly photo.net column:    Feb 22nd , 2004

I'm going to indulge myself a bit here and assume you may not be a grizzled old veteran enthusiast with stacks of Camera 35 and Modern Photography in your closet or basement, and therefore might like a bit of an explanation of some of the properties of lenses. For those to whom these notions are old hat, my apologies in advance.

I've been known to have a pretty jaded and cynical and sometimes even (heavens!) sarcastic view of "lens tests." I feel for the most part that they're relatively useless. It's not just that lens tests are sops to insecurity, although they are that. It's not that most of them are silly, or misleading, or wrong, either, although they are often that, too. It's that, often, they're just not helpful at all.

One of the things that troubles my hearers when I go off on one of these rants is my scorn for single-number ratings. You know, this lens is a 3.8, this lens is a 4. The anxious lens-owner, looking for ego-bolstering or purchase justification, will feel pleased if he owns the 4 lens and may start biting his nails with a distracted look in his eye if his is the 3.8. I hope by the end of all this, you'll have an idea why I think this is silly — and why I think you should think so, too.

I don't want to talk about lens design per se (I don't really know much about it, to be honest. Talk to Ron Wisner or Erwin Puts). What I want to look at is how a lens is specified. When most experts talk about how lenses are made, they seem to start from the cozy assumption that the problem is primarily (or entirely) a technical one, that the goal is simply to make it as good as possible, and that the designer has carte blanche in terms of resources. Such articles are loaded to the rafters with happy tech-talk about aspheric elements and how multicoating defeats reflections and so forth.

All that's well and good, but as you've probably already guessed, the technical part of the design is not the whole story.

Over the years, there's been a gradual, herky-jerky shift in the business model of how companies come up with products. To simplify it a lot, under the old assumptions, the engineers got ideas for products and built them, and the poor beleaguered marketing departments then took whatever they were handed and attempted to sell it. Under the new paradigm, it's the other way around — the marketing department decrees what it wants to sell, and the engineering slaves burn the midnight oil until they come up with what they've been asked for.

I'm sure a book-length exegesis could be written expanding on the preceding paragraph (and perhaps has been). But let's move on. The fact is that the specification of the product when the product is a lens begins not with a technical problem, but with a marketing one. As such, the chief puzzles faced by the engineers is one of dealing with limitations that may have little to do with absolute exercises in design. Only seldom is the marketing problem to make something of higher quality than the competition has.

Let's look at some of the limitations:

• MONEY.This is the biggie. If you could put it on a scale, it might outweigh all the others combined. Hobbyists typically have in mind a set hierarchy of who's best — Leica first, then Zeiss, Canon and Nikon tie for third, then Pentax and Minolta, and so on down the line. I get outraged stares when I say that any competent lensmaker could build the best lens in the world it it had enough money. But it's the case. All this earwash about "we have the best glass" and "we have the most expertise" and "we use the best computers" and so forth is marketing posturing. There may be optical sweatshops that simply don't have the expertise or the equipment to do well, but believe me, most lensmaking companies could build the world's best lens if it knew in advance it could sell 3,000 of them at $5,000 each.

The actual cost constraints are considerable. Let's take a brief look at two lenses that have appeared in recent years, the Leica 50mm Elmar-M and the Nikon Nikkor-P 45mm f/2.8. Both are derived from a simple 4-element Tessar design (originally developed by Zeiss) that's more than a century old. Both are easy as pie to manufacture. But at $700, the Elmar-M is the cheapest rangefinder lens Leica offers and (well, sometimes) gets talked up as a good value, while at half that price, $350 or so, Nikon users grump and grouse about how costly the Nikkor-P is "for what you get." So tell me, what's the price you think the market would bear for a Tessar-type labeled Phoenix or Samyang? Think those companies could find a market for a $700 one? But if you went to Schneider or Elcan or Perkin Elmer or Cosina and asked if they'd like to build 100 Tessar-types for you for $200,000, you think you wouldn't get a lens at least as good as any on the market?

In fact, a good deal of the engineering problem addresses questions the manufacturers would rather you not hear them ask, such as, How poorly can it perform before buyers start to bitch? How few elements can we get away with? How little coating can we get away with? How cheap can we churn 'em out? How much profit can we build in? Well, perhaps not these actual questions, but you can be sure the issues behind them are not far from the corporate consciousness.

• Size, weight, and physical lensmount constraints.Size is a major design constraint. In general, the larger the designer is allowed to design a lens to be, the easier it will be to make a good one.

Pros have learned over the years that they're going to have to lug some weight if they want the best performance. Consumers aren't so well trained. Consumers just don't care for great big heavy lenses, and they tend not to buy them. Again, a brief example: a number of years ago, Contax (Kyocera) commissioned a 35-135mm zoom from Zeiss To keep standards up, Zeiss delivered a lens that was approximately the size and weight of a mortar barrel filled with bricks. A magnificent performer, it sells at the rate of approximately four per year. (Okay, I'm exaggerating. I'm weak.) Ten or a dozen years later (I'm not looking it up...told you I was weak), Contax brought out the nifty little Aria, a camera more or less expressly designed for Japanese females. The winds of the industry were changing by that time, so direct comparison would be meaningless, but Contax had learned its lesson: the 28-70mm marketed with the Aria had lots of polycarbonate in it and, while not wee small, it was indeed wee small for a Zeiss — a lot lighter at 11.5 oz. than the older 35-70mm, which is 17.5 oz., not to mention the aforementioned 35-135mm, which weighs 34 pounds. (Okay, actually 25.5 oz.). Over the years, a great deal of optical and engineering expertise has gone into making lenses "just as good, but smaller."

Physical lensmount constraints are another often inflexible limitation. Just as long lenses are often limited by the allowable size of the objective (outermost) element, fast lenses are often limited by the size of the exit pupil (the element you see when you look at the back of the lens). Nikon designers might like to make a fast lens with a two-and-a-half-inch exit pupil, for instance, but the project is not likely to gain approval since the Nikon F-mount is one and seven-eighths inches across. Wide lenses are sometimes constrained by the amount of backfocus that is or is not needed, and leaf-shutter lenses by the size and the speed of the lens and shutter. Why shutter speed? Because making a leaf shutter open very wide for very short amounts of time requires a more expensive shutter mechanism. Case in point are the slow "fastest" normals for medium-format rangefinders such as the Mamiya 7 and Bronica RF645. It's not that the lens's maximum apertures couldn't be bigger, it's that the leaf shutters would also have to be. Interestingly, this is a limitation for leaf-shutter rangefinders in that buyers expect normal lenses to be fast, small, and inexpensive, because that's the way it is with focal-plane-shutter cameras. Really, it would make the most sense for the normal lens for a leaf-shutter rangefinder to be the fastest and the most expensive in the camera's lens lineup, and probably also not the lightest, but that's just not what buyers expect, and the manufacturers of such niche cameras know better than to try to re-educate the entire market.

• Mechanical robustness, manufacturability, and durability. Many years ago, in the salad days, when cameras were sold in camera stores and camera manufacturers had reps and all was right with the world, a friend of mine attended a demonstration by a Leica rep. According to my friend, this man gave a short talk on mechanical robustness, during which he took a short section of barbed wire and rubbed the barb against the outermost element of the lens. Then he took the lens in his hand, crouched, and launched it like a bowling ball across the floor, where it skittered and bounced until it banged into the wall. He walked calmly over, picked up the lens, snapped it into his camera, and said, "Ready to take pictures."

That's mechanical robustness. Old lenses are often more robust than the cameras they fitted. In a trend that is likely to continue, Cosina / Voigtländer has recently introduced several cameras intended solely to make use of old lenses. The Bessaflex takes M42 screwmount SLR lenses, the Bessa R2C takes classic Zeiss Contax rangefinder lenses, and the Bessa R2S takes classic Nikon rangefinder lenses.

Manufacturability is another design issue that has seen incremental improvement over the years. Obviously, a product that can be made by semi-skilled labor in 10 hours is going to sell for a lower price and have more profitability built in than a similar one that requires highly trained workers 20 hours to make. Some of this ease of manufacture can be "designed in" — in the case of a lens, for instance, element edges that are fatter and have more contact area are likely to be easier to collimate (align) properly. Lenses with fewer elements and fewer moving groups are also likely to be easier to make. Complex shapes molded from plastic or magnesium may be easier to make than the same shape machined out of a billet of metal. You get the picture.

In the case of cameras, manufacturability has a lot to do with economies of scale — the number of units of a lens projected to be sold. A soap manufacturer with a $250,000 high-speed boxing machine is likely to be more cost-efficient than a soap manufacturer that employs thirty people in a room boxing the soap by hand, assuming it sells more.

More in a couple...

Well, hmm. I notice this is getting long, and you must be getting tired, because I am. And I haven't even really begun to cover the subject I started out to cover, because what's still to come are all the properties of lenses that photographers, as opposed to designers, engineers, and manufacturers, really care about. This will continue two weeks from now. Or maybe, given my tendency to procrastinate, three.

—Mike Johnston

Click here to support SMP with a few centavos.

SMP archives