Research Vault

Mechanical + Microcontroller Project Catalog

Properties1
tagsresearch, hardware, mechanical, microcontroller, project-ideas, 3d-printing, laser-cutting

16 min read

Mechanical + Microcontroller Project Catalog

A deep-research roundup of medium-to-high-complexity projects that combine mechanical build (3D printing, laser cutting, gears, linkages, cams) with a microcontroller (ESP32, Arduino, RP2040, RPi). Curated for someone who has a split-flap display and a QLOCKTWO-style word clock under their belt, with a Bambu-class FDM printer and a diode-laser cutter on hand.

Every project is tagged with three axes so you can pick by appetite:

AxisLowMedHigh
Build time1 weekend1–4 weekends1–3 months
Cost (parts only)<$50300$300+
Skill ceilingsoldering + Arduino basicsstepper drivers, I²C, multi-module bring-uppower electronics (HV, drivers), firmware architecture

Start-here rule: pick a project where your weakest axis (usually power electronics or motion tuning) is exactly what you want to learn. The mechanical/3D side is rarely the blocker — it’s the motor tuning, signal integrity, and firmware state machine that separates a finished shelf-queen from a half-built drawer project.

Category Index

1. Mechanical alphanumeric / pixel displays

The natural next step after a split-flap display. Each one solves the “I want a non-flat display” problem differently — what you pick depends on whether you want electromagnets, rotating servos, or POV.

1.1 Flip-Dot Display (DIY controller version)

  • What: Reclaimed/cheap flip-dot panels (Alfa-Zeta 7×7, 14×24, or generic Bulgarian 28×14) driven by your own ESP32 controller instead of the OEM one.
  • Why it’s great: Industrial bistable dots that hold position with zero power. Insanely satisfying tactile click.
  • Complexity: Medium-High. 20V pulse driver, current-limited darlington/SPI shift-register chain, careful in-rush limiting (H-bridge per column, not per dot). One burned chip per amateur on average — current sink/source is the #1 failure.
  • Cost: 300 for the panels + 60 driver board.
  • Time: 2–6 weekends for a 7×7 clock; scaling beyond that quickly becomes a multi-month power-distribution exercise.
  • Refs:
  • Skill ceiling: Power electronics, level shifting, impedance vs dot count (48 boards in series will not all flip — you must feed both ends).

1.2 Moving Pixel LED Clock (Erich Styger)

  • What: A grid of tiny stepper motors (28BYJ-48 or NEMA8) each push an LED-backed pixel up out of a black panel. LEDs shine through the raised pixel; lowered pixels look like black voids. RGB LED behind each.
  • Why it’s great: Has the same “physical pixel” feeling as split-flap but scales more easily (one cheap 28BYJ per pixel vs NEMA per character). 18-month dev project for Styger, but the learning is reusable.
  • Complexity: High. Lots of small 3D parts, multiplexing challenge (you’ll need 4–8 daisy-chained shift registers or an LED driver per row), stepper tuning, mechanical alignment.
  • Cost: 400 depending on pixel count.
  • Time: 1–3 months for a meaningful grid (8×16 minimum).
  • Refs:
  • Skill ceiling: Multi-channel stepper firmware (AccelStepper or custom), PCB design if you want it to scale.

1.3 3D-Printed Electromechanical Pixel (Cooley / Coolfactor “Pixel”)

  • What: Each pixel is a tiny servo rotating a propeller-like vane inside a black casing. Rotated one way = blank, the other way = reveals a colored face. Greyscale via intermediate positions is the brilliant trick.
  • Why it’s great: Pure 3D-printed mechanism, no coils, no HV. 64 pixels makes a proper matrix display.
  • Complexity: Medium. Manufacturing is the long phase (Cooley: each pixel is 7 printed parts + 1 servo + 2 nails). Servo jitter at intermediate positions is the only real challenge.
  • Cost: 200 (the servos are the bulk).
  • Time: 3–6 weekends, mostly print time.
  • Refs:

1.4 POV (Persistence of Vision) Propeller Display / Clock

  • What: A single-arm rotor with a column of LEDs (8–32). Spun at 1000+ RPM, the LEDs strobe in sync with the rotation to draw text/clock faces in mid-air.
  • Why it’s great: Single-sided, fast to build, 100% 3D-printable, can be done with just an Arduino Nano and a $5 motor. Sits nicely on a desk as a fan-blade clock.
  • Complexity: Low-Medium. The hard part is angular indexing (Hall sensor or optical interrupter), and balancing the rotor so it doesn’t shake itself apart. Software is straightforward.
  • Cost: <$30.
  • Time: 1 weekend (a usable version) up to 1 month (balanced, well-tuned, wall-powered).
  • Refs:
  • Skill ceiling: Just solid-state. Perfect “I haven’t done anything in a while” project.

1.5 Magnetophoretic / 3D-Printed Passive Magnetic Pixel Displays

  • What: An FDM printer with a magnetic-PLA filament + a small electromagnet matrix behind the print. Each printed “voxel” becomes a bistable magnetic switch. The matrix paints hidden patterns into a printed surface that only show when you bring a magnetic viewer near.
  • Why it’s great: Print your display instead of assembling it. Always-on, no power needed, doesn’t look like anything until activated.
  • Complexity: Medium-High. Magnetizer design is the trick; magnetophoretic patterns require careful FDM pause-and-swap (or multi-material printer).
  • Cost: 150 + magnetic filament.
  • Time: 2–4 weekends.
  • Refs:

1.6 PixelWeaver / rotating-cube pixel display

  • What: Each pixel is a small cube with two black faces and two white faces, rotated 90° by a tiny mechanism to toggle. PixelWeaver by Chris Fenton uses an X/Y string-driven gantry that drags a “turtle” that rotates individual pixels in a 2D grid — clever but complex.
  • Complexity: High (string-pulled-cartesian stage + per-pixel mechanism). More research project than weekend build.
  • Ref: chrisfenton.com/the-pixelweaver

1.7 Marble pixel clock (no electro-mechanics, just geometry!)

  • What: Steel balls held in 3×5 grids with backlit cells. Each “digit” is a 3×5 pixel font; a stepper-driven shuttle lifts/repositions balls in/out of cells on a schedule.
  • Why it’s great: Zero electromagnets — it’s all geometry. Quiet. Looks magical.
  • Ref: Designboom: mechanical marble clock

2. Clocks & time-based art

You already have a word clock — here’s what’s adjacent.

2.1 Nixie Tube Clock (IN-14, IN-18, Z568M)

  • What: Soviet-era neon display tubes. Each digit is a separate wire cathode glowing inside a gas-filled envelope. Driven at 170 V DC with multiplexed anode control.
  • Why it’s great: Genuinely warm, analog glow. The aesthetic is unmatched. Off-the-shelf kits mean almost zero mechanical work.
  • Complexity: Medium. The HV power supply is the only real hazard — everything else is straightforward 5 V logic + transistor arrays. Pick a kit that has a regulated boost converter + current limiting.
  • Cost: 250 for a kit (tubes alone are $5–15 each on eBay).
  • Time: 1–2 weekends.
  • Refs:
  • Skill ceiling: Just be careful around 170 V — this will kill a 9 V battery feeling and hurt you badly.

2.2 QLOCKTWO-style Word Clock (DIY)

  • What: 11×10 letter grid; an ESP32 lights the relevant words to spell out the time (“IT IS HALF PAST TWELVE”). You already know this one — these are the variations you haven’t done.
  • Why it’s great: Laser-cut diffuser + WS2812 strip + ESP32. This is your bread-and-butter toolchain.
  • Variation A: Stainless steel face — laser-cut brushed stainless, LED matrix bonded behind. Vigue’s version uses a custom PCB with ESP32-S3. (ref)
  • Variation B: Round word clock — concentric rings of letters instead of a square grid. Rare, beautiful.
  • Variation C: Multilanguage — ESPWortuhr supports German/English/French/Spanish layouts from a single sketch. (ref)
  • Variation D: Web-configurable + OTA — ednieuw/Arduino-ESP32-Nano-Wordclock style. Time zone, brightness curve, animation. (ref)
  • Complexity: Low–Medium (you’ve done this). Variations are mostly mechanical: diffuser material choice, stenciling method, case design.
  • Refs:

2.3 Flip-disc clock / split-flap clock

You have these. The next step is a flip-disc clock (each digit is a 7-segment grid of magnetic dots that click between two colored faces) — combines the dot flipper controller skill above with a 7-segment clock front end.

3. Kinetic art & automata

The “I want a desk object that doesn’t have a screen” category. These lean more on the mechanical/3D side and use microcontrollers mostly for sensors + drive patterns.

3.1 Sisyphus-style Kinetic Sand Table

  • What: A steel ball magnetically dragged through a thin layer of fine sand by a gantry hidden under the table, drawing endless geometric patterns.
  • Why it’s great: Endlessly watchable. Strong DIY community. Pattern generators exist (sandify.org) so you don’t have to design the art yourself.
  • Variation A (Sisbot, polar): Closest to the original. Two motors driving a single arm with a magnet on the end. Two-axis kinematics. Compact, beautiful, more math.
  • Variation B (XY ZenXY): Cartesian X/Y, more 3D-printable parts, easier mechanical build, slightly bigger. V1 Engineering ZenXY is the canonical kit; their V2 blog has the closed-frame EMT-conduit rail design that scales to coffee-table size.
  • Variation C (single-Arduino): Two NEMA17 + CNC shield + GRBL. Hackaday’s 2024 build used exactly this. (ref)
  • Complexity: Medium. Stepper tuning + GRBL/Marlin config + leveling the sand bed + finding non-static sand. Do not skip leveling.
  • Cost: 400 (Z-axis ball magnet is a $5 N52 + something to attach it to the carriage).
  • Time: 2–6 weekends.
  • Refs:

3.2 Marble Machines & Chain Lifts

  • What: 3D-printed marble runs lifted by an Archimedes screw or chain lift, dropping through tracks and ramps. From a desk toy to a wall-sized sculpture.
  • Why it’s great: Pure mechanical fun; micro is optional (a touch sensor + motor speed control is the typical scope). MakerWorld and Printables have hundreds of STLs.
  • Top picks:
  • Complexity: Low-Medium. Mostly print-and-assemble. Worth doing once just to understand how 3D-printed kinematic chains fail.
  • Cost: <5 geared motor.
  • Time: 1 weekend (small toy) to 1 month (large sculpture).

3.3 Cam-driven Automata (Mechanical Toys)

  • What: Hand-cranked or low-RPM motor driven scenes where cams (off-center rotating shapes) push followers that animate characters. Classic Cabriole / Theo Jansen lineage.
  • Why it’s great: Once you understand cam math, you can build anything that cycles. Add a sensor + an Arduino and the cam-driven scene becomes a teaching tool (count rotations, sync with sound, etc.).
  • Refs:
  • Complexity: Low (mechanical) to Medium (if you add a micro for synchronization).
  • Cost: <$20.
  • Time: 1–2 weekends.

4. Drawing & plotting machines

Plotter art is a huge maker rabbit hole. The aesthetic is in the lines, not the resolution.

4.1 Pen plotter (XY, flatbed) — DIY vs AxiDraw

  • What: Two stepper-driven carriages (X+Y), pen up/down (servo or solenoid), drawing on paper from SVG/HPGL input.
  • DIY routes:
    • CoreXY / H-bot with linear rails — most precise, most expensive.
    • Belt-and-pulley on MakerBeam/V-slot — cheapest; tolerates flex, fine for pen art.
    • MPCNC (Mostly Printed CNC) — modular, low-cost, scales to A2+. thingiverse 2994920 is the classic design.
  • Buy vs build: The AxiDraw SE/A3 is 150–$300 if you have a printer to make the parts. Performance difference is small at A4 scale. (discussion)
  • Complexity: Medium. GRBL + a CAD pipeline (Inkscape + axidraw extension or vpype + axibot) is the typical software stack.
  • Time: 2–4 weekends.

4.2 Vertical wall plotter (V-plotter / Polargraph)

  • What: Two steppers mounted at the top corners of a wall (or whiteboard), with string that suspends a pen gondola. The gondola position is computed from string length — a fun inverse-kinematics exercise.
  • Why it’s great: No Y-axis rigidity problems because gravity does the work. The pen hangs freely. Draws huge things.
  • DIY:
    • Minimal Pico Vertical Plotter — RP2040-based, ESPHome-friendly. (Hackaday.io)
    • Stringent, the $15 wall plotter — minimalist version. (Hackster)
    • Polargraph — the original open-source design (Sandy Noble), still the most documented.
  • Complexity: Medium. Inverse kinematics + string-length math + the gondola needs to be lightweight.
  • Cost: 150.
  • Time: 1–3 weekends.
  • Refs: Instructables: Vertical X-Y Plotter · Liz Melchor’s Wall Hanging Drawing Robot

4.3 Sand plotter / 3D pen plotter

If you really like the Sisyphus idea but want a plotter for actual pictures on flat surfaces, the same XY gantry with a pen (or extruder) becomes a flatbed plotter / 3D printer / CNC.

5. Walking machines & linkages

The mechanical engineering rabbit hole. Each project teaches a different kinematic principle.

5.1 Strandbeest (Theo Jansen mechanism) walking machine

5.2 Biped / quadruped walking robots

  • What: 3D-printed chassis + NEMA17 with cycloidal gearboxes + IMU for balance. Quadruped is the current frontier (Stanford Pupper, Open Dynamic Robot Initiative MIT Cheetah clones).
  • Complexity: High. State estimation, gait timing, balance control. A year-long learning project.
  • Skip if: you don’t already know what a Jacobian is.

5.3 6-DOF Robot Arm

  • What: Articulated arm with 6 servos or steppers + gripper. Pick-and-place, drawing, simple assembly tasks.
  • Why it’s great: Robotics staple — teaches kinematics, IK, inverse Jacobian. Lots of working designs to clone.
  • DIY routes:
  • Complexity: Medium-High.
  • Cost: 400 (cycloidal).
  • Time: 2–6 weekends.

6. Musical instruments

6.1 MIDI Laser Harp

  • What: Vertical array of laser beams (laser diodes + line generators or galvo-scanned single beam). Break a beam → trigger a MIDI note.
  • Why it’s great: Lasers + MIDI = instant showmanship. Jean-Michel Jarre played these live.
  • Complexity: Medium. Two main paths:
    • Multiple lasers + photoresistors — simple, one MIDI note per beam. 9–13 beams typical.
    • Galvo + single beam — one beam swept across the play area; position determines note. More dramatic, more involved.
  • Cost: 200.
  • Time: 2–4 weekends.
  • Safety: Class 3R lasers or below. Never class 3B or 4 in a home setup without proper engineering controls.
  • Refs:

6.2 Stepper-based percussion (auto-drummer, solenoid drum)

  • What: Electromagnetic solenoids striking physical surfaces (drum heads, metal bars, wine glasses) under MIDI control.
  • Why it’s great: Easy to make weird, organic sounds. Combination of 3D-printed linkage + simple 12V drivers.

6.3 Xenakis-style stochastic music box

  • What: Stepper-driven disc with pins plucking a tuned comb (like a music box, but programmable). Combine with probability logic in firmware to make algorithmic compositions.

7. Ambient / connected displays

These are the “lives on a shelf, talks to the internet” category — mostly electronics + mechanical case.

7.1 E-Ink Weather Display

  • What: ESP32 pulling weather + calendar + air quality from APIs and rendering on a 4.2” or 7.5” e-ink panel. Months of battery life.
  • Why it’s great: Always-on display with zero backlight, looks like paper. The case is the only mechanical part.
  • Complexity: Low (electronics) + Medium (if you design a beautiful 3D-printed/laser-cut case).
  • Cost: 100.
  • Time: 1–3 weekends.
  • Refs:

7.2 Smart / Magic Mirror

  • What: Two-way mirror + thin display panel (ideally USB-C powered 15.6” 1080p) + Raspberry Pi running MagicMirror².
  • Why it’s great: Wall-mounted home dashboard. Mostly a software + frame-building project — minimal electronics.
  • Complexity: Low-Medium. Lots of community modules (calendar, transit, weather, news).
  • Cost: 400 depending on display size.
  • Time: 1–3 weekends.
  • Refs: magicmirror.builders · Raspberry Pi slim smart mirror tutorial

7.3 Connected art frame (e-ink or OLED behind a picture frame)

  • What: Take an existing art print, add a thin e-ink/OLED behind a cutout, fetch art from an API daily. “Today’s painting” frame.

7.4 Mechanically animated clock displays (servo-arm clock, flip-disc clock)

  • Already covered above (sections 1.1, 1.2). Cross-listed here because the “always-on wall display + WiFi time sync + ambient feel” is the common thread.

8. How to pick

Match the project to what you want to learn, not what’s most impressive. The biggest project graveyard is people starting a moving-pixel clock because the videos are cool, hitting week 3 of stepper tuning, and abandoning it.

If you want to learn power electronics & high-current drivers

  • Flip-dot display (1.1) — you will learn current limiting, bulk caps, flyback, level shifting, by fire.
  • Nixie clock (2.1) — 170 V boost converter + HV safety practice.

If you want to learn multi-channel stepper / motion control firmware

  • Moving pixel clock (1.2) — AccelStepper at scale; or build your own Bresenham-based step queue.
  • Sisyphus / ZenXY sand table (3.1) — GRBL or Marlin kinematics, pattern buffering, leveling.

If you want to learn mechanism design / kinematics

  • Strandbeest (5.1) — 4-bar linkage math.
  • Cam automata (3.3) — cam profile design, follower geometry.
  • 6-DOF arm (5.3) — inverse kinematics, Jacobian.

If you want a fast win (one weekend, finish something)

  • POV propeller clock (1.4).
  • Strandbeest desk toy (5.1, hand-cranked).
  • Marble machine (3.2, print-and-assemble).

If you want to learn firmware + WiFi + API integration

  • E-ink weather display (7.1).
  • Magic mirror (7.2).
  • Any of the clocks with NTP + web UI.

If you want to learn art / aesthetic fabrication

  • V-plotter (4.2) — the lines themselves become the art.
  • Sand table (3.1) — algorithmic sculpture.
  • Word clock variations (2.2) — material science (diffusers, stencils).

9. Cameras & imaging

The cheapest path to a WiFi-connected camera is an ESP32 camera board + a printed enclosure. See the dedicated deep-research note: ESP32 + 3D-Printed Camera Projects — board/sensor selection, streaming constraints (MJPEG-only on S3), enclosure designs, and the power-engineering realities of trail cams.

Quick picks from that note:

  • Printer monitoring cam (fastest win): AI-Thinker ESP32-CAM (~$7) + snap-fit case + ESPHome — the cheapest documented OctoPrint camera.
  • Portable timelapse cam: XIAO ESP32-S3 Sense + magnetic-base housing, captures to SD, battery-pad powered.
  • Showpiece: retro point-and-shoot digital camera on XIAO S3 Sense (AVI with sound, on-device playback, wireless file manager).
  • Engineering challenge: PIR-triggered wildlife/trail cam — board-level deep sleep is the trap; expect to hard-switch power rather than trust ESP32 sleep modes.

Open questions for future notes

  • Magnetic pixel display (MagPixel) — practical feasibility with Bambu-class printers and magnetic PLA quality.
  • Split-flap → flip-disc transition — is a flip-disc clock worth it after split-flap?
  • Real-time GRBL on ESP32 vs Arduino Mega for the sand table.
  • Galvanometer-based laser systems vs multiple discrete laser diodes for the harp.

Graph View