Auto Harmonizer EEL 4924 Electrical Engineering Design (Senior Design) Preliminary Design Report 2 February 2012 Project Abstract: Team Name: Slubberdegullions Team Members: Josh Elliott and Henry Hatton, Jr. Our project consists of creating an Auto Harmonizer which will capture a pure tone and generate at least a two-part harmony output, although not in real time. This project will rely extensively on digital signal processing to accomplish the harmonization, but will also utilize analog circuitry to allow for the inclusion of adjustable equalization. The device will be able to take an analog input, perhaps from an XLR port or ¼ inch audio jack, and then output the harmonized signal to a ¼ audio jack. Additionally, users will be able to customize the response of the device through the use of knobs which will adjust the equalization of the output signal. The two most important technical components of this project are the programming of the DSP and the design of the analog equalization circuitry.
Table of Contents Introduction.....3 Objectives...3 Additional Goals....4 Design Aspect Digital / DSP...5 Design Aspect Analog....7 Analog Equalization Technology...9 Cost Estimate.....9 Division of Labor.....9 List of Tables 1. Center Frequency Bands...7 List of Figures 1. System Flowchart......3 2. Software Flowchart......6 3. Band-Pass Filter 1 Circuit......7 4. Band-Pass Filter 2 Circuit......8 5. Band-Pass Filter 3 Circuit......8 6. Band-Pass Filter 4 Circuit......8 7. Band-Pass Filter 5 Circuit......9 8. Gantt Chart.......10 2
Introduction The heart of this project lies in the development of a system that will allow the user to harmonize an input signal with at least one frequency shifted copy of itself. Furthermore, the user will need to be able to adjust the quality of the output to his tastes through an adjustable, analog equalization circuit. Figure 1 below shows the System Flowchart. INPUT Switches and Potentiometers PREAMP DSP Analog Filters AMP OUTPUT Figure 1 System Flowchart Project Objectives Digital / DSP Non-real-time harmonization of an input signal with a frequency shifted copy of itself. The input will be recorded for a short period of time; data capture will occur at the behest of an external trigger. Data output will also occur in response to a hardware trigger. The input signal will be a pure tone to simplify the process of finding the fundamental frequency. Digital modulation will be used to create the frequency shifted version of the signal. Some form of memory, most likely an SD card, will be used to ensure there is enough space to hold the sampled input. Project Objectives Analog Pre-amplifying the microphone input signal. Provide 5 bands of equalization to include the following center frequency channels shown in Table 1. Utilizing active low-pass, band-pass and high-pass filters. Minimize noise in the system. Maximize dynamic range. Reduce RF susceptibility. Amplifying the output signal. Providing adequate anti-aliasing filters at the input and output. 3
Additional Goals: While these are not expected to be in the final product, they are nonetheless important objectives that will be pursued if possible. Real time harmonization of an input. Additional samples that can be used to create the harmonies, as opposed to the original input signal. Effects implemented through analog circuitry 4
Overview Design Aspect Digital / DSP This portion of the project will rely extensively on the use of software algorithms implemented on a TI MCU to produce the required digital signal processing necessary to accomplish harmonization. Some key factors that guide the design process are as follows: Utilization of TMS320F28335 MCU (Used in EEL4744) Chosen because of familiarity and processing capabilities Not a heavy hitting DSP, but it's use will greatly simplify potential design problems Immediate availability of devboard for early testing and development Extensive amount of resources and support readily available on campus Non-real-time Harmonization Significantly reduces processing speed required Makes the use of FFTs much more feasible Assumption of Pure Tone Input Will certainly make discerning the fundamental frequency more attainable Only one main frequency component will make frequency domain operations simpler. Apparent Challenges There will undoubtedly arise more unforeseen challenges as the project undergoes development, but there are a few fundamental issues that are salient. Fast Fourier Transforms Difficult algorithm to get properly functioning on a small MCU Utilizing existing TI libraries will hopefully make the process easier Pitch Shifting Captured input signal needs to be copied from memory and altered into a pitch shifted variant of itself. Current plan is to modulate the input signal with a sinusoidal carrier signal that will have a frequency that corresponds to the desired harmony. 5
Preliminary Software Flowchart The flowchart in Figure 2 below describes the basic structure of the algorithm that will be used to accomplish the harmonization of the signal. The flow depicted below will occur every instance in which an external trigger is detected. Figure 2 Software Flowchart 6
Design Aspect Analog Provide a multi-band active graphic equalizer to fine tune the harmonized output signal. Active filters are favored for the low cost, light weight, small size and gain availability. Using 2 nd order filters consisting of 2 capacitors and 2 resistive pairs. The center frequency (fₒ) and bandwidth are fixed for each channel. Table 1 Center Frequency Bands Band fₒ (Hz) 1 80 2 240 3 750 4 2200 5 6000 We need to experiment with input pre amplification circuitry and output amplification circuitry. Note that the following filter circuits are a preliminary design. Component values will most likely change due to the fine tuning of the frequency response by adjusting the quality factor, gain and corner frequency attenuation. Figure 3 below shows the 2 nd order filter for Band 1 designed with a center frequency of 80 Hz. Figure 3 Band-Pass Filter 1 Circuit fₒ = 80 Hz 7
Figure 4 below shows the 2 nd order filter for Band 2 designed with a center frequency of 240 Hz Figure 4 Band-Pass Filter 2 Circuit fₒ = 240 Hz Figure 5 below shows the 2 nd order filter for Band 3 designed with a center frequency of 750 Hz Figure 5 Band-Pass Filter 3 Circuit fₒ = 750 Hz Figure 6 below shows the 2 nd order filter for Band 4 designed with a center frequency of 2200 Hz Figure 6 Band-Pass Filter 4 Circuit fₒ = 2200 Hz 8
Figure 7 below shows the 2 nd order filter for Band 5 designed with a center frequency of 6000 Hz Figure 7 Band-Pass Filter 5 Circuit fₒ = 6000 Hz Analog Equalization Technology Equalization filter circuitry is used in applications such as program enhancement, sound reinforcement, telecommunications and data acquisition. There are many different types of equalizers. A passive equalizer consists of inductors, capacitors and resistors and does not require power to operate. The advantages for the passive equalizer are low noise performance, good reliability, low RFI interference susceptibility and high dynamic range. The disadvantages for passive equalizer are large size, weight, cost and the need for shielding. The active equalizer features operational amplifiers and various other components and require power to operate. The advantages for active equalizers are small size, light weight, low cost and gain availability. The disadvantages for the active equalizer are increased noise, small dynamic range and RFI susceptibility. Cost Estimate The preliminary estimate cost for all the components required to assemble one device is $100. However, there are development costs associated with the purchase of a development board as well as a TI DSP emulator that will be used to program assembled boards. The estimated cost of the development parts is $200. Division of Labor Josh Elliott Interfacing of MCU with peripherals DSP software algorithms for MCU PCB for MCU and digital components Henry Hatton, Jr. Design equalization circuitry Design input pre amplification circuitry Design output amplification circuitry Design PCB board for the analog circuitry 9
Figure 8 Gantt chart 10