Nikhil Bajaj

Design and Implementation of a Tunable, Duffing-Like Electronic Resonator via Nonlinear Feedback

To date, many vibration-based sensing modalities have relied upon monitoring small shifts in the natural frequency of a system to detect structural changes (e.g., in mass or stiffness), which are attributable to the chemical or biological species, or other phenomena, that are being measured. Often, this approach carries significant signal processing expense due to the presence of electronics, such as precision phase-locked loops, when high sensitivities are required. Bifurcation-based sensing modalities, in contrast, can produce large easy to detect changes in response amplitude with high sensitivity to structural change if applied appropriately. This paper demonstrates the design and implementation of a tunable, Duffing-like electronic resonator realized via nonlinear feedback electronics, which uses a quartz crystal tuning fork as the device platform. The system in this manifestation uses collocated sensing and actuation, along with readily available electronic components, to realize the desired behavior. The sensitivity of the device is tunable via the control of feedback gain and the type of Duffing-like response (hardening or softening) is also selectable, thus creating a versatile bifurcation-based sensing platform.

Reduction of memory footprint and computation time for embedded Support Vector Machine (SVM) by kernel expansion and consolidation

Support Vector Machines (SVM) are a family of algorithms widely used in classification and regression tasks. When faced with large and structurally complex data sets, the classification or prediction by SVM can become memory and time intensive, and especially so in non-sparse variants of the SVM such as the least-squares SVM (LS-SVM). This is of particular importance for implementing classifiers in embedded systems where memory and computation capabilities are limited. In this paper, decomposition methods for SVM classification functions are developed and discussed, using polynomial approximation methods. The SVM decision function is expanded into a polynomial form and consolidated into classification function with a significantly lower memory footprint and computational cost. The amount of reduction is analyzed for polynomial kernels, and in three demonstrated example systems, the classifier is made two orders of magnitude faster, with a memory requirement that is two orders of magnitude smaller. The methods are tested on standard classification data sets, and guidelines for use of the methods are provided.

Shaping the Frequency Response of Electromechanical Resonators Using a Signal Interference Control Topology

The recent study of signal interference circuits, which find its origins in earlier work related to active channelized filters, has introduced new methods for shaping the frequency response of electrical systems. This paper seeks to extend this thread of research by investigating the frequency response shaping of electromechanical resonators which are embedded in feedforward, signal interference control architectures. In particular, mathematical models are developed to explore the behavior of linear resonators that are embedded in two prototypical signal interference control topologies, which can exhibit a variety of qualitatively distinct frequency domain behaviors with component-level tuning. Experimental approaches are then used to demonstrate the proposed designs’ utility. [DOI: 10.1115/1.4034948]

Characterization of Resonant Mass Sensors Using Inkjet Deposition

Microand millimeter-scale resonant mass sensors have received widespread research attention due to their robust and highly-sensitive performance in a wide range of detection applications. A key performance metric associated with such systems is the sensitivity of the resonant frequency of a given device to changes in mass, which needs to be calibrated for different sensor designs. This calibration is complicated by the fact that the position of any added mass on a sensor can have an effect on the measured sensitivity, and thus a spatial sensitivity mapping is needed. To date, most approaches for experimental sensitivity characterization are based upon the controlled addition of small masses. These approaches include the direct attachment of microbeads via atomic force microscopy or the selective microelectrodeposition of material, both of which are time consuming and require specialized equipment. This work proposes a method of experimental spatial sensitivity measurement that uses an inkjet system and standard sensor readout methodology to map the spatially-dependent sensitivity of a resonant mass sensor – a significantly easier experimental approach. The methodology is described and demonstrated on a quartz resonator and used to inform practical sensor development. ∗Address all correspondence to this author. This work was supported by the U.S. Department of Homeland Security’s Science and Technology Directorate, Office of University Programs, under Award 2013-ST-061-ED000

Numerical and experimental studies of ultra low profile three-dimensional heat sinks (3DHS) made using a novel manufacturing approach

The continued increase in electronic device packaging densities is placing ever more challenging performance requirements on air-cooled heat sinks. In cases where the state-of-the-art heat sink technology is unable of to meet these requirements, this often results in either a relaxation of design specifications, or the exploration of other thermal management technologies better able to handle high heat density applications, such as liquid cooling. Both of these approaches provide challenges to equipment designers, as relaxing requirements does not allow for a scale-able path to increased device densities and their associated functionality, while incorporating new thermal management technologies often requires major hardware redesigns, which has significant cost implications. In this work, we explore the use of air-cooled heat sinks incorporating three-dimensional features, so-called three-dimensional heat sinks (3DHS), that enhance heat transfer through a number of different physical mechanisms, as an approach to further extending the limits of air cooling. An ultra low profile (5.7 mm) heat sink application is targeted due to the significant thermal challenges associated with restrictions on heat sink height. We also present details on a novel manufacturing method that has significant cost advantages over other fabrication methods such as investment casting and direct metal printing. Experiments on 3DHS and conventional heat sink are conducted in a wind tunnel test apparatus as a function of inlet air mass flow rate and flow bypass above the heat sinks. The experimental results show a strong correlation between heat sink permeability and thermal performance, as measured by heat sink thermal resistance versus ideal pumping power. The results also illustrate the important effects of flow bypass on heat sink performance. The best performing 3DHS design is observed to have up to a 19% improvement in thermal performance relative to a conventional parallel fin heat sink of the same form factor. Comparison of the experimental results with finite-volume simulations of the laminar, steady equations for mass, momentum and energy transport shows good agreement for heat sink thermal resistance and pressure drop across the heat sink. For the case where the fluid flow is modeled as transitional and steady, there is a greater discrepancy between simulations and experiments, suggesting that the experimental flow conditions are predominantly laminar.

Simultaneous thermal/flow characterization of thermal interface materials

Thermal interface materials used for dissipation of heat in electronic assemblies are often exposed to working conditions that may change the mechanical as well as thermal behavior over time due to both intrinsic change as well as environmental influence. Currently, among thermal interface material solutions, particle filled greases are common. For these materials, the relation between observable change in its mechanical behavior and degradation of thermal performance is not well understood. This may be attributed to challenges associated with characterization of thermal resistance while the sample is subjected to cyclic thermal and/or mechanical loading. An important goal of the present work is to study the change in flow behavior and its effect on thermal performance of the thermal interface material under varying environmental conditions. Squeeze flow characterization is carried out on a custom built tester while simultaneously measuring thermal resistance. The change in flow behavior and the consequent change in thermal performance is tracked using a custom-built tester. The tester enables simultaneous measurements under a single setup. For different material sets, this relation may be different and so a quick characterization technique as the one proposed is desired. Test cases for commercially available particle filled thermal greases are presented where the Non-Newtonian flow behavior and thermal properties are obtained using the proposed technique.

Expert-prescribed weighting for support vector machine classification

Support Vector Machines (SVM) are a family of algorithms that are used in classification and regression tasks. Often, multiple SVMs are combined in a coding scheme to provide multi-class classification capabilities. Generally, multi-class classification systems are evaluated on their accuracy of producing a correct coding by using test data and successful predictions are counted as a percentage of the whole, assuming that the test data set is a “good” representation of what the classification algorithm will see in its applied use. However, in practical applications, there may be situations where certain mistakes/confusions in classification are inconsequential to system operation. In this work, a method for integration of expert-defined allowable confusions into SVM systems is introduced, with an example implementation in a least squares support vector machine (LS-SVM) tested on industrial data, and shown to improve overall performance of a multi-class classification system when an appropriate performance measurement method is formulated.

Syringe Position Control for Back Pressure Modulated Drop Volume in Functional Inkjet Printing

Inkjet printing technologies have been common and well developed over the past few decades, and more recently have gained significant acceptance in functional printing and additive manufacturing applications. Control of dot gain in the deposition process is a desirable capability for a printing system from the perspective of process control and throughput, and preliminary data suggests dot gain and drop volume can be controlled in inkjet systems through manipulation of the reservoir back pressure. In order to help facilitate further exploration, the design of a back pressure control system is proposed, and the system modeled, with linear and nonlinear control designs proposed and compared in simulation for this nonlinear plant application, where the nonlinear control design, a sliding mode controller, outperforms the tested linear control design. NOMENCLATURE A Syringe surface area, m2 Vp Piping volume, m3 Vs Syringe volume, m3 Vt Total reservoir volume, m3 PR Reservoir (absolute) pressure, Pa PA Ambient pressure, Pa Pb Back pressure, Pa x Syringe deflection from initial position, m x0 Initial syringe position, m Vt,initial Total reservoir volume at x = 0, m3 PR,initial Initial reservoir pressure, Pa n Number of moles of gas (air), mol R Ideal gas constant, J/(K*mol) T Air temperature, K Kg Gas right-hand side constant, N*m Fs Syringe nonlinear spring force, N Ff Syringe friction force, N Fc Coulomb friction force, N Fe ”External force” for friction model, N Fv Voice coil actuator force, N Fk Voice coil spring return force, N c Syringe viscous damping coefficient, N/(m/s) Kc Voice coil return spring constant, N/m ms Syringe plunger mass, kg ma Voice coil armature mass, kg mt Total mass, kg Proceedings of the ASME 2014 Dynamic Systems and Control Conference DSCC2014 October 22-24, 2014, San Antonio, TX, USA

Squeeze flow characterization of particle-filled polymeric materials through image correlation

Particle-filled polymeric materials are common choices for thermal interface materials (TIMs). During assembly, TIMs are applied between the surfaces across which heat must be transported. The goal of the present work is to develop a test procedure that is consistent with the length scales characteristic of real thermal interfaces. A mechanical tester (Instron 5848 Micromechanical Tester) is employed with closed-loop capacitive sensor-based control. During the squeeze flow process both direct (capacitive sensor, glass gauge encoder, and load cell) measurements and indirect (large working depth lens, video camera) measurements are made in order to record transient and steady-state loads and displacements. Discrete two-dimensional Fourier transform-based digital image correlation methods are used with microscope images to track surfaces relative to each other. Parameters describing Herschel-Bulkley and Bingham models are extracted using experimental data. Transient measurements are used to determine viscosity and strain-rate index, and the steady-state bond-line thicknesses corresponding to various loads are used to determine the yield stress. Advantages of the testing method include non-destructiveness, relative ease of deployment, and the ability to characterize materials at realistic assembly conditions.

Topological Design Optimization of Nested Channels for Squeeze Flow of Thermal Interface Materials

Thermal interface resistance remains a bottleneck for thermal transport in electronic systems, comprising a significant portion of overall system thermal resistance. Performance of thermal interface materials (TIMs) is largely dependent on the bulk thermal conductivity of the TIM but also on the bond-line thickness (BLT) of the applied material as well as interfacial contact resistances. Recently, Hierarchically Nested Channels (HNCs), created by modifying the surface topology with hierarchical arrangements of microchannels in order to improve flow, were proposed to reduce both required squeezing force and final BLTs in interfaces. In this paper, a topological optimization framework that enables the design of channel arrangements is developed. The framework is based on a resistance network approximation to Newtonian squeeze flow. The approximation, validated against finite element method (FEM)-based solutions, allows efficient, design-oriented solutions for squeeze flow in complex geometries. A comprehensive design sensitivity analysis exploiting the resistance network approximation is also developed and implemented. The resistance approximation and the sensitivity analysis is used to build an automated optimal channel design framework. A Pareto optimal problem formulation for the design of channels is posed and the optimal solution is demonstrated using the framework.© 2010 ASME