From waste to micro-factory: 25 open-source projects for a TRIFAB

What if a reuse centre did not just collect objects, but became able to understand them, repair them, transform them — and even build the machines needed to transform them?

That question is the starting point of a long article by D. Dambreville, the author of the TRIFAB manifesto, published in French as TRIFAB : 25 projets pour un atelier qui fabrique ses propres machines. This guide adapts it for an international audience. It takes the third link of a TRIFAB — the fablab — and pushes it as far as it can go: a local, open-source micro-factory fed by what arrives at the recycling centre.

The principle: make the means of making

Take each element separately: a Raspberry Pi, a few Arduino boards, sensors, a CNC, a 3D printer, a laser engraver, recovered motors, circuit boards, aluminium profiles, wood, plastics, components from broken appliances, GitHub, free software, and an AI coding assistant. On their own, they are a collection of tools. Brought together in a coherent architecture, they become something else: a local infrastructure for design, repair, reuse and fabrication.

The traditional chain for used objects is linear: object → waste → sorting → recycling or disposal. A reuse centre introduces a break: object → reuse → repair → new user. A TRIFAB adds a third path:

object → dismantling → components → materials → design → fabrication → new object

And it can go further. A recovered motor can become part of a CNC. A circuit board can become the controller of a machine. An aluminium profile can become a frame. A broken printer can provide linear rails; a hard drive, magnets; a fan, an extraction system; a power supply, a lab bench supply. An old machine can provide enough parts to build a new one.

A TRIFAB should not only make objects; it should progressively make the means that allow it to make more objects.

This creates a particular dynamic — recovered resources → prototype → machine → new capability → new machine → micro-factory — in which each completed project increases the technical capital of the place. The goal is not absolute autonomy. It is cumulative local capability. The idea has a respected lineage in the open-source hardware world, from the RepRap project — a 3D printer designed to print many of its own parts — to open-source plastic-recycling machines shared by communities such as Precious Plastic.

The technical stack: nine building blocks

1. Raspberry Pi — supervision, interfaces, data logging.
2. Arduino and microcontrollers (including ESP32) — control of motors, sensors and actuators.
3. Sensors — temperature, current, weight, distance, cameras.
4. CNC — precise machining of wood, plastics and soft metals.
5. 3D printing — prototypes, spare parts, jigs.
6. Laser engraving and cutting — flat parts, panels, signage, templates.
7. Free software — FreeCAD for design, Blender for 3D, KiCad for electronics, Linux everywhere.
8. AI-assisted development — a coding agent such as Claude Code for software, automation and documentation.
9. GitHub and open documentation — the technical memory of the place.

The ninth block is the most important. A machine that is not documented remains a local achievement. A documented machine, whose plans, software, procedures and bill of materials are published, becomes a reproducible project.

Level 1 — Understand and recover

The first level turns the flow of discarded objects into knowledge and stock.

1. The smart component bank

A physical library of motors, bearings, screws, connectors, cables, sensors, fans, power supplies, switches, gears, profiles and plates — each with a photo, an internal reference, a quantity, a location, dimensions and, when available, a datasheet. A user can then ask: “Which available motors could suit this project?” and the system queries the database.

2. The reverse-engineering lab

A broken appliance becomes an object of study: dismantle it, photograph each step, identify the parts, measure, draw, document, understand. Appliance → dismantling → analysis → measurement → CAD → documentation → repair or redesign. Mechanics, electronics, electricity, computing, technical writing and scientific reasoning are developed at once.

3. The electronics recovery lab

Discarded electronics are a considerable reserve of components: motors, connectors, fans, heat sinks, buttons, cables, enclosures, switches. A dedicated station recovers, tests and classifies them, and feeds the inventory automatically.

4. The 3D scanner for recovered objects

A broken part is not necessarily useless; the problem is often knowing its geometry. Object → photographs → 3D reconstruction → model → clean-up → CAD → new part. Free tools such as Blender and FreeCAD, combined with photogrammetry libraries, can take part in this chain — particularly valuable for parts whose plans no longer exist.

5. The spare-part generator

Someone brings a broken part. Photograph → measure → identify → search documentation → model → prototype → test → fabricate. A 3D printer makes the prototype, a CNC a sturdier version, a laser a flat part — the process is chosen according to the material. This is where a TRIFAB becomes a local ally of the right to repair.

Level 2 — Build the machines

The second level is where the workshop starts building its own tools.

6. The self-buildable CNC

The CNC is a foundational machine: it machines parts with a precision impossible by hand. Building one is a major educational project in itself — design → calculation → mechanics → electronics → motors → software → calibration → machining. An Arduino-class controller drives the motors (open-source firmware such as GRBL exists for this), a Raspberry Pi provides the interface, FreeCAD the design.

7. A large-format 3D printer

Why always buy a large machine when you can study building one from profiles, motors, lead screws, guides, boards, suitable power supplies, printed and machined parts? It can produce furniture, prototypes, moulds, enclosures, educational pieces, technical parts and jigs — and some of its own parts can be made in the TRIFAB.

8. The laser engraver and cutter

Signage, panels, templates, enclosures, flat parts, models, teaching aids. A laser requires particularly serious design for beam enclosure, interlocks, emergency stop, fume extraction, fire protection and operating procedures. Safety is not an accessory of the project; it is part of the project.

9. The universal machine with interchangeable tools

One mechanical platform, several heads: CNC spindle, laser, deposition head, pen, cutting tool, camera. Chassis, motors, guides, controllers, power supply and software are pooled; only the tool changes. Fewer machines, more operations.

10. The robotic arm built in the TRIFAB

Aluminium or steel structure, 3D-printed and CNC-machined parts, recovered motors and bearings, Arduino or ESP32 electronics, Raspberry Pi supervision. The arm learns to move a part, grip an object, place a component, repeat a sequence, feed a machine. It becomes itself an educational and industrial object of the TRIFAB — the concrete proof that mechanics, electronics, computing and digital fabrication converge.

Level 3 — Close the material loop

The third level turns waste back into raw material — and measures it.

11. Recycled plastic to 3D-printing filament

Recovered plastic → sorting → cleaning → shredding → extrusion → filament → 3D printing → new object. The process requires rigorous characterisation of materials and careful risk management, but pedagogically it shows something fundamental: a material considered as waste can become a raw material again. Measure the incoming mass, transformed mass, losses, energy consumed, filament produced and objects made: the loop becomes measurable.

12. The battery characterisation bench

Recovered batteries require a particularly cautious approach. With suitable equipment and procedures, a bench measures voltage, current, capacity, temperature and behaviour under load, then classifies cells by condition. Damaged, swollen or abnormally behaving lithium batteries must be excluded from ordinary handling and follow specific procedures. The aim is less to “save every battery” than to understand, characterise and secure the flows.

13. The energy lab

How much energy does it take to make this part? How much material have we diverted from waste? How much has really been transformed? How much human time was needed? Instrumenting machines for instantaneous power, energy consumed, usage time, consumption per project and per kilogram transformed turns environmental talk into verifiable measurements.

Level 4 — Instrument the place

The fourth level gives the TRIFAB a nervous system.

14. TRIFAB OS

As machines and parts multiply, a question arises: what does the TRIFAB own? TRIFAB OS is not an operating system in the strict sense but a technical management platform for objects, components, materials, machines, tools, projects, stocks, users, documentation, maintenance and measurements. Each item gets an identifier — TRIFAB-MOT-000183 for a motor, TRIFAB-SEN-000421 for a sensor.

15. The TRIFAB IoT network

Each machine becomes a data source: a CNC reports temperature, consumption, cycles and maintenance; a 3D printer its temperature, duration, filament and status; a laser its cycles and extraction. A Raspberry Pi centralises the data (for example over MQTT), and the TRIFAB builds a simplified digital twin of itself.

16. An open-source weather station

Temperature, humidity, pressure, light, rainfall, CO₂, particulates, noise. Sensors → Arduino/ESP32 → Raspberry Pi → database → charts → open data. The TRIFAB becomes a small measurement laboratory for its territory.

17. The smart greenhouse

Built with recovered materials and controlled electronically: sensors for temperature, humidity, light, soil moisture, water level and CO₂; actuators for pump, fan, valve, lighting and vents. A simple loop — measure, analyse, decide, act, measure again — is a concrete introduction to cyber-physical systems.

18. The water treatment test lab

Experimental systems to measure, filter, control, automate and compare processes, with a Raspberry Pi logging measurements and an Arduino driving sensors and actuators. The project must strictly separate technical experimentation from the production of water intended for human consumption.

19. The miniature smart house

A model house with lighting, temperature, ventilation, shutters, presence detection and consumption monitoring, built with ESP32, Arduino, Raspberry Pi, MQTT, Home Assistant and Node-RED. Visitors and pupils immediately see the consequences of a line of code.

Level 5 — Automate, learn and replicate

The fifth level is where AI, automation and open source meet.

20. The AI-assisted sorting robot

A camera observes an object; the system looks at shape, colour, dimensions, apparent material, connectors, screws, type of enclosure, visible components, and proposes a category: wood, metal, plastic, electronics, cable, motor, sensor, hardware, complete object, spare part. The first version does not need to be an autonomous robot: it can simply assist the human operator. AI as a sorting aid, not a substitute for the person.

21. The TRIFAB AI Engineer

This is where an AI coding agent such as Claude Code becomes particularly interesting. AI should not be seen as a magic machine that fabricates objects directly. It is rather an interface between human intention and the technical tools of the TRIFAB. Given photographs, measurements, constraints, available materials and components, and a goal, it can help to analyse the problem, search existing projects and repositories, read documentation, compare architectures, produce a first design, generate code and configuration files, prepare documentation and a bill of materials, and plan tests. The human remains responsible for design, validation and safety.

22. The mini robotic production line

Stock → feeding → robot → CNC → laser → 3D printing → assembly → inspection → storage. The point is not to copy an industrial plant, but to understand — and teach — how a production chain is designed, instrumented and automated.

23. The micro-factory

All the equipment brought together into one coherent chain: recovery → sorting → identification → dismantling → storage → design → fabrication → assembly → test → documentation → reuse. The Raspberry Pi supervises, Arduinos control physical devices, the CNC makes mechanical parts, the 3D printer prototypes, the laser cuts panels, free software designs and drives, GitHub keeps the data, the AI assistant helps develop and automate.

24. A TRIFAB that builds part of itself

The 3D printer makes parts for the CNC; the CNC cuts plates for the laser; the laser cuts panels; Arduinos automate a new machine. Part of the productive capacity of the place serves to increase that same capacity: the self-extension of the workshop.

25. The open-source TRIFAB kit

The most ambitious project is not a machine. It is the documentation of the whole system, so that another territory can progressively reproduce a TRIFAB. A central repository might look like this:

trifab/
├── machines/
├── electronics/
├── robotics/
├── software/
├── cad/
├── documentation/
├── safety/
├── education/
├── procedures/
├── maintenance/
└── inventory/

Each machine comes with plans, bill of materials, wiring diagrams, firmware, software, procedures, documentation, maintenance and calibration notes, indicative costs, build time, difficulties and alternatives. A local achievement becomes a technical commons.

Safety and responsibility come first

Several of these projects involve real hazards: lasers, high-current batteries, rotating machines, plastic fumes, water. Three rules apply across the board:

• The safety design is part of the project, not an add-on: enclosures, interlocks, emergency stops, extraction, fire protection, procedures, training.
• Regulated and dangerous flows stay separate, and experiments are never confused with products for the public (drinking water, electrical goods sold without testing).
• AI accelerates design; it does not replace human validation, testing or safety rules. A bad part generated quickly is still a bad part; a bad procedure automated is still a bad procedure. AI must sit inside a validation chain.

Documentation: GitHub as technical memory

GitHub should not be seen as a place to dump code but as the technical memory of the TRIFAB. Each project can carry a README, licence, bill of materials, CAD files, schematics, firmware, software, documentation, photos, tests, safety notes, maintenance and a changelog. An object made in a TRIFAB can therefore have a complete technical history.

The rule is simple: document what was done, what worked, what did not, why, how to reproduce it, how to repair it, how to improve it. Failure itself becomes data: a failed prototype is useful documentation for the next project — and protects the place against the departure of the one person who knew.

A technical school built on real objects

Instead of teaching mathematics, physics, mechanics, electronics and programming separately, start from a real object: “Let’s build a mobile robot.” Geometry, speed and coordinates; forces and energy; wheels, axles and gears; motors, power and sensors; programming; communication; CAD; 3D printing, CNC and laser; documentation on GitHub. One project becomes a multidisciplinary learning path. A TRIFAB can offer progressive levels:

The 25 projects at a glance

Conclusion: a factory that learns

A TRIFAB can start modestly: a Raspberry Pi, a few Arduinos, a 3D printer, a small CNC, a properly secured laser, some sensors, a Linux computer, free software, a GitHub account — and a lot of recovery. Its real ambition is not the number of machines in the building. It is the capacity of the place to turn what it receives into new technical capabilities.

An object becomes a part. A part becomes a prototype. A prototype becomes a machine. A machine increases the capacity of the TRIFAB, which can then build other machines — and every experience is documented so that others can reproduce it. Four memories work together: material (parts, materials, machines, tools), digital (software, models, files, databases), documentary (plans, procedures, tutorials, lessons learnt) and human (skills, know-how, experience, mistakes, innovations).

We move from “we have a workshop” to “we have a system able to learn, repair, design, make and pass on”. A recycling centre recovers. A reuse centre re-employs. A TRIFAB transforms — and when it documents everything it learns, it passes on the capacity to transform.

Want the bigger picture first? Read What is a TRIFAB? The complete guide, then see how to launch a pilot site.

• Open source
• Fablab
• AI
• Robotics
• Repair