George Bryant

Electro-sax mouthpiece

The mouthpiece and breath sensor are the key parts of the electro-sax. They take it from being a weird shaped keyboard to an actual wind instrument. That makes it vital that they feel right. I want it to feel familiar to an alto sax player.

There are a few things that come together to make the mouthpiece feel good. One is the shape - it doesn’t (as I soon found out) have to be shaped exactly like a real sax, but somewhat similar is nice. It has to sense your breath, and allow air to continually flow through. Just sensing air pressure without allowing airflow feels very wrong. And finally, sensing the pressure of your bite isn’t essential but it’s nice for an extra level of expression.

Prototyping

The part I wanted to try out first was the breath sensor. It’s the core of the sax, and it’s the part I was most curious about. Based on a video by KontinuumLab I made a simple breath sensor using the top of a bottle, a straw and a balloon.

Numbered of a breath sensor - a tube (1) leads through a bottle cap into the chopped-off top of a bottle (2) with a balloon (3) stretched over the cut opening. There's a second hole in the cap (4). On the other side of the balloon there's an LED shining at it (5) and reflecting into a light dependent resistor (6).

Blowing on the straw (1) into the chamber (2) partially inflates the balloon (3), and the air escapes out of a smaller hole in the bottle lid (4). The more you blow, the more the balloon moves. That movement is sensed by a light-based distance sensor. An LED (5) shines on the balloon, which reflects light back at a light dependent resistor (6). A higher intensity reaches the LDR when the balloon is closer, which is sensed by a microcontroller.

It was instantly very convincing! Blowing into a straw isn’t very sax-like but honestly didn’t feel that odd. The sensitivity wasn’t great but it had enough range to get some variation as I blew harder or gentler. It responded a little slowly - playing rapid staccato notes wouldn’t be happening on this prototype. Nonetheless this prototype convinced me that I could make something compelling, playable and fun! Most of my later prototypes are based on the same principle, just with something a bit more custom-made than the top of a bottle.

5 prototype breath sensors in a row

Breath sensor prototypes

Prototype 2: Testing different sensors

Prototype 2 was a 3D printed mouthpiece on which I tested different sensors. This is where I identified the slow response of the LDR solution. Next I tried a barometric pressure sensor, which responded even slower! I also had concerns about exposing it to the moisture in my breath. Finally I tried a TCRT5000 IR LED + phototransistor combination. This was more sensitive, responded fast enough and was a single component (plus a resistor) so was easy to set up.

Three square pieces about 4cm across, each with a different sensor in the middle. The left one has an LED and an LDR, the middle has a barometric pressure sensor PCB and the right has a module with an LED and a phototransistor in an LED-like package

Sensor sections of breath sensor prototypes - LDR, barometric pressure and phototransistor

For all of these prototypes I used either a section of nitrile glove or a piece of balloon as a membrane. The latex balloons are much more flexible, but the nitrile material is flatter and easier to get into place. There’s probably a good source for thin, flat latex, but I didn’t find it. The membrane gets clamped with screws between a mouthpiece section and a sensor section.

Prototype 3: Tuning TCRT5000

The TCRT5000 test was a success, so I made an adjustable version to finetune it. This allowed me to move the sensor up and down to find the distance from the membrane which gave the best signal. Once the membrane gets too close to the sensor it begins to block light more than it reflects. If you hit that zone it’s very weird to play - you blow harder and it gets quieter!

Prototype 4: More sax-like shape

The next two prototypes kept the TCRT5000 sensor but played with the mouthpiece shape. Prototype 4 kept a similar breath sensor but had a detachable mouthpiece that was closer in shape to an alto sax mouthpiece. The detachable part caused problems - my FDM prints didn’t seal together well, leading to hissing as air escaped. The mouthpiece shape felt good except for the positioning of the air inlet, which was on the bottom side instead of the end. That made it easy to block accidentally with my lip. The opening of the mouthpiece needs to be on the end, without making it too bulky or too fragile.

Sketch page of ideas of how to implement a more sax-like mouthpiece, including separating it from the breath sensor and using a separate fake reed Breath sensor block with a mouthpiece attached to it as a separate piece. The mouthpiece is shaped vaguely like a mini version of an alto sax mouthpiece

This prototype felt like a really cool idea at the time, but just didn't live up to the hype.

Prototype 5: Single-block mouthpiece

Prototype 5 was an attempt to make a nicer looking mouthpiece section. It’s shaped to sit on top of the main box of the sax, and has an angular shape that matches what I was considering for the overall aesthetic at the time. The mouthpiece and breath sensor are once again joined together - I didn’t have a good way to connect them if they were separate. It’s also designed in a way that made it easy to 3D print without supports. Despite not being shaped much like a real sax, it felt satisfying to play and had good sensitivity.

Sketch page around ideas for how to lay out a nicer single-piece mouthpiece in a way that the wires go into the body of the sax but the wet air doesn't Cross section from CAD of the unibody breath sensor, showing the air coming in through a tapered mouthpiece on the front, expanding the rubber membrane and exiting through a narrow passage on the top The 3D printed prototype in orange and pink - the balloon membrane is now held in place with a plastic ring instead of taped on

Taking the single-block mouthpiece from ideas and sketches, into CAD and into a final form. This was my favourite mouthpiece for a while and I played it a lot!

Prototype 6: Pressure sensor

I haven’t yet mentioned the big problem with the TCRT5000. Because it’s an IR sensor, it’s sensitive to stray infrared light from outside. After one too many design modifications trying to stop the sax playing while pointing at a light bulb, I gave in and bought a pressure sensor chip. These have a small port to connect to an air tube and output an analogue voltage. Using silicone tubing allows me to separate the mouthpiece from the breath sensor. The breath sensor can be inside the body of the sax, making the external part smaller (just the mouthpiece).

Close-up of Y-splitter with a large silicone tube connected to 2 of the ends. The third smaller end goes to a pressure sensor which is shaped like a 6-pin DIP package with a plastic box on top of it

The pressure sensor attached to a Y-splitter off the input and output tubes

The pressure sensor is a huge improvement in consistency of the breath sensor. It’s just as sensitive, but much easier to construct as it only needs a connection to a breath tube from the mouthpiece. The downside is that it’s quite a bit more expensive - at the time of buying it an appropriate sensor was £15! It’s been worth it for the lack of hassle, although the longevity of the sensor when exposed to damp air remains to be seen.

Adding a bite sensor

The reed of a real sax isn’t static - it vibrates and that’s what creates the sound. A skilled player will adjust the pressure and position of their lip against the reed to shape the sound. A famous example is 0:35 in Baker Street - the saxophonist Raphael Ravenscroft repeatedly bends the long note. I want to be able to sense this bite pressure on the electro-sax to capture some of that expression.

That leads us to the latest prototype. I tried to make a more sax-shaped mouthpiece with a reed and bite sensor. The sensor is a force-sensitive resistor, which the reed presses on. The resistor is half of a voltage divider, so the voltage in the middle varies with the force on the reed. That voltage should vary the sound being output, adding an extra degree of expressiveness.

Bite sensor mouthpiece with the reed removed, showing a small force-sensitive resistor mounted to it. It's near the base of the reed so there's a little bit of mechanical advantage

The bite sensing mouthpiece disassembled, showing the force sensitive resistor that the reed presses on

The downside of adding this reed is that it leaks air. In order for it to move it needs some clearance at the sides, but this clearance allows air to escape, making a hissing noise. Along with the air comes condensation from the user’s breath, eventually causing a drip.

So there’s still some work to do - the mouthpiece doesn’t feel quite right yet. But it’s pretty close, and definitely good enough to test out how it should be shaping the sound. I suspect the next version will be shaped less like a real saxophone mouthpiece - perhaps more like the single-block prototype. For now, it’s time to move on to other parts of the sax.