Loudspeaker Active Spectral Divider Design Project Team 3: Aaron, Sam, Brandon, & Matt

Project Report Evaluation Team ID: 03 Team Member 1.0 2.0 3.0 4.0 App Tech* TOTAL Aaron Barnes Sam DiCarlo Brandon Diebold Matt Wolfe Maximum Points Possible 10 30 10 10 20 20 100 * technical content, writing style, professionalism, clarity, completeness Instructor comments: 2

TABLE OF CONTENTS Abstract 1 1.0 Introduction 2 2.0 Design Narrative 3 3.0 Results 6 4.0 References Appendix A: Activity Logs 8 Appendix B: Schematic (Analog) or Tuning Parameters (DSP) 13 Appendix C: Frequency Response Measurement 1 Appendix D: Subwoofer Design Documentation 21 Abstract Your job is to design an active 3-way spectral divider (crossover network) for a stock 3-way (tri-amplified) labyrinth loudspeaker system. Both the loudspeaker system and class D amplifier modules (4 x 100W) with power supply will be provided. Your spectral divider can either be analog (op amp based) or digital (DSP based, implemented using a stock development board). Level controls should be provided for each loudspeaker output drive. Primary goals include optimizing frequency response (target ±3 db over the range of 40 Hz to 16,000 Hz) and minimizing THD (total harmonic distortion). For bonus credit an (add-on) subwoofer can be designed and constructed, which will require a 4-way spectral divider. For analog designs, a PCB for the filter circuit can be created for additional bonus credit. 1

1.0 Introductions My name is Aaron Barnes and I am a junior studying electrical engineering. I am taking this course as an elective because I was interested in understanding the physics of sound propagation and learning more about the electronics that are used in audio systems. Additionally, I enjoyed the project-based learning of ECE 362 and thought this course would offer a similar experience. I learned how to balance a speaker s frequency response using pink noise during this project. Although I am not actively searching for a career in audio right now, this course is a good foundation for any future audio-related hobby work I may do. My name is Sam DiCarlo and I am a senior majoring in Sound for the Performing arts. The reason I took this course is to learn more in depth the theory of how sound works and to get a better understanding of the physics and math behind sound. I know a lot on how to use sound boards, amps, speakers, etc. as I have been mixing and setting up sound systems since I was in middle school. From taking this class so far I have learned a lot on the physics and mathematics behind sound. This project especially, helped me learn how a speaker is built and how to build one from scratch. It has taught me a lot on the equations used like the inverse square law and the physics on how to make a speaker louder by adding a vent. My goal in life is to go on tour with a band being a production sound engineer or a Front of House engineer for a band. Taking this course has really made a lot of questions I have had about sound more clear. My name is Brandon Diebold and I am a senior in electrical engineering. I decided to take this course because I had little experience with audio systems. Rather, my interests were in communications and optical systems. I was interesting in learning how the concepts of analog and digital communications systems would be used to design a physical audio system. I was also curious to see how the physics of sound systems compared with optics. Working on this project has taught me a lot about speaker design. I now understand how they are built and why they are designed that way. I also learned how pink noise can be used to balance the frequency response of the speaker. Although I am not currently searching for a career in audio, this course provided a good basis for any future work involving one or more sound systems. My name is Matt Wolfe and I am a senior in electrical engineering. I chose to take this course because I had very little to no knowledge of how audio systems work. Furthermore, I am interested to apply the mathematics and physics concepts I have learned in other courses to everyday devices (such as speakers). I also took EEC 40020 due the project based nature of the course. I believe that this is the best way to fully understand engineering material. Having no prior knowledge as to how sound systems worked, I found the use of frequency crossovers to be fascinating. As it pertains to this project, I learned how important it is to get these crossover points correct, and how much of an impact they have on the acoustic result. I have always been an avid gaming computer enthusiast and would love to implement what I have learned in this course. My goal, after taking this course, is to build my own speaker system for my gaming rig. 2

2.0 Design Narrative We chose to design an analog spectral crossover instead of the DSP approach because the team wanted the challenge of picking out the values of resistors and capacitors to implement filters with operational amplifiers on a breadboard. Since we all had experience using software design tools, we were excited for an opportunity to once again build RC circuits rather than simply use software to implement the circuit by specifying filters and their cutoff frequencies. While we knew that the analog approach would require more time to design and be harder to troubleshoot, we considered these acceptable tradeoffs for what we considered the more interesting approach. The only modification made to the stock 3-way labyrinth loudspeaker system was to replace the tweeter driver with an alternative driver found in class. Replacing the driver of the tweeter was deemed necessary after the listening to the stock tweeter with the midrange and woofer disconnected. Regardless of how much the gain of the tweeter output channel was increased, the sound coming through the stock tweeter driver was muffled and distant. By replacing the tweeter driver a much brighter high frequency sound was achieved. The team created a design using fourth order Bessel filters which were realized with TL02 operational amplifiers. According to an article published by Rane Audio, Bessel filters have a maximally flat group delay with a linear phase response in the passband (Bohn 2006). In addition to the desirable phase response of the Bessel filter, the magnitude response at the crossover frequencies of Bessel filters seemed to be relatively flat. After determining the type and order of the filters to be used, the filter component values were calculated using Texas Instrument s online webench software. The tool was useful for the initial calculation of component values which were then tweaked to match the components that were available on hand. The crossover frequencies used in the design were as follows. The subwoofer operated from 0 to 0 Hz, the woofer from 0 to 900 Hz, the midrange from 900 Hz to 6 khz, and the tweeter output all frequencies above 6 khz. The bandpass filter for the midrange was constructed using a low pass filter with cutoff frequency at 900 Hz in series with a high pass filter with cutoff frequency at 6 khz. A unity gain buffer was constructed using an op amp to provide the input to all of the filters. The crossover frequencies were selected using the frequency response of the labyrinth speaker with stock drivers, displayed in appendix B. 0 Hz was chosen as the cutoff frequency for the subwoofer since the woofer frequency response was at or above 83 db SPL from that point onward (until its cutoff). The low frequencies of the woofer which overlap with the sub were not filtered out using a high pass filter due to the idea that you can never have too much bass. 900 Hz was selected as the crossover from the woofer to the midrange since the intersection of the frequency response occurs at around 900 Hz. The previous intersection at around 400 Hz was not used in an attempt to avoid the large peak that occurs at 450 Hz in the mid-range frequency response. 6 khz was selected to avoid a similar midrange peak that occurs at about 6.5 khz. Although the magnitude of the tweeter falls off sharply after 12 khz, the driver which was used to replace the stock tweeter seemed to maintain a steady magnitude in the 12 khz to 16 khz range. 3

A number of difficulties were encountered while constructing and tuning the crossover system. First, a short in the circuit was preventing the proper voltage from being achieved to bias the op amp transistors. After this problem was rectified it appeared that no high frequency audio was being passed through the system. After using a function generator to successfully pass an 8 khz sine wave through the tweeter channel the team reached the conclusion that the problem was with the stock tweeter driver and not the crossover circuit. An unfortunate problem was encountered while testing the system with pink noise. The pink noise generated from the analyzer was at too high of a volume so a 10 kohm potentiometer was used to attenuate the signal before it was sent to the input of the unity gain op amp. To adjust the level of each channel a non-inverting amplifier was created using extra TL02s. A 10 kohm potentiometer was placed between the inverting input and output of the op amp and a 100 ohm resistor connected from the inverting input to ground. Thus the gain of the filtered output of each channel could be adjusted easily using the potentiometer. The design procedure resulted in a system capable of producing a pleasant listening experience, however it did not provide nearly as much tuning flexibility as a digital system might have. Subwoofer Design: The type of enclosure group three constructed was a sealed box enclosure for our subwoofer speaker. The primary reason for picking a sealed box enclosure over a ported box enclosure is because a sealed box provides a more tight bass. With a ported box enclosure the sound is a lot louder but sounds less controlled. Alternatively, with a sealed box the bass is tight and crisp. The team ultimately decided it was more important to have a tight bass sound rather than louder lows. Consequently, the team paid a great deal of attention to the materials used, the dimensions implemented, and the sealing of the enclosure. The components used for the design were 1/2 plywood for the enclosure of the box, nails to hold the box together, wood glue, and caulk to seal the joints. A speaker terminal plate was added in the back of the back of the box for easy connection. The speaker driver chosen was a 10 polypropylene woofer that had a power capacity of 85 W / 150W RMS/Peak. This speaker was specifically chosen given that our amp only produced 100 watts on the single channel so caution was necessary to not purchase a woofer that had a higher power capacity than what the amp could handle. After the box had been constructed and the speaker screwed into the face, all seams were caulked over to insure that the box was sealed. 4

Small Signal Parameters: **The speaker the team chose had a Qts of 0.53 because it was a sealed enclosure. The unconstrained dimensions are length: 16.25 inches, width: 26.34 inches, and depth 10.04 inches. The constrained dimensions are length: 16.391 inches, width: 16.2 inches and depth 16.2 inches. The length dimension is off by 0.12 of an inch. The width is off by 10.14 inches and the depth is off by 6.2 inches. Unfortunately the hole cut into the front was slightly too large for the speaker, so caulk was used to seal it to the box. Calculated Parameters: Vb: 0.4401 Liters Fb: 45.283 Hz F3: 40.6261 Hz Since the caulk was still drying on the subwoofer during testing the response curve provided in the lab report was based solely on the performance of the tweeter, mid, and woofer. The subwoofer the team constructed will be used in the final sound-off competition at the end of the semester. 5

3.0 Results The output of the spectrum analyzer shown in Appendix C was measured without a subwoofer since the subwoofer was still under construction. The speaker s response to pink noise hit the target of ±3 db over the range 40 Hz to 16 khz for most frequencies. While the frequency response was not completely flat over the target frequency range, only the frequencies around 630 Hz failed the constraint of ±3 db. This unexpected peak in the frequency response about 630 Hz proved difficult to eliminate by decreasing the woofer without creating a large dip in the 200 Hz to 500 Hz region. While it is possible to achieve ±3 db over the target range, the woofer is so sensitive that the slightest change would push the system past the target. The issues with the 630 Hz region may be due to the crossover from woofer to midrange being below the intended 900 Hz, creating a bit of overlap between woofer and midrange. This could have occurred since the components used to build the circuit did not match those determined by our design. It may be possible to remove the peak at 630 Hz by finding resistors closer to the ones specified in the initial webench calculations. The system produced a pleasant listening experience. The system was tested by playing a mix of salsa music and several rock tracks by the Black Keys. Instruments and voice were well balanced and distinct. While the tweeter initially sounded slightly tinny, adjusting its gain and playing it alongside the midrange brought out the liveness of the music. However, the lowest frequency components of the system seem a bit quiet when compared to the highs. The team concluded that the system needed more bass. Adding the test subwoofer improved the warmth of the system and created a more intimate listening experience. After replacing the tweeter driver the design had a crisp high end. The bass of the system without the subwoofer is a bit too distant, but adding the subwoofer greatly improves the experience. Further tweaks to that would benefit the design would be to replace the driver of the midrange and possibly crossover from the woofer to the midrange at a lower frequency. This could fix the frequency response imbalance that was measured from 630 Hz to 1.6 khz The current circuit is difficult to setup and maintain consistent quality because it is a fairly large breadboard circuit. A printed circuit board will solve many of the problems that have been occurring due to shorts from components that get jostled around. Another issue that needs to be addressed is loud distortion noises that can be heard when the input to the circuit is left floating. These noises are not heard when the input is plugged into an audio player but are quite annoying if the speakers are turned on during setup. Finally the printed circuit board needs to be modified to include the gain adjustments that were made to each channel during tuning. If the team chooses to participate in the spark challenge it may also be necessary to find a smaller power supply to provide the requisite ±5V DC to the circuit board. 6

4.0 References Works Cited Bohn, D. (2006, April). A Bessel Filter Crossover, and Its Relation to Others. Retrieved April 05, 201, from http://www.rane.com/note14.html Texas Instruments. (n.d.). TL01 Low-Noise JFET-Input Operational Amplifiers. Retrieved from http://www.ti.com/lit/ds/symlink/tl01.pdf Sealed Subwoofer Box Enclosure Calculator Speaker Cabinet Program. (n.d.). Retrieved April 06, 201, from http://www.ajdesigner.com/speaker/bcs.php

Appendix A: Activity Logs 8

Activity Log for: Aaron Barnes Role: PCB/Prototype Time Activity Date Start Time End Time Spent Crossover Frequency Selection, Filter type 2/2/1 10:30 am 12:30pm 2 hours research Op amp selection and ordering, Webbench filter calculations 2/28/1 12:00pm 3:15pm 3.25 hours Orcad Pspice schematic and simulations, BOM generation 3/3/1 1:00pm 2:30pm 1.5 hours Team Meeting, Subwoofer discussion 3/22/1 Breadboard prototype construction and 3/22/1 oscilloscope verification of crossover points, schematic updated with accurate component values 10:00am 10:30am.5 hours 3:30pm 9:00pm 5.5 hours Initial testing of breadboard circuit with speaker 3/24/1 2:30pm 3:30pm 1 hour Tweeter level troubleshooting 4/4/1 10:30am 11:30am 1 hour PCB Eagle Schematic/Layout 4/4/1 8:30pm 1:30am 5 hours 4/5/1 5:30pm :30pm 2 hours System tuning Report writing 4/5/1 8:30 11:30 3 hours Report editing 4/6/1 9:00 9:30.5 hours 9

Activity Log for: Sam DiCarlo Role: Research and Build Subwoofer/ Write Report Activity Date Start Time End Time Time Spent Team meeting and discussion 3/22/1 10:00 am 10:30 am 0.5 Hrs Research of sealed box designs and play 3/22/1 3:00 pm 6:00 pm 3 Hours around with A.j. Designer Build speaker in A.j. Designer and order Speaker Parts Initial testing of breadboard circuit with speaker Build Speaker Enclosure and buy parts to build box 3/23/1 3/24/1 3/25/1 5:00 P.m. :00 Pm 2 Hours 2:30 pm 3:30 pm 1 Hour 11:00 am 5 P.m. 6 Hours Report Writing 4/5/1 :00 Pm 11:00 Pm 4 Hours 10

Activity Log for: Brandon Diebold Role: Writer and Design Assistant Activity Date Start Time End Time Time Spent Team meeting and discussion, Op amp 2/2/1 1:00 pm 1:30 pm 0.5 hours research, Crossover frequency review Team meeting, Subwoofer discussion 3/22/1 Subwoofer design and discussion, Researched 3/23/1 speaker parts 10:00 am 10:30 am 0.5 hours 10:20 am 11:20 am 1 hour Initial testing of breadboard circuit with speaker 3/24/1 2:30 pm 3:30 pm 1 hour Checking and adjusting circuit and adding 3/25/1 4:30 pm 5:30 pm 1 hour potentiometers Tweeter troubleshooting 4/4/1 10:30 am 11:30 am 1 hour System troubleshooting, tuning 4/5/1 4:30 pm :30 pm 3 hours Report writing 4/5/1 10:00 pm 2:00 am 4 hours 11

Activity Log for: Matt Wolfe Role: Subwoofer/wiring Activity Date Start Time End Time Time Spent Team meeting and discussion 3/22/1 10:00 am 10:30 am 0.5 Hours Build speaker in A.j. Designer and order 3/23/1 5:00 P.m. :00 Pm 2 Hours Speaker Parts Initial testing of breadboard circuit with 3/24/1 2:30 pm 3:30 pm 1 Hour speaker Picked up necessary pieces for Sub 3/24/1 4:30 5 0.5 Hours Build Speaker Enclosure and buy parts to build box 3/25/1 11:00 am 5 P.m. 6 Hours wired and sealed Subwoofer 4/3/1 5:00 pm 8:00 Pm 3 Hours Tweeter troubleshooting 4/4/1 10:30 am 11:30 am 1 hour Write Report 4/5/1 :00 Pm 11:00 Pm 4 Hours 12

Appendix B: Schematic w/ Optional PCB (Analog) -or- System Tuning Parameters (DSP) 13

Orcad Pspice circuit schematic 14

Printed Circuit Board Layout 15

Orcad Pspice frequency sweep simulation 16

Appendix C: Frequency Response Measurement 1

Spectrum Analyzer Output Frequency Response of Stock Labyrinth Speaker 18

Picture of setup used for frequency adjustment testing 19

Prototyped Circuit with potentiometers added for gain adjust 20

Appendix D: Subwoofer Design Documentation (required for DSP, optional for analog) 21