During the manufacture of molded plastic lenses, one item that can be monitored and adjusted is the relative decenter of the two surfaces, which can have a profound effect on the performance of the lens.  We were able to measure this decenter, as well as other internal errors, to a high degree of accuracy and repeatability by using a point-source microscope (PSM) in conjunction with the computational and optimization capability of ZEMAX.

The measurement was performed on an actual commercially available injection-molded plastic lens, the Philips CAY003.  The prescription of this lens is available in ZEMAX under Tools…Catalog Lenses.  Without the going into detail regarding the operation of the PSM (see www.optiper.com), it suffices to say that the PSM is able to locate both the vertex (by focusing on the diamond turning marks on the surface) and center of curvature (by retroreflecting from the surface) of an optical surface.

In principle, after measuring the location of the vertices and center of curvature (CofC)’s of each surface, one knows (to first order) everything about the as-built properties of the lens (tilts, decenters and thickness error).  However, a complication arises in this situation because when measuring the location of the vertex and center of curvature (CofC) of the rear surface, it is necessary for the beam to travel through the lens in double-pass.  Fortunately, we were able to fully account for the refraction at the front surface by modeling the test setup in ZEMAX.

The lens file used is attached. The lens data editor shows the catalog lens, with its front face tilted and decentered by a pair of coordinate breaks:

The variables consist of the tilt of the front surface, thickness error, and the tilt and decenter of the rear surface.  The (x,y,z) location of the front surface was designated as the origin with no decenter (Configuration 3):

A sample layout is shown below. Configuration 4 is the top layout, and 1 the bottom. The center of curvature of the rear surface is being located in Configuration 4.

Since this example is not an imaging system, it was necessary to create a custom merit function in order to simulate the test setup.  REAR operands were used for the two configurations in which the beam was focused on the vertex of a surface.  Arrays of RAID operands were used for the configurations in which the beam was focused on the CofC.

The measured data consisted of 4 sets of (x,y,z) coordinate locations, corresponding to the 2 vertices and 2 CofC’s.  These measured coordinates were then input into the ZEMAX multi-configuration data editor (rows 2,3 and 4 above). Since the (x,y,z) location of the front surface was designated as the origin with no decenter, it was only necessary to input 9 data points.  Once the data is input to the model, one clicks on the OPT(imize) button, and then the information regarding the internal errors of the lens are contained in the optimized model.

After taking and analyzing several data sets for various lens tilts relative to the measurement axes, we found the accuracy and repeatability of our measurement of the relative surface decenter to be on the order of 1µm.