Blog · · 5 min read

How high can a toy car jump on bicycle-pump air? Jets vs. pistons in numbers

In the model, a toy car with a 3D-printed air tank hops 3 cm on air jets but 30 to 60 cm on kick pistons. Why jets are limited by the mass of air, why pistons push on the ground, and how to test a printed tank safely.

The brief sounds like a toy designer's dream: a remote-controlled car that can jump and maybe flip, powered only by air from an ordinary bicycle pump. No pump on board. The tank, the air channels and the nozzles should be printed together with the body, in one or at most two parts.

That is Roadstoy, an open-hardware project on makemachine.ai. Nothing has been built yet, so every number below is an estimate from the project's models. But the way the design changed between two concepts shows something that matters far beyond toys: how you use compressed air decides more than how much of it you have.

What a bicycle pump can store

Many floor pumps are rated to about 11 bar (160 psi). The first concept planned for 8 bar.

One litre of air at 8 bar weighs about 11 grams, and a little under ten grams of it can leave the tank. That is not much mass. It is a lot of energy, though: several hundred joules. For scale, lifting 600 grams by 75 centimetres takes about 4.4 joules.

So energy is not the problem. Getting it into the car is.

First concept: two bottles and five jets

Version 1.0 put two half-litre plastic bottles into a 1:10 car, filled to 8 bar. Four air jets point down at the corners, one boost jet points back, and a small gyro controller keeps the car level.

The flight model gave one litre at 8 bar about 3 newton-seconds of push. That is enough for one trick per fill: a hop of about 0.75 m with a backflip that lands upright after about 0.9 seconds in the air, or a boost of 15 km/h in half a second.

Hovering is out of reach: about 0.6 seconds per litre, so three seconds would need five to six litres of tank.

Second concept: the tank moves into the body, and the hop collapses

The brief asked for very few parts, so the next version did exactly that. Version 1.1 is a 1:14 car, 307 mm long, made of two printed parts. The main body holds the tank cells, channels, valves, nozzles and four small cylinders; a snap-on nose holds the battery and the controller.

Then the model gave the bad news: with jets alone, the car hops about 3 cm at 6 bar. A tank inside a 30-centimetre toy body is simply too small for jets.

Why jets need air mass

A jet pushes the car by throwing air the other way, and its total push is roughly the mass of air times the speed at which it leaves. In the first concept, about 3 newton-seconds from a little under ten grams of air means the air leaves at a bit over 300 metres per second, close to the speed of sound.

Better nozzles gain a little, but not a factor of ten, so the main lever is more air. Halve the tank and you halve the push. Hop height grows with the square of the take-off speed, so half the push gives about a quarter of the hop. Even less in practice: while the thrust is below the car's weight, nothing lifts at all.

Jets also waste most of the energy. To push a car that moves at a few metres per second, they accelerate air to hundreds of metres per second, and that air carries the energy away. At hop speeds only a percent or two ends up in the car.

Why pistons work: push on the ground

The recommended option in version 1.1 keeps the printed tank and valves, but the air drives four kick pistons built into the body that push against the ground. The model gives a hop of about 30 to 60 cm, at only 3 bar.

A piston's force is pressure times area, and its work is force times stroke. As an example, four pistons with a 16 mm bore at 3 bar push with about 240 newtons together, some forty times the weight of a 600-gram car. Over a 15 mm stroke that is about 3.6 joules, enough in theory to lift 600 grams by 60 centimetres.

A jet has to accelerate air. A piston pushes against the ground, which does not move, so the pressure turns into lift almost directly. Same tank, same valves, ten to twenty times the hop, at half the pressure.

The part that can hurt: printed pressure tanks

Compressed air stores energy, and a tank that cracks releases it at once, with sharp plastic flying. Printed parts are weakest between their layers, and two prints of the same file are not equally strong.

The research also checked bottles: the lowest published burst pressure it found for two-litre PET drink bottles was 7.5 bar, below the planned 8 bar, and half-litre bottles had no burst data at all. The plan dropped to 7 bar until the project's own tests are done, still close enough to 7.5 that the burst test comes first.

The project's rule: no untested pressure vessel. Every printed tank is burst-tested before use, like this:

  • Test with water, not air. Water barely compresses, so a failing tank leaks or splits instead of exploding. Pressure vessel codes for metal tanks test with water at 1.3 times the maximum working pressure or more, and allow air only where water cannot be used.
  • Leave more margin for plastic, not less. Test several prints, behind a shield, well above the working pressure.
  • Fit a relief valve set at the working pressure, so the pump cannot overfill the tank.
  • Use the lowest pressure that works. Stored energy grows with pressure, and the piston design needs 3 bar instead of 6 to 8.

Valves decide the details

Version 1.1 uses sliding steel pins as valves. They open and close single channels in about 25 milliseconds, but without O-rings they would let the tank leak empty within seconds. Timing and sealing are measured on the bench, not looked up.

What is still open

Nothing is printed, pumped or measured yet. Next come the mechanical CAD, burst tests of the printed tank and a first hop, plus a stress simulation of the printed body like the one in a jumping robot's FEA. The printed parts also have to follow the rules for strong prints.

The short version

A jet's push is limited by the mass of air you carry, so jets need volume. A piston's push is pressure times area, so pistons need area and something to push against. For a small printed tank, kick pistons beat jets by ten times or more in the model, at half the pressure. Whatever you build, test the tank with water first.

Follow Roadstoy to see whether the models hold up on the bench. To check your own idea the same way, start at makemachine.ai/try.

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.

Get started See the live projects

More from the blog

Open-hardware projects built with AI