Blog · · 5 min read
How to make a machine: seven steps from idea to a prototype that runs
How do you make a machine from scratch? Seven steps from idea to working prototype: the job, a physics check, the working principle, CAD, electronics, parts and safety. With a worked lift example and where AI saves time.
Many machines that never work fail in the first hour, when someone starts drawing the machine instead of writing down the job.
Here is a sequence that works for a toy robot as well as for a workshop lift.
1. Write down the job, not the machine
"A machine that lifts boxes" is an idea. "Lift a 20 kg box by 0.5 m in 2 seconds, 200 times a day, in a dusty workshop, with a mains socket nearby" is a job. Only the second one can be checked.
A good job description has numbers for loads, distances, speeds and cycles. It names the environment, the power you have, how many units you plan, and what must never happen, such as the load dropping when the power fails. Write down what success looks like, because step 7 will measure against it.
The business model belongs here too. A washing machine that customers pay for per wash, instead of buying it, is designed differently from one that is sold once. Washing machine as a service shows how much that one sentence changes.
2. Check the physics before you open CAD
Ten minutes of arithmetic saves weeks of drawing. For the lift above:
- Work per lift: 20 kg × 9.81 m/s² × 0.5 m ≈ 98 J
- Average power: 98 J / 2 s ≈ 49 W at the load
That is before losses. A plain lead screw often turns less than half of the motor's work into lift. At about 40 %, the motor has to deliver around 120 W. A ball screw at about 90 % needs roughly 55 W. Then add the torque to accelerate the load, and a margin for friction you have not measured yet.
The same numbers show a trade-off that is expensive to discover later. The efficient ball screw can be back-driven, so the load sinks when the power is off unless a brake holds it. A plain lead screw with a small lead usually locks by itself. How to size a motor walks through torque, speed, gearing and holding in detail, and Will my hardware idea work? shows a full feasibility check on a real idea.
3. Choose the working principle
Now decide how the machine works. For the lift, that could be a screw, a belt, a scissor mechanism, a rope and winch, or a pneumatic cylinder. Sketch two or three, then score them against the job from step 1: force, speed, holding, noise, cost, and how hard they are to build.
Sometimes the principle is the whole idea. Roadstoy, a remote-controlled toy car on makemachine.ai, drives on an electric motor but can briefly jump and flip with compressed air. A short, hard push is far easier to store and release as gas pressure than to produce with a motor that fits in a toy.
4. Build the CAD around the parts you buy
Start the 3D model with the parts you will not make yourself: motors, bearings, rails, screws, the power supply, the circuit boards. Their real dimensions, from the datasheet or the maker's model, fix where everything else can go.
The model then grows in four stages, from a rough layout to something a workshop can build from. CAD models for purchased parts shows where to get the bought ones. On makemachine.ai the model is written as code, so a longer screw or a bigger motor regenerates the parts instead of redrawing them.
5. Add electronics and control
Most machines need a power supply sized for peak current, a motor driver, a controller, sensors such as end stops, and an emergency stop that cuts power to the motion without relying on software. Build the first version from development boards and modules. Design your own circuit board once the wiring works and you know what you need. Can AI design a PCB? covers that step.
6. Decide what to make and what to buy
Each custom part needs a process. 3D printing is fast for the first fits, CNC machining or sheet metal suits parts that carry real load, and 3D print, CNC or mold? compares them. Put everything into one bill of materials with quantities, suppliers and lead times. The first complete list is usually where the budget surprise happens, as What will my prototype cost? shows.
7. Make it safe, then test against the job
A machine moves, so it can pinch, crush, cut, burn or shock. Do a risk assessment while the design can still change: list the hazards, estimate the risk, remove what you can by design, then add guards, then warnings. ISO 12100 describes this method. If you plan to sell the machine in the EU, it needs CE marking under the Machinery Directive, which the new Machinery Regulation (EU) 2023/1230 replaces from 20 January 2027.
Then test, and measure instead of guessing. Does it lift 20 kg by 0.5 m in 2 seconds? Does the load stay put when you pull the plug? How warm does the motor get after 200 cycles? Every "no" sends you back to the step where the assumption was made. For strength, a simulation before the first part breaks is cheaper than one after. Can AI run an FEA? shows what such a check found in a jumping robot.
Where AI helps, and where it does not
AI is fast at the parts of this sequence that are mostly writing and arithmetic. It turns a one-line idea into a list of questions, does the power and torque estimates with units, compares working principles, writes the CAD model, proposes parts with datasheets, drafts a schematic, and prepares the files a workshop needs for a quote.
It does not hold the prototype in its hands. Specs and part numbers must be checked against the source, as Can you trust AI with engineering numbers? explains. The trade-offs, the safety sign-off and the test on the bench stay with a person.
The short version
- Write the job down with numbers before drawing anything
- Check power, force and holding with ten minutes of arithmetic
- Compare two or three working principles against the job
- Start the CAD with the parts you buy
- Give the emergency stop its own path, not just software
- Choose processes and write the full bill of materials early
- Assess the risks while the design can still change, then measure every requirement
You can follow these steps in open-hardware projects on makemachine.ai, such as the Jump & Run Robot, the BBQ Modular Grill and The forever Washing machine. If you have a machine of your own in mind, you can describe it here.
Watch hardware get built by AI. Follow public projects on makemachine.ai, read every result in full and get an email when there is something new.
More from the blog
- Will my electronics enclosure overheat? A five-minute heat estimate for a sealed box
- Joining 3D printed parts: screws, heat-set inserts, snap fits or glue?
- Free or paid AI 3D modeling? What you really pay for, and a 30-minute test to compare tools
Open-hardware projects built with AI
- DIY Automatic Plant Watering System: An open-hardware watering system for indoor plants: a small pump, a refillable tank and a soil moisture sensor in each pot keep up to six…
- DIY BBQ Temperature Controller: A temperature controller for charcoal grills and smokers that makers can build themselves. A pit probe, two meat probes and a small fan on…
- ZählerSchau: ZählerSchau is a small, battery-powered unit that clips onto an electricity meter in Germany (water meters via a clamp adapter), takes a…
- Roadstoy: Roadstoy is a remote-controlled toy version of the new Tesla Roadster that can briefly jump and flip using compressed air. It drives on an…
- The forever Washing machine: The forever washing machine is a washing machine offered as a service instead of a product. Customers do not buy the machine but pay for…
- Bike Light: I want to build a bike light that needs no battery. Magnets sit in the spokes close to the hub, and a coil on the fork picks up their…
- Rooficle: I want to develop a light, efficient and affordable e-bike that can be used legally on public roads in Singapore. Its core is an…
- BBQ Modular Grill: I want to design a modular grill. Instead of one sealed block, it is made of separate parts that can each be bought on their own online…
- Go FANUC Go: A mobile base that moves an existing FANUC M-710iC industrial robot (about 560 kg, with a milling spindle at the wrist) out of a room and…
- Jump & Run Robot: I want to build a small, low-cost toy humanoid robot that people can put together themselves. It should be about 30 cm tall, or as small…