Sie sind auf Seite 1von 4

Rashed Al-Rashed

Week 2: Cinematic Coupling


Motivation / Initial Planning:
Through my Master’s thesis, members of my lab were constantly using cameras for data
gathering and material property evaluation (tensile tests, bending stiffness, etc). Most of these
operations required careful placement of the tripod holding the camera. There was no easy way
of adjusting the camera’s settings without adversely affecting the placement of the camera.
Additionally, for camera-related tasks that were to be repeated on a near-daily basis, the tripod
mount would need to be disassembled immediately after use (to not take up any more floor
space than is necessary), only to be reassembled the next morning. I set out to address both of
these issues by designing a “cinematic coupling” that would act as a camera mount.

Functional Design Analysis References Risks Countermeasures


Requirements Parameters
Repeatability Under 3µm Abbe Fundamentals Cannot use Consider other
[Translational], error textbook as camera geometries
Within 0.1° analysis mount
[Rotational]
Stiffness Less than 10 Hertz Fundamentals Deformation Reconsider
µm deflection Contact textbook, generates material selection
of top plate Stress Slocum’s KC seventh
under load design contact
(4.27 N) spreadsheet point
Repeatability is the most important functional requirement for this kinematic coupling,
as the goal is to be able to place the camera at a consistent orientation repeatedly. Stiffness
was less of a concern, as minor deflection of the top or bottom plate would not affect
repeatability, so long as the top plate does not deform enough to close the gap and form a
seventh contact point with the bottom plate. The camera I designed this cinematic coupling
around was a Nikon L330. The total load generated by the camera and its mounting screw
equaled 4.27N, which constitutes the load that the kinematic coupling needed to withhold.
Initial Design:
I considered a number of concepts at this stage, including deciding between Maxwell
and Kelvin couplings, and whether or not to use bolts to generate a preload. Most cameras
have a mounting hole at their bottom face. The Nikon L330 had a threaded hole that fit a ¼-20
screw, allowing me to attach the camera to the top plate. I did not need a fixed center of
rotation around one of the contact points, so I opted for the Maxwell coupling, as it was easier
to manufacture. Due to personal experience with cameras and tripod mounts, I knew that the
load applied by the camera’s weight (4.27 N) would not be sufficient for the top plate to not tip
over if buttons were pressed on the camera without either preload or large spacing between
the top plate balls. I decided to not use bolts (or any other form of preload), as it allowed for
effortless removal of the top plate. In this situation, I could remove the top plate assembly,
modify all the settings on the camera, set a timer on the camera, and then reassemble the
kinematic coupling. This configuration meant that the kinematic coupling did not need to
handle any load other than the constant load provided by the camera and mounting screw,
vastly simplifying the accompanying analysis. Since my functional requirements were very lax,
and there were no concept decisions made requiring analysis until deciding on material choice,
a modified reference spreadsheet (“Kinematic_Coupling_3Groove_Design_Rashed.xls”) was
used to analyze my chosen concept.
Material Choice and Fabrication:
Using a reference spreadsheet (Kinematic_Coupling_3Groove_Design.xls, modified and
included in reference folder), I calculated the deformation caused by the camera load for
different materials that we had available in our lab space. I found that the deformation of the
top plate when using plywood was acceptably small (approximately 0.42 µm), and opted to use
plywood for ease of manufacturability. I could not identify the exact type of plywood I had on
hand, but found that the deformation was acceptably small no matter the type (or
directionality, as plywood is not an isotropic material) of plywood selected. For my final
calculations, I used the material properties of Douglas-Fir plywood, as it seemed to be the most
commonly used plywood with average properties. I used steel balls found in lab (0.5” diameter,
exact type of steel unknown) as the top plate contact points. I found the center of mass of the
camera experimentally (by finding the X-Y coordinate of the point where the camera could be
balanced upon a sharp point), and placed the mounting hole such that the center of mass of the
camera would be directly above the center of the kinematic coupling, as to avoid any moments
acting upon the coupling.
After selecting my materials, I started developing my CAD model of the kinematic
coupling. Many parameters were decided by the materials and tools I had on hand (ball size,
plywood thickness, groove angle). As the tolerances on most of the system dimensions would
not affect repeatability, most tolerances were selected as reasonable tolerances that could be
achieved by the CNC router that I had access to (Shopbot D2418). Any tolerance issues would
be fixed when positioning the bottom plate to the desired location and orientation. Once the
CAD model was finished, I then used it to generate files that I could import to the CNC router.
After fabricating the top and bottom plate, I then used an adhesive XXXX to attach the balls to
their respective sockets. After attaching the camera to the top plate, the kinematic coupling
was complete (see “Finished Coupling 1/2.jpg”).
Measuring Repeatability:
To calculate repeatability, I set the camera on a timer, and set it down in front of
a ruler multiple times (ruler placed 0.75” away from camera lens). I used these images
to find the precise location of a selected marker on the ruler, and used that to measure
repeatability. It seemed appropriate to use the camera itself as the testing process
(detailed methodology and data in spreadsheet “02 - Kinematic Coupling Analysis and
Data.xls”).

I then wrote a simple matlab script (“camera_accuracy.m”) to threshold the


images (example input image and threshholded image shown above), thus allowing me
to easily find the location of a chosen edge of a given marker. Repeating this process
over a number of images allowed me to get a measure of repeatability in the exact way
that would matter when using the camera itself, essentially answering the question “how
many pixels off will you be on repeated use of the coupling?”. The end result was an x
error of 0.27 pixels, and a y error of 0.32 pixels, which, at that distance, represented an
x error and y error of 1.3 and 1.5 µm respectively, and a rotational error of well below
the set goal of 0.1°.
Dial Indicator

Kinematic Coupling

Spring Scale

Next, I tested the stiffness of the kinematic coupling. A spring scale was attached
to a zip tie that ran through the kinematic coupling bottom plate and attached to the
center of the top plate, and a dial indicator was set to measure the deflection at the
center of the top plate. I measured a stiffness of 1.42 [N/µm], compared a predicted
stiffness of 6 [N/µm]. I believe the discrepancy comes from the fact that the wood
underneath the ball might have undergone plastic deformation at some point during
testing or inspection of the kinematic coupling, by me or others. Looking back at the
spreadsheet, an addition of 11N would lead to yield, which is an amount easily
achievable by accidental rough treatment.

Das könnte Ihnen auch gefallen