[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Rollei] lens manufacturing

At 06:55 AM 06/22/2000 -0400, you wrote:
 >      i have hesitated for many months to get into the occasional
 >debates about zenar vs, tessar ad infinitum but here goes..
 >      granted my information and knowledge about these matters is
 >sparse but, but, but.
 >      my impression has been that at one time the figuring of a lens
 >formula, the grinding of a lens etc. was all a matter of much mental and
 >physical effort. We have all read stories of teams of mathematicians
 >spending months or years to compute a new lens. and of course as far as
 >grinding and polishing a lens let`s not discuss it. anyone who has ever
 >ground a telescope lens with his son or his dad has some input on that
 >      In fact to this day Leitz supposedly hand matches elements in
 >their camera lenses rather than taking one element after another and
 >placing it into the lens cell.
 >      However it was my understanding that, with computers, lens formulas
 >can be calculated in minutes in what used to take months or years. The
 >grinding and checking of lens elements i have assumed have progressed
 >      Now all of the above is based on a minimum amount of knowledge.
 >But from this i have derived the fact that schneider can copy whatever
 >zeiss has or visa versa unless it is patented or unless for whatever
 >the reason do not wish to.
 >      My own opinion? Pre- ww.11, tessar's are contrastier than most
 >other lens of similar age.
 >      Recently I built a collimator- i was shocked to find Tessars not
 >that sharp wide open. Now they were sharp but barely, and two stops
 >down perfect.
 > I know i am supposed to have known this beforehand but until i have
 >the time to test other tessar lens i am skeptical of what my eyes
 >    Comments appreciated,
 >         regards fro the u.s.a.
 >           ellis feldman

    Computers do speed up the process considerably. Lens designs are 
usually evaluated by ray tracing, a mostly trigonometic calculation. It can 
take half an hour to trace a single ray using a log table. A hand 
calculator will reduce this to a couple of minutes. A computer lens design 
progam will trace hundreds of rays and calculate the results, and chart 
them in a second.
   The lens design program can not replace the designer but can help 
tremendously with the calculations. Theh program can also calculate such 
things as the sensitivity of various lens parameters, like glass constants. 
This can avoid having a design which calculates well but is impossible to 
build practically.
   Spherical surfaces are a lot easier to generate than the parabolic ones 
needed for a telescope mirror. Modern computer controlled grinding 
machinery can make aspherical surfaces economically, as demonstrated by the 
number of commercial lenses with them, but, in the past they were very 
expensive to make and need to be made one at a time. Spherical surfaces are 
made many at a time on the same machine.
   Where elements are to be cemented the mating surfaces must be very 
precisely ground and polished to exactly the same curvature. This is done 
one at a time, which adds to the cost of cemented surfaces.
   The method of checking the figure of the lens is to check it with a very 
precise guage using interference patterns. Probably a laser interferometer 
would make a more precise guage.
   After grinding and polishing the elements are "edged" by placing them on 
a centering machine and grinding the edges concentric and coaxial with the 
optical axis. For air spaced elements the precision required is not too 
great since the lens mount will automatically center spherical surfaces.
   Elements which are to be cemented must be centered to a much greater 
degree of accuracy since the edge is the usual reference surface for 
aligning the elements for cementing. Again, this adds to the cost of a 
cemented surface.
   Before any lenes are actually made the glass is checked for its 
constants. For the design to work the glass must be very close to the 
catalogue values the designer used. While slight adjustments may be made in 
the curvature of the surfaces to compensate for variations in glass, each 
change would amount to a new design so, in general, such an adjustment 
would be avoided.
   No lens is at its best wide open unless designed that way especially. 
Lenses which must be best when wide open are generally either a faster lens 
which is effectively stopped down when wide open or lenses which are 
designed for rather narrow coverage, as is the case for projection lenses.
   There are seven basic aberrations of lenses, sometimes called the Seidel 
aberrations after the mathematician who calculated them. They are: 
Spherical aberration, coma, astigmatism, longitudinal chromatic, lateral 
chromatic, curvature of field, and distortion. Of these spherical, coma, 
and astigmatism are reduced by stopping down the lens. Chromatic seems to 
be reduced as does field curvature because of the increase in depth of focus.
   Beyond these first order aberrations there are higher orders. Up to 
seventh order aberrations must be taken into account in a complex lens. It 
is the balance of these higher order aberrations which often determine the 
subtle differences in lens performance among lenses of essentially similar 
   In addition to all this cerain aberrations can be minimised for only one 
subject distance. This is due to the geometry of the system. The rule is 
sometimes known as Abbe's sine condition, and sometimes as "the optical 
invariant".  In any case, its why lenses are said to be "optimised" for a 
certain distance or degree of magnification. The lens can be corrected for 
spherical aberration, coma, and astigmatism at one point in the image field 
for only one distance. This is usually infinity for camera lenses and 1:1 
magnification for process and copy lenses. For enlarging lenses it is 
usually the center of the range of magnlification the lens is intended to 
be used for. Some types of lenses are much more sensitive than others to 
change in subject distance. Generally the slower and simpler the lens the 
less its performance depends on object distance.
   Tessars, and many other lens, vignet at full aperture. If the iris is 
looked at from an angle you will see that it changes from round to an 
eliptical shape. At some point the lens mount also cuts off part of the 
iris. A lens must be stopped down to the point where there is no longer any 
mechanical vignetting at the edges of the field before it has a chance of 
having good performance. Very often this is where the iris is about half 
the diameter it is when wide open, or about two stops down from maximum. 
Some lenses are better this way then others. For instance, the Kodak Ektar 
tessar type lenses have less residual spherical aberration in the center of 
the field when wide open than Tessars of the same age. Both the Ektar and 
Tessar have less coma than comparable Xenar or Wollensak Raptar or Optar 
(the last made for Graflex) lenses. Well designed f/4.5 Tessar type lenses 
will have best corner performance at around f/11. An f/3.5 lens at around 
f/8 to f/11. The center performance will be better for slower lenses, an 
f/6.3 Tessar generally outperforming an f/4.5 lens, etc.
   The quality of glass is a key to good lenses. Glass does not bend light 
of all colors the same amount, blue light being bent more than red light. 
Optical glass comes in a variety of _indexes of refraction_ and 
_dispersion_.  The Index of Refraction is the measure of how much the glass 
bends light. It is the _average_ of the bending over the range of colors 
the glass is intended to be use for. Dispersion is the amount the Index of 
Refractin changes with color. These two must be chosen carefully if a lens 
is to have a minimum of chromatic aberration.
   There is a good tutorial and FAQ on lenses by David Jacobson on his web 
page at the Rochester Institute of Technology, there are links to it from 
the Graflex web page at http://www.graflex.org
   For an example of a computer design program do a web search for "Oslo". 
Discard the travel stuff for Norway and you will come to the web page for 
this program. There is a free ware version which will handle up to (I 
think) twelve surfaces, the object and image surface are incuded. This 
program has a good graphical interface. It is not hard to set it up for the 
lens prescriptions from patent date. It comes with several sample lenses. 
This program will allow you to see what sort of differences are produced 
when a lens parameter, like bending or spacing, is changed.

   Enough already!
- ----
Richard Knoppow
Los Angeles,Ca.