Mechanical
engineering
with a bias for building.

The problem. Polycarbonate (PC and PC-CF) is a structural-grade thermoplastic, but the temperature gradient between extrusion and ambient causes warping and delamination. To print PC reliably, the chamber needs to hold ~80 °C steady-state — well above what an off-the-shelf printer can do.

What I built. An integrated thermal management system on a custom-built CoreXY printer: reflective insulation lining, dual 5015 blowers under the bed for forced convection, and a 100 W PTC chamber heater module mounted to the lower wall.

Why build them. Most off-the-shelf headphones cost $$$ for the wood cup and the brand. The driver itself is often the easy part. I wanted to learn the acoustic side — how the cup, the baffle, and the damping pattern shape what you actually hear — by designing and building the whole thing from scratch.

The cup. Walnut on the outside, a 3D-printed structural baffle on the inside. Wall thickness, internal volume, and the bevel angle behind the driver are all tuned to push resonant nodes out of the audible band.

The baffle. A printed spoke pattern that doubles as structure and as damping geometry — the spoke spacing controls how much air gets coupled to each part of the driver, which is the lever you pull to shape mid-treble.

The gap. Hobbyist CNC routers come in two flavours: cheap toys that can't cut aluminum, or industrial machines priced out of a home shop. The team's brief was a router that fits in a one-car garage, costs under CAD $4,000, and actually cuts aluminum with a safety factor of 4.

My contribution. Force calculation and structural analysis. I derived the worst-case 280 N cutting force from spindle power, ran ANSYS static structural on the 4080 aluminum gantry, and confirmed a 12.5× safety factor before we committed to the extrusion size.

Final architecture. Y-moving gantry, dual SFU1605 ball screws on X and Y, T8 lead screw on Z, MGN12H linear rails on all three axes. The motion-system tradeoffs — backlash vs. cost vs. scalability — came from a weighted decision matrix across six candidates.

Bill of materials. 4×4 ft work area, ±5 µm theoretical resolution per microstep, CAD $3,717 total — all parts sourceable from Misumi, Hiwin, and McMaster.

The work. Designing non-structural composite airframe panels for UTAT's liquid-propellant rocket. Modular geometry that integrates with the propulsion bay and gives the test team rapid access during pre-launch — without compromising aerodynamic stiffness.

The simulation work. ANSYS FEA against flight loads — shaved 15% mass off the panels while meeting stiffness targets.

The manufacturing pivot. Pioneered the team's transition from positive to negative-mould fabrication for these panels. Halved production time (2 weeks → under 1) and gave us cleaner aerodynamic surfaces with fewer post-cure defects.

Why. A small, beautiful exercise in tolerance stacking and precision turning. The whole pen is a single linear mechanism — a piston riding inside a transparent reservoir, sealed by an O-ring, advanced by a threaded collar — but every dimension has to be right or it leaks, binds, or drips.

What's interesting. Designing for the manual lathe means thinking in operation order: which features get cut first, which surfaces become datums, where the tolerance has to live. The CAD looks simple. The setup sheet is where the engineering shows.

What it does. A motorized linear slider for macro photography — a Raspberry Pi drives a stepper motor through custom Python software that automates focus stacks, panoramic tracks, and time-lapse shoots. Sub-millimeter positioning, zero step loss across 100+ continuous cycles in endurance testing.

Designed for FDM. Carriage and rail topology optimized for desktop 3D printing — 30% less filament than the first version, and stiffer where it matters: mid-span, where vibration becomes visible at long exposures.

Build status. Mechanism prototype is running; mounting hardware and the final electronics enclosure are still in progress.