Motion Control · System Identification · Force Control
I build high-performance motion systems through modeling, frequency-response analysis, simulation, controller design and machine validation.
Complete evidence chains: problem → model → identification → design → validation → reusable tools.
force control
Machine TestedModeling and validation of a nonlinear contact-force loop for a bonding head Z-axis: unilateral contact model, friction compensation, and analysis of the delay between force command and displacement response.
View project
system identification
Machine TestedTwo-input two-output frequency-response identification of a gantry stage: linear and yaw excitation, 2×2 FRF matrix, coherence assessment, DAC input correlation and its impact on decoupling.
View project
motion control
SimulationA structured sizing methodology for motion axes: from motion requirement through profile kinematics, force/torque, power and thermal margin, encoder quantization, delay budget and resonance target — to predicted motion performance. Implemented as a tested TypeScript module and an interactive calculator.
View project
Each capability links to the project or note that demonstrates it — claims backed by evidence, not keywords.
Motion Control
Servo loops from model to machine: trajectory, feedforward, loop shaping.
System Identification
FRF measurement campaigns that survive scrutiny: excitation, Welch, coherence.
Frequency Response Analysis
Reading resonance, anti-resonance and coupling out of measured spectra.
Force Control
Contact-force loops with unilateral contact, friction and mode switching.
Mechanical Dynamics
Contact stiffness, preload, friction and clearance — and their control impact.
MATLAB / Simulink
Nonlinear plant models, identification pipelines and controller verification.
C++ / Engineering Tools
Real-time implementation and tested engineering software.
Size a motion axis from move profile to resonance target: peak velocity/acceleration, force margin, mechanical power, encoder error budget and recommended minimum resonance frequency.
Load exported FRF data (CSV) and inspect Bode magnitude/phase, coherence and peaks directly in the browser.
Generate linear/yaw transformation formulas for dual-drive gantries from geometry and port numbering, with live preview and copyable output.
Upload time-domain step data and an error band; get settling time, overshoot and a marked-up response plot — all client-side.
A force command is issued, the DAC output changes — yet detectable displacement appears only ~150 ms later. A structured checklist of physical and signal-chain causes, and how to separate them.
Segment length, window, overlap and averaging count decide whether an FRF is smooth and trustworthy or noisy and biased. A practical recipe with the trade-offs made explicit.
A 280 Hz anti-resonance that moves between measurements is telling you something about contact stiffness, preload and clearance. A debugging plan for unrepeatable anti-resonances.
Why a dual-drive gantry is naturally described by sum and difference coordinates, what the transformation actually assumes, and how mode separation simplifies both identification and control.