Noise Avenger - Tame your Unruly Neighbor
Noise Avenger displayed in Georgia Tech Fall 2018 Mechatronics Expo |
It was back in 2017, I moved in to the designated apartment in Alabama with 3 other interns for summer internship at Hyundai. Everything was in peace until we noticed the constant loud stomping noise coming from the ceiling. This disturbing noise would not seize and soon it began to traumatize all my roommates, especially me. We tried hitting the ceiling with a broom stick, threw balls at it, and even stuck a racing drone on it, giving it a full throttle for minutes. None of them worked. We were all shy people so instead of walking up the stairs and have personal encounters with strangers, we planned to make a device that retaliates neighbors by knocking back on the ceiling.
But then again, we were also lazy responsible with loads of internship work so our plan came in halt as we just decided to endure the noise throughout summer.
The idea of the device was within the subconsciousness until Fall 2018 semester in Georgia Tech mechatronics class (ME 4405), my teammate (Jaemin Yoon) and I were given a task to come up with a novel device that uses multiple sensors, and do cool stuffs. Hence the creation of Noise Avenger.
Project proposal can be viewed here
Of course, CAD model is designed before the fabrication. This was done with SolidWorks. The wooden platform on which the device is attached is designed to simulate the wall.
The motor used for raising up the pendulum is rated for 1000 rpm at 12V, which is too fast for our application. Therefore, 16:1 gear reductions were 3d printed to bring down the speed while increasing the output torque.
Basic structures are water-jetted from 6061 aluminum plates. The tolerance of the water jet was awful (off by -0.011 inch when cutting inner diameter) so majority of the pieces had to be fixed by either dremeling or drilling, especially where the bearings would stay. This process produced a lot of metal dusts which easily went into the open bearing, often preventing them from operating smoothly (should have used sealed bearings). Transparent acrylic plates holding the aluminum structures together were nice for displaying internal parts to the audiences at the expo. Also, maintenance was easy since I was able to catch the plastic gears falling apart before failure. Plastic gears are held down onto the shaft via set screw. This was a mistake since the screws are threaded onto the plastic itself, becoming loose over time.
The idea of the device was within the subconsciousness until Fall 2018 semester in Georgia Tech mechatronics class (ME 4405), my teammate (Jaemin Yoon) and I were given a task to come up with a novel device that uses multiple sensors, and do cool stuffs. Hence the creation of Noise Avenger.
The goal of this project is to:
- use MSP432 microcontroller to control robot's behavior
- knock the wall in proportion to the input noise (bigger the noise, louder the knock)
- display functioning device in the Mechatronics expo
- get an A
Parts I used for this build:
Basic structures
|
6061 aluminum
|
Ball bearings
| 8mm ID open bearing (part #5972K91 from McMaster) |
Main Motor
|
9.7:1 Metal Gearmotor 25Dx48L 12V
|
Motor controller
|
VEX 29 motor driver
|
Servo
|
HXT900 9g servo
|
Arm angle feedback
| Potentiometer |
Sound sensor
|
SEN0232
|
Battery
| 3S 500mah 35C * 2 in parallel |
Micro Controller
| MSP432 |
Overall CAD model |
Of course, CAD model is designed before the fabrication. This was done with SolidWorks. The wooden platform on which the device is attached is designed to simulate the wall.
2 sets of gears for replacements in case something goes wrong |
The motor used for raising up the pendulum is rated for 1000 rpm at 12V, which is too fast for our application. Therefore, 16:1 gear reductions were 3d printed to bring down the speed while increasing the output torque.
Getting assembled |
Basic structures are water-jetted from 6061 aluminum plates. The tolerance of the water jet was awful (off by -0.011 inch when cutting inner diameter) so majority of the pieces had to be fixed by either dremeling or drilling, especially where the bearings would stay. This process produced a lot of metal dusts which easily went into the open bearing, often preventing them from operating smoothly (should have used sealed bearings). Transparent acrylic plates holding the aluminum structures together were nice for displaying internal parts to the audiences at the expo. Also, maintenance was easy since I was able to catch the plastic gears falling apart before failure. Plastic gears are held down onto the shaft via set screw. This was a mistake since the screws are threaded onto the plastic itself, becoming loose over time.
Sound sensor placement on the "wall" |
The sound sensor used is SEN0232 which was selected due to it's cheap cost and availability (fast shipping). It was mounted directly on the wall so that it has higher chance to pick up noise coming from the other side.
Potentiometer placed inside the gear shown above |
Potentiometer is installed inside the gear which spins correspondingly to the gear behind it. This is to measure how much the knocking pendulum is brought up since the attached gear spins at the same angular velocity. The microcontroller reads the voltage change due to the potentiometer adjustment via ADC and determines when to stop the main motor.
Here's how it works. When the sound sensor detects the noise, the motor will turn, bringing up the arm that has a small 9g servo attached on its end. This servo has it's horn extended at the moment so the arm, while moving up, also carries the knocking pendulum with it. Once the pendulum reaches its desired height in proportion to the noise input, the main motor stops. The height feed-backed by a installed potentiometer which turns at the same time the arm is moving up, proportionally with 1:1 ratio. Once the motor stops, the servo releases the pendulum, which falls down and knocks the wall. Being a friendly neighbor, the "noise value" is the average of 4 measurements in 1 second from the sound sensor so that the accidental/loud but short noises can be ignored as a courtesy.
Following video is a demonstration of it working.
Haha!
Document used for presentation is attached here
To see technical review of this project, click here
The expo was a success! The device worked most of the time and picked up deliberately made noises quite well. Although not perfect, pendulum was brought up to the desired height somewhat in proportion to the loudness of the noise made. A lot of people came by and made noises to trigger the device to hit the wall.
Device wasn't perfect however, as several improvements can be thought which are listed below.
Improvements for future iteration:
- higher quality sound sensor for larger range of reading
- timer algorithm implementation - to decide return period and magnitude depending on the input noise
- use stepper motor instead of DC brushed motor for better resolution
- find better way to secure gears on the shaft (relying on plastic threads were a bad idea)
- sealed ball bearings are must (prevent dust going in)
- use absolute encoder instead of potentiometer (for higher angular resolution)
- laser cut basic structures from wood instead of using metal (save weight - less stress to the wall)
Thanks Jaemin for being an awesome teammate, helping me make this into realization. Although not so much of an ethical idea was used into this project, it was fun and the plan my roommates and I had from 2017 Summer was finally fulfilled!
P.S. I don't use this device in my room (hopefully, never have to!). I love my neighbors.
P.S. I don't use this device in my room (hopefully, never have to!). I love my neighbors.
No comments:
Post a Comment