When this project started, our main goals where to create a parametric model of a mouthpiece which could be adjusted with the click of a mouse and an artificial mouth (AM). This artificial mouth could be used to eliminate the saxophonist while comparing mouthpieces. This is useful because a saxophonist immediately  adapts when he or she is given another mouthpiece. With an AM we could compare the sound spectrum of mouthpieces with different geometries with the exact same lip pressure, lip position and air pressure. Afterwards, when a saxophonist would describe how he or she thinks the mouthpiece sounds this would give us information about which geometrical aspects of the mouthpiece influence the sound in a preferred way and we could try to maximize this by looking at the sound waves. This in turn would be of use in for example printing personalized mouthpieces, which could be printed with the type of ‘sound’ the customer is looking for.

Even though in the end we were able to play the saxophone with a bicycle pump using our homemade mouthpieces, there were some minor and major problems which, in retrospect,  could have been prevented easily.

Planning is key. Although we had a concept for the AM relatively quick and knew how to build it mostly, nine out of ten times you  will come across some problems while making it though. For example, the lower lip mechanism was thought all the way through before starting to manufacture the definitive version. Also a test version was made to see how it would work. If all stages of designing the mechanism inside would have been applied to the rest of the box as well it would have come together a lot better in the end. So in a future research, when designing the AM, it would be a great start to make a complete CAD model of it before actually starting to build. This way everything would be dimensioned beforehand, so no calculations, even the simplest ones, could be done wrong in the workplace. In our model, this led to the wall thickness of the polycarbonate box walls not being taken into account. Which in turn led to having to eliminate the system to read out an X-value for the lower lip positioning system.

Don’t make assumptions. I.E. Our artificial mouth was designed using a 2 mm rubber gasket in between the walls and the steel. After cutting the steel, we found out only 3 mm rubber was sold at the inventory where we got most of our materials. Even though this meant just a 2 mm difference compared to our design,  this made it really hard to fit the polycarbonate walls into place and actually made the AM less airtight.

Another assumption we made was the availability of pressurized air, because there are air outlets through most of the building we tested in and at the start of the project there was a lot of talk about a pressurized vessel. This was to be able to use the AM in an acoustics room, in which there are no air inlets because it is a closed environment. However, we didn’t really take into account this air had to be build down from around 10-20 Bar to 30 to 60 mBar. This requires high precision air regulators which are hard to come by. These kinds of specialty equipment should be ordered long before starting to build the AM. A good guideline would be to try to have everything for the build there before the last week (in which the build takes place).

Although we made some small mistakes along the way we thing it was a good project which gave some great insights in how to build an AM which could do the research very well.

Science fair

On the 30th of October the final moment was there: the science fair. Before it started however, we needed to finish the artificial mouth and to put all things together. Unfortunately the dimensions of the box were 8 mm too short in the length and height, so we couldn’t put the electronic ruler into the box. We couldn’t chance it either, because there was no time for that and we just putted everything together and hoped it would work.

At 12:30 the science fair started. We made a small stance at the designated location and placed our equipment there. That was also the place where we first tested our artificial mouth in combination with the 3D printed mouth pieces and the saxophone. At first, it didn’t really work. We tried to chance the lip position and power of the pumping of the bicycle pump. After a half an hour small peeps or hisses could be heard from the mouthpieces and after an hour a single tone could be heard! After 3 weeks we have proven that our box worked!

Making the 3D model and 3D printing the mouthpieces

We used a section view of the mesh to model the mouthpiece. With the section view we could see the inside contours as well as the outside contours of the original mouthpiece. We used these contours to create our geometry with the use of surfaces in Solidworks.

The first step in creating the model was to form a solid. Then we made surfaces which followed the outside contours of the mesh. These surfaces were used to cut into the solid and thus creating the outside geometry.

After this we used the inside contours to create the internal cavity.

This was used to cut out the inside of the mouthpiece.
We modeled it in such a way that we could adjust the tip opening by simply modifying two different parameters. This resulted in achieving our goal of creating an adjustable 3D model for the mouthpiece.

We 3D printed three different mouthpieces each with a different tip opening. With these mouthpieces we could test certain functions of the artificial mouth. However the prints were not fully completed due  to a printer error. Luckily we still could use these prints because the part of the mouthpiece that was not completed was not that significant. This is mainly because the inside parameters stayed the same and no holes were created.  The only difference with the model is that the upper shape is flattened.

The 3D modelling of the mouthpieces

The artificial mouth is going to be used to investigate the effect of the characteristics of the mouthpiece (e.g. tip opening) on the sound produced by the saxophone. Therefore we will need a variety off mouthpieces with different chracteristics. The mouthpieces will be 3D printed. In order to achieve this we will need a 3D model in which we can adjust parameters.

First we scanned four different mouthpieces in a CT-scanner (computed tomography). These scans will give us the actual 3D shape of the mouthpieces. Two alto mouthpieces and two tenor mouthpieces are scanned. The CT-scanner is located on the faculty of Civiele Techniek.

The problem here was that the mouthpieces were to big to be scanned in one scan. Thus every mouthpiece needed to be scanned two times. In total we had 8 scans and every scan is worth for around 5 gigabytes of data. So we had 40 gigabytes of data which needed to be converted into dicom files. These dicom files are actually section slices of the mouthpiece:

These image slices will be computed in a program to make a mesh:

A mesh is a kind of 3D image. Since we had for every mouthpiece two scans we needed to combine the meshes in to one mesh. We used a program called Meshlab for this. This created a mesh were we could work with.

A mesh does not have real geometry, it is just a collection of points. So we now have to make a CAD model in Solidworks with real geometry. The aim here is that we make an adjustable model. So that by changing some parameters we have a 3D model with different characteristics.
Our basic 3D model will be based on the mesh. We will actually use the contours of the mesh and aproximate our geometry as closely as possible.

Week 4: Sticking it all together

Since the end of the project is in sight we have to run our legs off to finish it off properly.

We are just processing and manufacturing, the design phase is over. Jan and Ruben have done most of the machining for the lip mechanism. All parts are there, but their not completely done yet. For example some polishing has to done to make the block slide over the rails smoothly and to make everything fit. The mechanism is simple in it’s design, but to make it is something else: Tolerances of up to 5/1000 of a millimeter are taken into account to make it work smoothly, which is quite a task for a team with no more than 24 hours of experience in milling. We tested the acme thread today, and the nuts fit great in the machine. An adaption had to be made to make it machinable though: instead of actually milling out a hexagonal shape, a rectangular shape with the height and width of a nut was milled, which works perfectly. This part of the machine will probably be done by monday morning.

The box still needs a lot of work. The main reason for this is that we didn’t have the right information about the digital ruler. The dimensions of this implement became so big that we actually thought of leaving it out, to maintain a reasonable volume. However, a precise measurement in the x-dimension is needed when repeating the experiment so we decided to keep it in. To reduce volume in the box, a filling material can be put in to keep it to a minimum.

Right now all materials have been cut to size, but not yet welded. This will take some time and we hope to get it done by Monday afternoon.

The pressure sensor also came in and it works like a charm. When plugged into a  multimeter, it will give a certain value when pressed. This value can afterwards be looked up in a diagram which translates it to either grams or force (newton).

Sensor which will be put onto the lip mechanism

Week 3

This week we worked out the details of the artificial mouth and we scanned the mouthpieces and putted the data into the various programs.

We have made the box smaller and tweaked the lip raiser slightly. Furthermore we tested if the polycarbonate plates could resist drilling near the edges.  It turns out the plates didn’t sliver, so we will use screws to keep the plates and the frame together instead of using silicone kit.

The scanning of the mouthpieces was successful and we are now putting the data together to make a replica in SolidWorks. We also designed a ring that would fit on the mouthpieces and that will fit in the hole of the lid of the artificial mouth.

Next week we are planning to put all the pieces together and to really build the artificial mouth.

Prototype for displacement in Y-axis

This mechanism shows how replacing a block in the X-direction will be translated to a displacement in the Y-direction. This allows us to move the lower lip up and down without having to make a slot in the box, but simply a hole which can be made (almost) airtight. To see it in motion, watch the youtube clip below.

In the final artificial mouth, the block will be moved with acme screw thread, this allows us to position it very precisely. There will also be incorporated some polyamide (nylon) sleeve bearings to reduce play in the mechanism and thus create a more precise measuring device.

Week 2

This week we tried to determine the materials that were needed to make the artificial mouth. We also tried to work with the programs that were needed to open the CT scans.

We searched roughly throughout the internet and came up with the following list:
-Steel profile 20 mm angle profile.
-Sealing ring
-Trapezoidal thread 10 mm smallest possible pitch.
-Aluminum for milling
-Plexiglas 6mm
-Aluminum round, for the rails and guides.
-Silicone for lip
-Clips for fixation
-QTc pills (pressure sensor)
-Digital ruler
-Electrical resistance meters.
-Steel Plates for QTc above the pill
-Nuts and bolts to attach plexiglass
-Sealing rings for plexiglass on steel
-closing bracket for the nozzles

The next step was to use the raw drafts from last week to work out the details in Solidworks. When this was finished, we used the technical sketch function and with that we made the prototype parts of the artificial mouth.  A lump of Aluminum was milled to create the pieces. This was done to see if the mechanics of the ‘’lippresser‘’ really worked


A problem occurred when we tried to use the programs that should view the CT scans. The software did not want to show any piece of the scans at all. This problem is however  solved and we are still trying to get used to the software.

Next week we are planning to scan the mouthpieces and to make more parts of the artificial mouth.

Week 1

This week we first gathered around to brainstorm on the problems concerning the design of an artificial mouthpiece and to create several solutions for them. We started with a list of criteria for the artificial mouth.
Here are a few main criteria:

  1. The artificial mouth provides the ability to hold several sizes/different kinds of mouthpieces.
  2. The product is made in a way which makes it possible to do a particle flow test.
  3. It contains a valve to insert air.
  4. The product must be able to withstand a maximum pressure of 150 mBar.

Afterwards, we used these criteria to design an artificial mouth. The general idea was to make a small box in which we would stick the sax mouth piece. This box contains two ”lips” that fit to the mouthpiece from the inside. Furthermore, it is connected to an air tube which causes the box to fill up with air and flow through the sax mouth piece.

We came up with the idea to make the lips from silicon to represent a humanlike feel. The bottom lip can be moved back and forth and up and down by a system based on two axes. The system can be seen in the image below.

These axes will be turned from the outside of the box. Below the bottom lip, a pressure sensitive foil will be placed to make sure an equal amount of pressure will be put on every mouthpiece. The movements in the X- and Y-direction would make it possible to fit the lip to the different mouthpieces that are diverse in length and width. The box will most likely consist of an aluminum framework that has transparent Perspex sides. This is done to see through the box while doing experiments. This way particle flow within the mouthpieces can be measured in future experiments.

The amount in which the lips can be adjusted is dependent on the size differences in several mouthpieces. For this reason the dimensions of a tenor and soprano mouthpiece have been measured to see the maximum differences. Perhaps the dimensions of a baritone mouthpiece will be measured or a safety factor will be implemented to guarantee every regular mouthpiece will fit our artificial mouth. The found dimensions can be seen in the image below.

Unfortunately we can’t start the CT scanning of certain mouthpieces next week, therefore our next step is to obtain the materials and to determine the dimensions of the parts.

The first day

We started trying to discover how a saxophone works. Ruben, our sax player in the group, explained this with his sax. He showed the importance of the mouthpiece. A big part of the sound is dependent on the mouthpiece, the closer the parts are near the reed the more influence they have on the sound.

A big part of our job is to create an artificial mouth so we can make a controlled comparison of mouthpiece designs. We redesign the artificial mouth made by Valerio Lorenzoni. He gave us helpful advice to improve his core design. A big difference between his artificial mouth and our future design is the purpose of the research. He researched the wind flow in the mouthpiece, our goal is to compare sounds of different mouthpieces.