Parfocal Lens and Autofocus System

ME94 Poster1024_1.png

After an injury left me unable to compete on the cross country team my sophomore year, I became the unofficial team photographer. Around this time I was also learning how to use the various machines in the engineering lab, so I thought it would be a fun project to try and design my own camera lens.

After watching all of the MIT OpenCourseWare material on intro to optics I came up with a simple design for a fixed focal length (aka prime) 30mm f1.2 lens. I wanted a lens that would let in as much light as possible at the cost of chromatic aberration and distortion. Based on my goal specifications and the flange-focal distance of my Fujifilm XE-2s mirrorless camera body I determined the specifications of the four glass elements I would need for this project. I made a CAD model of my lens design in SolidWorks to verify that the rotational focusing mechanism I had envisioned would work. I found a supplier in Rochester NY who was selling off old Eastman-Kodak optical quality glass for pennies on the dollar and bought all of the required glass.

I first 3D printed a prototype of my design to make sure everything looked OK before moving on to machining it on the lathe.

Sample photo taken with this lens on Franconia Notch, NH

Sample photo taken with this lens on Franconia Notch, NH

My next project was to design a 70-200mm Parfocal zoom lens. Many modern zoom lenses use a varifocal design, which means that the point of focus changes as the focal range shifts when the user adjusts the level of zoom. This offers more freedom in lens design but requires feedback between the camera body motors to adjust glass elements in tandem to maintain focus.

I calculated the trajectories that the two moving lens groups would need to follow to achieve a parfocal configuration. I then used the equation curve tool in Autodesk Inventor to wrap each function around cylinders that would contain each lens group. These curves were cut from the surface of each cylinder to create cams that would cause the lens groups to follow the desired trajectories as the focus and zoom rings were turned on the lens barrel.

IMG_20180416_170709_212.jpg

I then printed my design using the FormLabs Form 2 I had access to at my summer internship.

MVIMG_20180426_184918.jpg

Here are some sample photos I took with this lens on a trip to Vancouver BC.

For an independent research project my senior year I continued work on my camera, this time working on an autofocus mechanism. The main advantage of a Parfocal lens design is the fact that the focal plane for a given object remains constant regardless of the level of magnification. This means that the lens can hold a focus as it is zoomed in and out without any communication with the lens body. Furthermore, this allows the focusing mechanism to be moved from the lens into the camera body. Having the focus mechanism inside the camera body allows more space to prototype and also means that the overall cost of a full camera kit is much cheaper as only a single motor control system is required, rather than a different motor in each lens.

X mount cutaway-01.jpg

This prototype was made using the image sensor of a Raspberry Pi camera with the original lens torn off. Because the footprint of the pi’s sensor is so much smaller than the APS-C image sensor on my mirrorless camera, images taken through this prototype appear much more zoomed in than photos taken with my real camera body. For that reason, I did all of my autofocus testing with my Fujifilm 24mm Prime Lens.

VID-51580721-121445-658 (1).gif
Previous
Previous

Deep RL Bipedal “Walker”