Robotic Glass Harp

Elias Benevedes, Kyle Mitchell, Robert Starr

HU 3910 Musical Robotics B20

Have you ever tried rubbing your finger around the rim of a wine glass? If you apply just the right pressure with the right amount of water on your finger, you can cause the glass to start resonating with a pure tone! This concept has been incorporated into instruments known as Glass Harps. These instruments are usually played by one or two musicians, as the number of notes that can be played at once is limited by the proximity of glasses and available fingers to play them with. We look to solve this limitation with Glass Tango: a robotic glass harp.

One fully assembled module, with tuned wine glasses and servos ready to actuate.

A CAD model of three five-glass modules assembled together. In the middle is a mounting plate for brushed DC motors and electronics hardware.

Instead of spinning fingers around the rims of the cups, we decided to have the glasses spin inside of 3D printed cups. This reduces mechanical complexity of having a separate motor mount for each glass, and reduces the number of motors required overall. Five glasses are grouped together in one module and are driven together with a single 53” long, 6mm tall belt.

Originally, the plan was to have these modules driven by brushless DC motors. These motors are much quieter and have less internal vibration, making them the perfect kind of motor to drive a system that we want as quiet as possible. However, driving these motors requires complex and expensive electronics, especially at the speeds we desired. After hours trying to design, develop, and debug a custom solution, we were not able to drive the motors at a high enough torque to overcome mechanical resistances spinning the 3D printed cups. In the end, we utilized brushed DC motors to drive the belts.

The trickiest part of the project was getting the glasses to sing with a material that can be mounted to a servo motor horn. Also, keeping the material and the rim of the glass lubricated is critical to maintain resonance. The first material that was tested with some success was latex (from latex gloves). When wrapped around a solid object, they acted similar to a finger and would cause the glass to sing. However, keeping that finger lubricated was a complex design problem. We considered multiple solutions, but decided to try using a sponge to lubricate the rim of the glass. To our surprise, the sponge not only lubricated the glass, but would resonate the glass on its own! This became our solution: mount pieces of sponge to 3D printed servo horns that would attach to the servos.

The exciter assembly. A 3D printed servo horn is mounted to a servo, which is mounted to a bracket, and finally mounted to the tower attached to the base plate.

The servo was mounted to a bracket, which was then screwed in along a slot. This allowed the vertical position of the servo to be adjustable in case there were differences between the heights of each glass. This unit was mounted at the top of a tower made of laser cut wood to position it above the rim of the glass.

Unfortunately, the team ran out of time to complete this project, and thus were not able to get the machine to consistently create resonance on its own. The servos we selected (9g hobby servos) could not apply enough pressure to the glass from the sponge piece to cause it to sing consistently, even when powered with an independent power source. Furthermore, small pieces of sponge did not stay lubricated well and would deform in shape when mounted to servo horns for extended periods of time.

In the future, there are a few solutions to these problems to be considered. First, to solve the excitation problem, either a more powerful servo would be required, or have the rim of the glass closer to the finger mounting point. This would allow more pressure to be applied to the rim of the glass. Second, using larger pieces of sponge on the servo horns would allow the sponge to stay lubricated longer, and would allow the team to slightly change how the sponge was mounted every so often so it did not “wear in”. Finally, reimplementing BLDC motors as the main driver for each module is almost necessary. As heard in the video below, the brushed DC motors are quite loud, detracting from the musical quality of the machine. BLDC motors would allow for much quieter operation of the machine.

We got VERY close to getting this machine working. With a few small tweaks and changes to the underlying hardware, we are confident that this machine would work great, producing beautiful music with complex chords and harmonies.

**CLICK ME to view video of one glass being played by a servo**