Blog - Sketching with Hardware

Different ways to build a catapult

Published on: | Author: Anja Mainz | Categories: 2018a, Tutorials

As part of the practical course “Sketching with Hardware” we built a catapult among other things to complete our project. On the way to a functioning catapult we got to know different possibilities to build a catapult. I would like to present these different possibilities as well as the structure of our functioning catapult in the following.
Generally speaking, there are four basic ways to accelerate the ball in the catapult (or whatever you want to load it with – in our case, however, it was a ping-pong ball): With the help of a spring, air pressure, a magnet or the impulse of motors.

“Spring-driven catapults.”
A fundamental problem, if you want to build a spring driven catapult, is the selection of the spring. It makes sense to choose a coil spring or alternatively a suitably strong rubber. If you don’t want to accelerate by pulling, you need a compression spring. What is interesting here is the choice of the “right” spring, which depends on what should be hurled and how far. From a purely theoretical point of view, each spring has a so-called spring force, which must be provided to tension the spring and which is inserted into the acceleration of the object when releasing the spring. In the case of a spring, this force should be given by the seller (typically in Newton) – this may be more difficult with a household rubber. In addition, it is not easy to find out which spring force you need for your project. Newton’s law says:

a being the desired acceleration and m being the mass of the object. Here the friction is completely neglected (which might be a factor for a very light ball with the air!). Obviously, however, this approach does not really help in practice.
Acceleration during shot put is about 10 m/s². The ball is considerably heavier than a ping-pong ball and is thrown 15-20m wide. Nevertheless, this number can serve as an order of magnitude. My advice would be to choose this as the upper limit and calculate the power required for your ball. Make a selection of springs whose maximum spring force is at most equal to the calculated force and experiment with it to find the correct spring. If it is known which spring or rubber is to be used, the following three variants are available:

  • The “simple” variant: At first, this variant appears to be the easiest to implement. Picture material for the description below. a suitable tension spring, servo or solenoid, two pipes and a small “paddle” are needed to hit the balls. At best, the surface of the paddle should have the same diameter as the ball. The length of the handle depends on the diameter of the selected pipes. One of the two pipes should have a slightly larger inner diameter than the diameter of the ball (the “launching pipe”). The spring should be built into the other one, so this one should have at least the length of the tensioned spring. Cut out a gap from each of the two pipes over the entire length. The “spring pipe” also requires a hole opposite the gap through which the servo can later hold the spring in place. Fasten the pipes together so that the gaps lie next to each other. Attach a small, circular wooden plate to one side of the spring, whose diameter is slightly smaller than the inner diameter of the “spring pipe” (it is helpful to attach something like a string to the plate so that the spring can be tensioned later). The “paddle” must now also be attached to this wooden plate. The next step is to place the spring-paddle-construction into the pipes. At last attach an extension to the servo, which can hold the spring plate, fix the servo at the top of the “spring pipe” and control the servo via the Arduino. The catapult is finished. The spring can now be stretched by pulling on the string – the servo can either be pushed gently to the side or moved sideways via the Arduino. If the spring is tensioned, the servo must return to its place and now holds the spring tensioned. Now you can throw the ball into the front of the pipe and as soon as the servo receives the command from Arduino, it pulls the pin holding the spring, the spring contracts and accelerates the ball. This construction has the disadvantage that the board which is supposed to accelerate the ball can get stuck in the pipe. This can be solved with rails, which in turn makes the construction more complex.
    A possibly easier variant would be to work with a compression spring and only one pipe. For this purpose, a board is attached to both ends of the compression spring (one wider than the inner diameter of the tube, one with the diameter of the ball. Then the smaller board and the spring are inserted into the tube and the larger board is attached to the end of the tube. Now you only have to make a cut in the tube where the smaller board comes to rest when the spring is tightened. Attach a servo next to it that can hold a pin in the gap and pull it out. In the stretched state, the servo holds the board with the pin, when it releases the pin, the spring relaxes, accelerates the smaller board and thus the ball lying in the tube. However, even this version does not solve the problem of blocking.

    Conclusion: At first glance, this is a very simple solution, but there is a risk of being stuck and fixing by means of rails is rather complicated. This version must also be tensioned by hand. It would then be almost easier to build something that works like a crossbow.
  • The prototype of Group 2: Group 2 built a bird feed station called “Birdinator” during the “Sketching with Hardware” course, which automatically killed flies for the birds. In the first version this flytrap was nothing more than a catapult. This catapult was a bit more advanced than the one described above – it could also tension itself: Here, the object is not shot out of a pipe, but is hurled into the air as with an antique catapult. The throwing arm was a fly swatter, which was attached flexibly to a board with one end and the other end was attached to the board with rubber bands. In the opposite direction was a motor connected to the end of the fly swatter by a cord. When the motor was switched on, it started to turn and wind up the string. That pulled the end of the fly swatter towards the engine, the rubbers stretched. When the cord is fully rolled up, a servo secures the fly swatter in its position and the motor unwinds the cord again. If the line is completely unrolled, the servo can release the “throwing arm”, which can then shoot its “charge” as does a classic catapult.Conclusion: A simple solution that can also be realized in the short time of a “sketching with hardware” practical course.
  • The variant for advanced users: I came across this variant when researching for our catapult. The original video can be found at This solution is very elegant (although obviously quite elaborate): on a large wooden plate, a circular plate is fixed in a bearing on a metal pin so that the plate can rotate. Even more important, however, is the helix-shaped plate (see below), which is attached to it. The large, round plate is powered by four small wheels (you can move them with motors, for example – if you work with motors in your project, you will need two H-bridges (you can connect two motors to each of them). It is important that all motors run in the same direction. In the video only a separate motor is used, but this motor needs more power). In the next step, a board is now installed across two of the wheels, on top of which a second one is fixed horizontally but orthogonally to the first board, which is mounted in a bearing on the first board and thus remains movable. The tip of the second board should be above the centre of the circular board. A further wheel is attached to it, which, if you turn the round plate, starts to circle the helix-shaped board. Now the two boards only have to be connected with a suitable tension spring (see below). The catapult is almost finished with this, it already has a fully functional “hitting arm”. The ball only has to be placed in such a way that the striking arm hits it (in the video it is simply placed on two parallel boards). I personally consider this solution to be very elegant, as the helical shape of the board makes it easy to retighten.Conclusion: A very elegant solution if you have enough time to build it.

The “air-driven” catapult

One solution I have stumbled upon repeatedly during research is catapults that work with air pressure. One possibility I have found is to use two normal drainage pipes, one short and one long and mounted between both pipes a T-shaped pipe (in the hardware store this part is called “branching”). A leaf blower is inserted into the short end of the resulting “pipe with a throw-in hole” (the finished product can be seen below) and a catapult which can shoot tennis balls is the result. You can find the catapult at This type of catapult is temptingly easy to build, but has one big problem: the air pressure. It is not without reason that a leaf blower is used in the video to shoot tennis balls. In order to move something, a lot of air pressure is needed and a lot of power. The leaf blowers that I found during the research require about 230 volts. This means that they are clearly out of the range in which you should find yourself as a newcomer in the “Sketching with Hardware” course. If you want to move something less heavy than tennis balls, you have the option of using a hairdryer. Nevertheless, this should not be overestimated. I only experimented with a 220V hairdryer: with this one you can accelerate a table tennis ball to move it shorter distances. In order to avoid the problem of high voltages, however, you have to fall back on travel hairdryers – there are actually hairdryers that only need 12 volts. Though, they have only a fraction of the power (my 220V hair dryer has 1000 watts, a 12 volt travel hair dryer has 120 to 168 watts maximum). This problem can of course also be avoided, but then you have to work with compressed air accumulators and reciprocating pistons and the very simple approach becomes quite complicated.

Conclusion: If you don’t want to control the catapult with an Arduino and have a leaf blower at hand, you can build a very effective catapult. Air pressure is regardless probably not the perfect solution for the “sketching with hardware” course.

Magnet-driven catapults: For magnets similar to springs, you have to think about how strong a magnet must be in order to move the desired object over the desired distance. You can think about how much Newton you will need and then experiment with different magnets in this spectrum (for guidance: in the variant I describe in the next paragraph magnets were used, which hold 24 kg (this corresponds to 235N) – with that a juggling ball can be shot 3 meter according to the mentioned source.

  • The magnetic catapult: The simplest version of a magnetic catapult can be found here: In this version a traditional catapult with a throwing arm was built. A permanent magnet is built into both the arm and the point of the catapult where the arm lies in the tensioned state. Note that the magnets have to be mounted with opposite polarity. The catapult is mechanically tensioned and fired in the mentioned source. With a strong motor, however, it is certainly also possible to clamp and fire the catapult as described above in the prototype of Group 2.Conclusion: Fully automatic clamping and firing may require some tests – especially with strong magnets a lot of force is required – but the magnet construction itself is simple and reliable.
  • The matter of the electromagnet: After we had thought about the various possibilities mentioned above during the research, we came up with the first idea that we wanted to realize. Inspired by the magnetic catapult and the so-called Gaussian rifle, we wanted to build our catapult with the help of an electromagnet. The basic idea is the following: An electromagnet has the advantages of a permanent magnet with the additional advantage that it can be switched on and off. Thus, the catapult would not have to be clamped against the force of the magnet like the normal magnetic catapult. Instead, only the magnet in the throwing arm would be a permanent magnet, while in the base there would be an electromagnet. By switching on the necessary power, the electromagnet also transforms into a magnet and the two magnets push each other away, as they do in the case of a magnetic catapult. A visit to the Conrad taught us besides the realization that the employees often have not much more knowledge than we students do, that an electromagnet with a core is completely unsuitable for the project to build a catapult, because the permanent magnet holds on to the core and the electromagnet has to overcome this magnetic force in addition to accelerating the ping-pong ball. If you want to build a catapult with an electromagnet, I would like to give you the following information:
    If you want to wind the coreless electromagnet yourself, it is important to use a wire that is as thin as possible (e. g. 0.05 mm²) and at the same time isolated. The wire must always be wound in the same direction!
    – The more windings, the more powerful the effect
    – Thin wires are very sensitive to high voltages. Electromagnets should not be permanently switched on, but only for a short period of time.
    – To help you decide whether or not you want to work with an electromagnet: Our supervisor advised against it, because you have to put a lot of energy into it for a comparatively small effect. That’s why we didn’t continue to work with this approach (besides the fact that we had worked with too thick wire at the beginning and were disappointed by the lack of effect). However, if you have more time than a few days: I think it is exciting to give it a try, because an acquaintance with 12V, 0.05mm² wire and not too many windings could shoot a small screwdriver (see photo) through the room (the magnet is also coreless – what looks like a core is the permanent magnet, the counterpart of the electromagnet. The effect increases when the permanent magnet is in the coil at the beginning).
    Conclusion: This variant sounds like a lot of fun to experiment with, but may require high power for little effect.


Catapults based on motor power:

  • Wheels: During the research for our catapult I came across many table tennis ball machines, which work on the principle that the table tennis ball rolls over a pipe to two small wheels. The two wheels are powered by motors and rotate against each other. The concept is based on the fact that the wheels have exactly the right distance from each other. They must be far enough apart so that the ball can still pass between them. At the same time, they must be close enough to “grab” the ball. Both wheels can be operated with one motor each. The direction in which the motor rotates depends on how you connect it to the H-bridge. You can also change the direction of rotation in the code with the code linedigitalWrite (in5, LOW);digitalWrite (in6, HIGH);or exactly the other way around. In5 and in6 must be the two exits of the Arduino, to which the H-bridge is connected. Such a design can be found at The “delivery” of the balls takes place here via a vertical pipe, which is cut horizontally so that a servo can hold a small piece of wood into the pipe. If a ball is to be given, the servo moves the piece of wood to the side, the ball falls through the pipe, rolls to the wheels and is accelerated into the room.Conclusion: A good and simple solution if you have two adequate wheels.
  • Rotors: With this, we have arrived at the construction method we have chosen in our project. The basic idea was given to us by our supervisor, that a motor equipped with rotor blades could also shoot our ball. In fact, this version is possible, but only works in 50% of cases: When the rotor blade hits the ball at an angle of 90 degrees, the ball is catapulted through the entire space. However, the rotor blade has many other angles where it can hit the ball and the steeper they are (the more the rotor blade is parallel to the desired direction of the ball shot), the less energy is transferred to the ball. We solved this problem by drawing inspiration from the wheel catapults: We installed a motor with two rotor blades on both sides of the launching pipe. This increased the number of “good” shots immensely, because one of the rotor blades hits the ball very well. If you want to rebuild this, make sure that the two rotors are slightly different in height, so that their tips cover the width of the ball together when turning – photo of our rotors can be found below. Similar to the design of the above mentioned catapult with the wheels, we added another pipe to our launching pipe, which was mounted at a rather steep angle on the launching pipe. We did not regulate the ball specification by servo, but drilled a hole in the upper pipe and stopped the balls via a solenoid. This design has two disadvantages compared to the servo design: You have to keep the upper tube quite thin (or artificially slim it at the point in question) because the pin of the solenoid is not very long and the balls simply roll over it. The second problem is the one I mentioned above about electromagnets: The magnet should not be permanently switched on, otherwise the wire will have to cope with high amounts of energy. In this construction, however, the electromagnet must be permanently switched on, as it is only switched off briefly to let a ball pass through. Conclusion: The rotor blades give the balls real momentum, but even with two motors not every ball is hit perfectly. For ball handover, the servo used above is better suited than the solenoid we have chosen.
linked categories 2018a, Tutorials


Write a comment