Final Design

    As the pins on Integrated Circuits (IC's) become smaller and smaller, Delta Design's pick-and-place handlers will not be able to rely on passive compliance to align and test their Integrated Circuits (IC’s). Ultimately, the goal is to prepare for the next generation of IC's by providing an accurate/efficient micro-alignment system to properly orient each IC before testing.  To explore the answer to this question, the sponsor wants the group to explore the use of squiggle motors to align the chip.  Here, Delta wants to develop a mechanism that can align a microchip ±0.5 mm in the X, Y, and θ directions.  Three prototypes were created to research motorized micro-alignment: (1) A 52 mm footprint model demonstrated one axis of motion and closed loop control. This model demonstrated 1 μm precision and a large holding force (2) Another 52 mm footprint model was able to utilize a lever supported by a flexural bearing for mechanical advantage.  Using a squiggle motor for actuation, this model displayed a 3.6:1 mechanical advantage (3) Lastly, a 4X scale model incorporating mechanical advantage, position sensing, and stepper motors was used to give a tangible representation of a model that could be used for X, Y and θ micro-alignment                                                                                

Results

In the 52 mm footprint prototype, three things were achieved:

First, it was shown that the Squiggle motor assembly could fit within the original size constraints.  This is significant because it proves the value in researching Squiggle motor actuation.  Further, it allowed the team to move on to a 4X scale model with confidence that it could be scaled down.  

Second, the prototype demonstrated closed loop control and high resolution of the NSE5310 hall effect encoder.  In the end product, this precision will be integral in placing the DUT safely.

Lastly, the assembly illustrated the viability of utilizing the high holding force of the Squiggle motor.  While the assembly included no mechanical advantage, 30 grams of force from the Squiggle motor was enough to move the top stage and hold its position.

Results

The 52 mm footprint mechanical advantage prototype showed that the theoretical calculations can be used to design future flexural based mechanical advantage assemblies.  Here, the measured mechanical advantage as a function of lever rotation agree with the theoretical calculations.  At its furthest lever angle, the prototype achieved a 3.6:1 mechanical advantage giving the Squiggle motor a full 1 N of force.

Results

The 4X scale prototype highlighted a combination of flexural bearings, linear actuation with mechanical advantage, and position sensing using hall effect sensors.  Furthermore, mapping was developed to coordinate motor actuation with stage position demonstrating controlled parallel X, Y, and theta motion.

Finite Element Analysis (FEA)

    After assembling the 4X scale model, the team observed an unexpected rotation and deflection out of the XY plane of motion. This deflection induced friction, an unacceptable issue when dealing with these environments, and so the team decided to focus efforts to understand the parameters that could minimize the deflection. While with other calculations for beams, simplified 2D assumptions were used, the 3D nature of the parallel stage proved to be more complicated.  For one, the physical model showed torsion out of the plane.  While equations exist for torsion in a circular beam, there was no simple way to parameterize the torsion constant to solve in the standard torsion beam equation.  Here the equation is:

    Note that this equation suggests that the out of plane rotation should be inversely proportional to the length of the flexures.  However, without knowing the torsion constant, JT, it was hard to account for the thickness of the flexure and different geometry of the stages under different load cases.

    To account for these three factors, a simplified version of the parallel stage design was modeled on Solidworks.  FEA analysis was run using Solidworks Simulation.  Here, the team was able to run three different load cases while (1) changing the thickness of the flexure and (2) changing the length of the flexure.  A summary of results can be found below.  Note that the exact dimensions of the flexures and the magnitude of forces were exaggerated due to the limitations in Solidwork’s meshing ability for geometry less than 0.25 mm thick.  For this reason FEA results are analyzed to show trends due to changing these variables with the ultimate goal of finding out how out of plane rigidity and in-plane elasticity relate to t and l.

Final FEA Results

Initial tests with the double flexure showed that increasing the length and the thickness helped to increase out of plane rigidity while maintaining XY flexibility.  However, even with this design criteria, the out of plane flexibility was still non-negligible.

In the latest FEA, the team was able to model quad-flexure configurations which showed two orders of magnitude better out of plane rigidity while maintaining XY flexibility.