ArchivedFabricationEmbeddedManufacturing

Harvard Digital Fabrication

PHYS-S-12. Full coursework archive.

RoleStudent / Builder
Timeline2023
UpdatedMarch 11, 2026
Duration11 weeks
Final ProjectModular TVC Rocket
InstitutionHarvard Summer School
Overview

PHYS-S-12 at Harvard Summer School: eleven weeks covering laser cutting, CNC milling, casting, microcontroller programming, embedded sensor integration, and machine building, culminating in a modular TVC rocket. Each week is documented with build notes, CAD files, and process photos. The full course site, including week-by-week writeups and the final project breakdown, is embedded below.


The problem

Complete a full fabrication curriculum covering CAD, laser, CNC, casting, electronics, and embedded systems, culminating in an integrated final project under real time pressure.

Constraints
  • Weekly scope limits. Each domain had to be understood fast enough to produce a result.
  • Chemical propulsion was safety-prohibited. TVC project pivoted to electronics-based stabilization.
  • Limited equipment access required adapting designs to whatever machines were available on the day
Design decisions
  • TVC gimbal used PLA's inherent flexibility as the compliant hinge mechanism. Eliminates the backlash you get from traditional linkages.
  • Twist-fit modular rocket sections for field assembly without tools
  • MPU-6050 IMU mounted at rocket base with OLED display for real-time orientation readout during bench testing
Build process
  1. 01

    Week 2: Laser-cut press-fit cardboard assembly. Kerf compensation was the key lesson.

  2. 02

    Week 3–4: Kinematic sculpture with gear-reduced DC motor and Arduino Uno, later upgraded with herringbone gears from Fusion 360 spur gear add-in

  3. 03

    Week 5: TVC mount prototyped on Prusa. First version failed on fit, second version used Polycam photogrammetry for reference geometry.

  4. 04

    Week 6: MPU-6050 6-axis IMU calibrated against angle finder, compliant mechanism contact sensor designed for launch detection

  5. 05

    Week 7: OLED display on I2C showing live gyro data, oscilloscope characterization of output timing

  6. 06

    Week 8: Roland SRM-20 CNC milling in machinable wax. First attempt failed when wax melted onto the end mill due to speed. Surface roughing fixed the adhesion issue.

  7. 07

    Week 9: ESP32 ESPnow peer-to-peer communication, both boards exchanging data via MAC address registration

  8. 08

    Final: Modular TVC rocket, Bambu Lab 3D printer for final parts, collaborative build with Hiranya


Result

Complete coursework documented. TVC rocket final project presented. All weekly builds archived with full process notes.

What went wrong / what I learned
  • CNC setup parameters matter more than the geometry. Losing a wax mold to a spindle speed error is an expensive lesson.
  • Photogrammetry (Polycam) outperformed the dedicated scanning hardware (RevoPoint) for complex organic geometry
  • Modular assembly needs to account for cumulative tolerance. A twist-fit that works at the bench can bind in the field.

Tools & Methods

Fusion 360Arduino Uno R3ESP32 (ESPnow)MPU-6050 IMUOLED display (I2C)Roland SRM-20 CNCPrusa / Bambu Lab 3D printersLaser cutterPolycam photogrammetryOomoo silicone casting

Specs

Duration
11 weeks
Final Project
Modular TVC Rocket
Institution
Harvard Summer School

External link

View site