Gun'na Fold 'em All

A mechanical paper airplane dispenser.

Five students from Olin College of Engineering set out to create a paper airplane dispenser for their Principals of Engineering class.

Our paper airplane dispenser takes a sheet of paper and performs a series of folds to crease the paper into a paper airplane.

Below is a system schematic which shows how our mechanical and electrical systems couple together.

Back Folder Module

The back folder module folds the first third of the paper airplane back onto itself.

Here are some key frames of how the back folder works:

The paper (red) enters the module on the feed ramp.

The paper continues forward until it hits the hard stop.

The paper buckles downwards and hits the two crimping rollers.

The crimping rollers help crease the paper in half and feeds the paper into the next module.

Click the gallery photos for an in-depth of our back folder iterations

We went through an iterative design process while figuring out how to fold paper back on itself, shown below are several iterations of the backfolder that we went through.

Corner Folder Module

The corner folder module folds the corners of the paper airplane onto themselves.

Here are some key frames of how the corner folder works:

The paper enters the module.

The edges of the paper ride up a tiny ramp and get caught above a string. The rest of the paper is kept down by a roof shaped triangle block.

As the paper is fed further the corners of the paper start folding backwards.

Finally two rollers grab the paper and crease the folded corners.

Click the gallery photos for an in-depth of our corner folder iterations

We went through the same iterative process with our corner folder module. Shown below are a few iterations we went through.

Spine Folder Module

The spine folder module creases the spine for our paper airplane.

Here are some key frames of how the spine folder works:

The paper enters the module, the center of the paper is pushed into a 1 inch gap by a large wheel.

The center of the paper is then coaxed into a smaller gap by two crimping rollers on the sides which creases the valley of the spine.

The final roller grabs the paper and makes the upper spine creases (mountain creases) more pronounced.

Click the gallery photos for an in-depth of our spine folder iterations

Below are the iterations we went through for the spine folder module.

Integrating Modules

Initially, we had significant issues integrating the back folder, corner folder, and spine folder.

Issues with initial integration

There were several problems with our initial integrated assembly. Several holes were too small, and others were too large. The rollers at the end of the triangle fold module were too close together and caused paper jams.

The large wheels the spine folder were too close together. We changed the position of the larger wheel in our final iteration. We used delrin rods instead of steel shafts. Delrin rods are flexible and as a result the large wheel was unable to provide the force needed to grip and pull the paper through the system.

We initially planned to launch the paper airplane after folding it. However, we had never tested the launcher before our final sprint, and it turned out that it did not work. We decided not fundamental part of our project and removed it in the next iteration.

Click the gallery photos to learn more about the failures of our first intergrated system

Click on the pictures to learn more about the failures in our first integrated system.

Final Integration

We fixed all of these issues, and remade the final integrated module. There were a few problems with the final iteration that we were able to fix. We shorted the distance between the triangle folder paired rollers and the largest wheel, which caused paper jams. The rod of the large wheel is geared to spin at the same speed as the rods for the paired rollers. However, the diameter of the large wheel is significantly larger than the rollers, so it spins at a significantly faster speed. We removed the rubber band from the wheel and tested whether the paper would slip and not be pulled as quickly by the wheel. The paper slipped too much--the wheel no longer had traction, and the paper did not continue through the module. We then cut up a rubber band and glued parts of it to the surface of the wheel, so that the paper would occasionally slip and occasionally be grabbed by the roller. This provided too much traction, so we removed all pieces of rubber band except one half-inch piece. The wheel then functioned as it should.

The other major issue we had was the paper curving upwards at the back folder instead of curving downwards and being caught and pulled through by paired rollers. We fixed this by attaching a metal sheet to the top of the acrylic, essentially extending the acrylic so that the paper could not curve upwards.

Gearing

In order to ensure that all the rollers spin the same speed, we geared the rollers together. We decided to separate the three modules into three separate gear trains. We felt that it would be impractical to build a single gear train to mechanically connect all the rollers because the rollers a far apart from each other and they spin in two different planes. In addition to being easier to design, fewer gears mean less friction and more mechanically efficient.

We were initially concerned about the three modules spinning at slightly different speeds because it might cause paper jams. Even though we used the same motors and gear ratio for all three systems, motors are not perfect so they will spin slightly differently. After some tests, we found that the slight speed differences did not effect the paper movement at all.

We decided to create our own gears because it would be easier to design gears based off the positions of the rollers rather than positioning the rollers based off of set gear sizes. We tested a lot of materials for the gears. In the end, we found laser cutting Delrin plastic sheets worked the best because it is durable, easy to manufacture and has a very low coefficient of friction with itself.

In order to securely attach the gears to the rollers, we press fitted Delrin collars to the gears. We then a stuck a metal pin through the collar and roller. Not only does this prevent the gears from slipping, it also helps constrain them in all directions.

Electrical

We wired three 12V, 46rpm geared motors up to a 12V wall power supply. Together these three motors powered the paper airplane dispenser, one per gear train.

With no load, the motors draw .8-.9 Amperes each, and with a gear train load can draw up to 2 Amperes each.

Software

Going into this project we knew the launcher would be very mechanical heavy but we still originally planned to have a software component. Our final goal was to mount the airplane launcher on a turret, so the user could remotely aim it. Unfortunately, we were never able to accomplish this goal.

At the beginning of the fourth sprint, we realized that our launcher mechanism was not going to work unless we invested a heavy amount of time into it. We decided that we would much rather allocate our time to the rest of the machine and to make it as robust as possible. Since we changed the scope of the project from a launcher to a dispenser, the turret idea got scrapped. It doesn’t make any sense to control which direction the machine dispenses the paper airplane.

Even though we were excited to have a software component it did not end up happening. If someone wants to pick the project up, they should definitely work on software from the very beginning instead of at the end. This would significantly increase the odds of it actually getting done.

Bill of Materials

By the end of the project we spent a total of $236.42 of our $250 budget.

Click the images below to learn how we used each material and their estimated cost!

Future Work

There are many electrical, firmware, and software ideas for the dispenser that did not come to fruition. The following section summarizes some of our backlogged ideas regarding these aspects of the project, and how we would have implemented them.

Motor control

An arduino or similar micro-controller could have provided us with smarter motor control. A simple program combined with a few switches could have allowed forward-run, reverse-run, and power-off control over each individual motor, which would have provided better control over paper jams and testing of various modules.

Rotary axes

Placing the originally planned launcher (now a dispenser only) on a rotary table, along with an arm for angle manipulation and another micro-controller would have allowed us to point and aim the launcher.

LED runway

A simple circuit consisting of LEDs and resistors would have made a spectacular runway in the dark. Oh well!

Target recognition

Utilizing the rotary axes mentioned above, plus some sensors and some OpenCV, we could have had target recognition and fired airplanes and passersby at Expo. We're really sad about this no-go.

The Engineering Team

Mugshots

Jiaying Wei

Gregory Coleman

Halley Pollock-Muskin

Ruby Spring

Ryan Louie