6-DOF PPOD
Overview
Our work on the portable 6 degree-of-freedom (DOF) programmable part-feeding oscillatory device (PPOD) is based off of the previous work of (INSERT LINK?) Tom Vose in the Laboratory for Intelligent Mechanical Systems (LIMS), as well as (INSERT LINK) Ankur Bakshi, Donald Redding, and Ben Tollberg from the previous class. The PPOD is a vibrating table driven by six speakers acting as actuators. The speakers vibrate at specific frequencies and phases to create force patterns on the surface of the disk that are capable of moving small objects on pre-determined paths. Previous versions used MATLAB to control the entire system.
For our project, we attempted to control the speakers and feedback operations using only the NU_PIC32 microcontroller board. The PIC32 periodically varies a pulse width modulation (PWM) duty cycle (INSERT LINK), which, when sent through a low-pass filter (INSERT LINK), forms a sinusoidal wave that is amplifies to drive the speakers. Readings from accelerometers mounted on the PPOD are compared with the sinusoidal waves using Fast Fourier Transform (FFT) (INSERT LINK), thus creating a transfer function matrix. This matrix characterizes the relationship between the input and the output of the system, and represents the PPOD’s calibration.
Team Members
Jonathan Drake (Mechanical Engineering, 2010)
Caitlin Ramsey (Mechanical Engineering, 2010)
INSERT TEAM PICTURE
Mechanical Design
There are two main mechanical components to our PPOD—the base and the disk. The base is a 20 inch diameter disk made of ¼” thick acrylic. This is where the speakers and electrical components are attached. There are 3 speakers used for vertical motion and 3, on shelves attached to the disk, used for horizontal motion. The motion of the speakers is transferred to the disk, the vibrating surface, through a series of mechanical connections:
- 1/2" diameter rod is superglued to the interior cavity of the speaker
- 1/4" diameter rod is inserted into a hole in the half inch diameter rod
- Flexible tube is then fitted over the 1/4" rod. On the other end of the flexible rod another small quarter inch rod is inserted
- The smaller quarter inch rod is inserted into rectangles that have a quarter inch hole cut out of the center. The rectangle is then glued to the underside of the disk.
- The horizontal speakers have an extra piece each, a two inch rectangle that is glued to the disk one half and the rectangle is glue to the side of the rectangle. The flexible tubing allows the disk to move without losing the connection with the speakers.
The accelerometers are attached to 1x1x1 inch cubes of solid acrylic that are glued to the underside of the disk. There are three accelerometers attached to the table. The accelerometer mount side of the cube has four holes drilled into each corner to screw the accelerometer board into place. A notch was milled out of the surface for the solder tips, allowing the accelerometer to be flush with the surface of the cube ensuring an accurate reading.
The disk is 11.5 inch in diameter cut from quarter inch thick acrylic. The speaker attachments are all 60 degrees apart, alternating between vertical and horizontal speakers. All speakers are attached in a symmetrical pattern ensuring the accuracy of the calibration program and the force patterns on the surface. Acrylic was chosen for all the components because of its workability in the Ford machine shop and for its ability to be glued together quite easily.
Circuit Design
The PIC32 controls the entire system. We will explain the circuit, following the path from PIC to speaker, for one speaker. Theoretically, if our project was extended, 6 speakers would be included, but we concentrated on a proof of concept with one speaker subsystem. We connected an output compare (OC1) pin to a low pass filter (LPF) circuit, which converts the square wave duty cycle of the PWM into a sinusoidal wave. We chose appropriate values for the resistor (66K) and capacitor (0.01uF) that gave us a good resolution and response time compared to the duty cycle period coming from the PWM. The output of the LPF was connected to the non-inverting input of a unity gain buffer, as well as back into an analog input pin (AN0) on the PIC. This simple LM741 opamp conditions the noisy sine wave of the PIC for input into the amplification circuit. We powered this opamp with +/- 12V. The output of the buffer is connected to the input of the audio amplification circuit, designated Vin on the diagram. This chip, the LM1875, is powered with +/- 12V, and has a gain of 20 (modified by a resistor value). To deal with noise, we boosted the capacitors connected across the rails to 1000uF. The output of the audio amp was connected to the positive terminal of the speaker. The negative terminal of the speaker was connected to ground. The accelerometer (again, theoretically 3, but we are explaining 1) is attached to the vibrating table disk, and powered by +3.0V. The X, Y, and Z-axis pins of the accelerometer are connected to analog input pins on the PIC: AN1, AN2, and AN3 respectively. We used a +/- 12V boardlet to power the buffer and amplifier, and the NU32 board’s +3.3V to power the accelerometer.