Home

AUDIO SPEAKER PROTECTION FROM UNSAFE LEVELS

1. 0 Fh Fh Fh Fh Fh Fh Fh Fh H void SPI MasterInit void 54 Set MOSI SCK output all others input SPI DDR 1 lt lt DD_MOSI 1 lt lt DD_SCK 1 lt lt DD_SS 1 lt lt POT_CS 1 lt lt LED Enable SPI Master set clock rate fck 16 SPCR 1 lt lt SPE 1 lt lt MSTR 1 lt lt SPRO SPI PORT 4 void SPI_MasterTransmit unsigned char cData Set SS low Bitclr SPI PORT 1 Start transmission SPDR cData Wait for transmission complete delay ms 100 Set SS high Bitset SPI PORT 1 delay ms 10 int calculateAttenuationBracket int currentVoltage int overloadVoltage int previousBracketNumber currentBracketNumber setting up the attenuation bracket definitions if currentVoltage lt 0 9 overloadVoltage umber 0 else S taeda NOR 1 0 overloadVoltage umber 1 else m NAT lt 1 1 overloadVoltage umber 2 else us ATR lt 1 2 overloadVoltage aren umber 3 else ie ers lt 1 3 overloadVoltage ieee umber 4 else ni lt 1 4 overloadVoltage umber 5 else ne lt 1 5 overloadVoltage umber 6 else ee lt 1 6 overloadVoltage UN RE umber 7 else E lt 1 7 overloadVoltage currentBracketNumber 8 else else else
2. eene eere 25 Figure 11 1 8 27 Figure 12 ADC step down vs high voltage cosine and high voltage music 28 Figure 13 Full wave rectification diode in series low pass 28 Figure 14 SPI communication wiring sses neret rene 29 Figure 15 General input vs output trend of attenuation circuit throughout the range of Ete eoa REN AS EUR eet ep PEE ur eee OR 40 Figure 16 Input vs output trend of attenuation circuit low range 41 Figure 17 Circuit Attenuation in dB for a Low Range of Input Signals 43 viii KEY TO ABBREVIATIONS ADC analog to digital conversion a general signal conversion method also a register defined on the Atmega 32 microprocessor for carrying out such a conversion CS chip select an input pin to the MAX 5411 digital potentiometer active low PCB printed circuit board copper platforms whereupon electronic circuits have been etched PCBs are rugged inexpensive and can be highly reliable They require much more layout effort and higher initial cost than either wire wrapped or point to point constructed circuits but are much cheaper and faster for high volume production QSOP Quarter Size Small Outline package a chip package with p
3. St 1 attsig o 212 nun 330nF TLC2272ACP x2 regulators to regulate 12 ground ved volts and 12 volts to 5 volts R3 DIGPOT small sig input 8 G 330nF 330nF HDR1X2 ground Figure 11 Analog schematic small pig HDR1X3 DIGPOT 2 0 and 5 volts A number of 4 3 330nF 330nF resistors were using to bias to high and low terminals which are sent to the MAX5411 Two large 33uF capacitors were used to couple and DC signal connected to the music inputted Two capacitors were used because 33uF capacitor are polarized electrolytic thus the terminal charges must be symmetric So the two positive sides of each capacitor were connected leaving the negative terminals on the outside The power was regulated to 5 volts and 5 volts because the op amps used the analog PCB circuitry required 5 volts Fast rail to rail op amp built by TI TLC2272 was used as a voltage follower for the small signal output 27 The TLC2272 was also used step down op amp Since the large personal amplifier outputs a high voltage signal of up to 100 volts peak to peak The first stage 05 of the step down circuit op amp reduces the gain of the signal to 10 volts peak to peak The next two op amps USB and U6A full wave rectify the signal To smooth out the signal a low pass filter was built with a capacitor and resist
4. It was decided that one printed circuit board PCB would be built to handle the digital components and another board would handle the analog elements The first digital board was built to house the atmega32 processor digital potentiometer maxim Max5411 12 volt to 5 volt regulator LM7805 and headers that plug into an LCD display Figure 10 shows how a cosine signal was attenuated using the digital potentiometer The 5411 was chosen because it had a logarithmic taper digital with 32 tap points each It can replace mechanical potentiometers in audio applications 2295 Ed D aiU eat Figure 10 Attenuation of a cosine signal 25 requiring digitally controlled resistors The chip also has a 5 1 serial interface that controls the wiper positions The MAX5411 has a factory set resistance of 10kQ A zero crossing detection feature minimizes the audible noise generated by wiper transitions Switching amplitudes through a digital potentiometer when there is any sort of voltage cause small audible cracks The zero crossing feature has the potentiometer wait until the low input and high input are equal to change its resistance For the purpose of mounting the MAX5411 on our PCB the 16 pin quarter size small outline package QSOP package was chosen 10 mil traces were routed to the 16 pin QSOP The digital board would receive the audio signal needing attenuation through its high input on the
5. LCD STRING overloadVoltage string LCD ADDR 0x40 to the beginning Print text to LCD Print ASCII adc value Print text to LCD Print ASCII overloadVoltage value Move cursor to second line LCD STRING Attenuation LCD STRING previousBracketNumber string LCD STRING dB delay ms 30 Pause for 100ms before looping again S EEKE KK kk k kk k k k k k k k k k k k k k kk kk k ke k k k k k k File Name avr32 adc h S EEKE KK k k k k k k k k k kk Ck kk k k kk Kk k k k k k ke kk k ke k k k k k k void ADC_INIT void ADMUX 0b00000000 ADCSRA 0b10000110 ADC clock prescalar of 64 int ADC_START unsigned char channel ADMUX ADCSRA while ADCSRA 6 return ADC channel 1 lt lt ADSC 1 lt lt ADSC Start conversion wait for conversion to complete S BRK KKK k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k File Name digpotcontrolfinalbethany h S BKK K K k k I k k k k k k k k k k k k k k k k k k k k k k k k k k k k IA k k k k k k k include lt avr io h gt include util delay h include math h define SPI_PORT PORTB define SPI_DDR DDRB define DD MISO 6 ine DD MOSI 5 ine DD SCK 7 ine DD SS 4 ine POT_CS 1 ine LED 0 ine Bitset port pin port 1 lt lt pin ine Bitclr port pin amp 1 lt lt
6. Ross Monty Limiters Unlimited 1997 RaneNote 5 Apr 2009 lt http www rane com note127 html gt Sound Equipment Resources 2009 Images Entertainment Superstore 11 Apr 2009 lt http www images2 co uk Sound_Equipment resources_loudspeakers_explaine d blowing a_speaker html gt THAT Corporation Application and Design Notes 2009 THAT Corporation 15 Feb 2009 lt http www thatcorp com appnotes html gt TruPower Limiting 2009 Meyer Sound Labratories Inc 11 Apr 2009 lt http www meyersound com products technology tpl_tech htm gt 49 APPENDIX ADDITIONAL PROJECT MATERIAL 50 51 S EEKE KK k k k k k k k k k Ck k k Ck Ck kk k k k k k k k k k k k k File Name Bethany4 10 c S EEK K K K k k k k k k k k k k k k k k k k k k k k k include lt avr io h gt include lt avr interrupt h gt include util delay h include spi h include lt math h gt include adcAveragingAndLCDfunctionsBethany h include digpotControlFinalBethany h include lt stdio h gt define SPI_PORT PORTB define SPI_DDR DDRB define DD MISO 6 define DD MOSI 5 define DD SCK 7 define DD SS 4 define POT_CS 1 define LED 0 int main void DDRC 0x00 SPI PORT 0x01 delay ms 10 int speakerImpedence int speakerPowerRating int overloadVoltage int previousBracketNumber 0 initially no attenuation unsigned
7. else else else else else else else else else else else else else else else i i i i current current current Current current current current current f curren f curren f curren f curren f curren f curren f curren f curren f curren f curren f currentVoltage f currentVoltage f currentVoltage f currentVoltage f currentVoltage f currentVoltage f currentVoltage currentBracket currentBracket currentBracket currentBracket currentBracket currentBracket currentBracket currentBracket currentBracket f currentVoltage Bracket Bracket Bracket Bracket Bracket Bracket Bracket Bracket tVoltage tVoltage tVoltage tVoltage tVoltage tVoltage tVoltage tVoltage tVoltage tVoltage lt 1 8 overloadVoltage umber 9 lt 2 overloadVoltage umber 10 lt 2 3 overloadVoltage umber 11 lt 2 7 overloadVoltage umber 12 lt 3 2 overloadVoltage umber 13 lt 3 7 overloadVoltage umber 14 4 5 overloadVoltage umber 15 5 4 overloadVoltage umber 16 lt 6 6 overloadVoltage umber 17 lt 8
8. 37 and 90 dB translates into 0 0002542 which would mean that using one of these attenuation levels would theoretically allow for an input that is either 141 8 or 3540 5 times the level of our overloadVoltage safety threshold level while still attenuating this input to attain an output value that is still only 0 90 times that threshold value or less However first of all the 90 dB wiper position is intended for use as a mute function and if we were to specify that our signal could function up until such extremely high input levels under the 90 dB setting we would no longer have a way of muting the input signal if the level was ever exceeded or for other emergency purposes More importantly as previously stated a level of 25 overloadVoltage should be high enough that regardless of the speaker s parameters the input would never reach levels high enough to even come close to this limit thus we would expect the mute function of our device never to have to be employed Our final attenuation brackets allow for inputs to be smoothly attenuated until the input reaches a level twenty times the threshold value Between twenty and twenty five times this threshold value we start attenuating the signal toward a zero output in order to alert the user that they are operating at extremely high levels and the amplifier s gain either needs to be turned down significantly or if the gain continues to increase our device will soon merely mute the input sign
9. However since we are taking the power rating of the amplifier into account in our calculations this should allow our device to prevent the amplifier from reaching the range where it would begin such distortion Therefore our signal not only functions to protect our speakers from being damaged but also to prevent distortion in its signal output LIST OF TABLES Table 1 overloadVoltage Calculations given all possible user input combinations 19 Table 2 Potentiometer settings and attenuation 39 Table 3 Division o labor 2 ecd edet guias 46 Table 4 materials st o as ute eee d ao te enses 47 Table Charte s e te ee e tun adeat etaed uu 48 LIST OF FIGURES Figure 1 Speaker components eene iii Figure 2 Eeedback control e ee EORR RERO Ed ORE HA E CUR RE HER 1 3 uie HARI e e Re RAD 3 Figure 4 Example of feedback limiting essere eene 5 Figure 5 THAT 4301 Analog Engine sse rennes 12 Figure 6 High level system ener trennt 15 Figure 7 Software 1 E 17 Figure 8 Revised software 23 Figure 9 Digital BGB 25 Figure 10 Attenuation of a cosine 51
10. Input vs Output from Attenuation Circuitry implemented that attenuation For instance if we detect a significant jump in the signal level if that signal was in fact a crescendoing signal increasing steadily in 10 15 20 volume by the time our device Figure 15 General input vs output trend of MUS signal will be even higher than we anticipated Under this 10 safety net system that signal would have room to still be higher than we expected and be attenuated to a safe level The other possible discrepancy this allows us to compensate for is the existence of peaks in the system that are either undetected by our ADC sampling or are overlooked by our calculation of the ADC statistic we decide to base our attenuation on The possibility of our system not detecting a peak in the system is very unlikely since based on a 8 MHz internal clock rate of the Atmega 32 microprocessor our use of a division of 64 in the initialization of our ADC conversion see avr32_adc h file in our code in the Appendix and the 8 clock cycles that is typically required to carry out each analog to digital conversion that gives us a sample rate of about 15 625 kHz If we consider that the human range of hearing can only detect tones up to 20 kHz and that 4 186 kHz is the highest frequency of a standard piano we can state with a reasonable amount of confidence that our system should be able to detect nearl
11. concept we knew that we needed to run a wire between the microprocessor and the potentiometer in order to enable communication to the device each time before any byte of information could be transmitted After some further research on our MSTR bit setting issue we found this sentence in a Wikipedia article entitled AVR SPI C Snippets When running master mode a low level at the slave select pin SS will force the device into slave mode At first this seemed completely contradictory to our basic understanding of SPI communication for clearly figure 14 shows the SS pin on the SPI master connected to the SS pin in our case called CS chip select on the slave both being active low with the master being in charge of selecting which device it will be talking to However we were not cautious in noticing a slight technicality in the operation of the SS pin on the microprocessor In our case we did not need the pin specified as SS on the microcontroller for its primary function was in fact to operate as its own version of our potentiometer s chip select pin in the case that someone were to need the capability of 33 having the microprocessor serve the function of a SPI slave device where this pin would be configured as an input and then driven low by an outside SPI master device when communication to this microprocessor was desired However the simple fact that this functionality was not needed by our system did not mean that we we
12. hence a mechanical failure on the part of the voice coil or the voice coil itself or the wires leading to the voice coil can literally melt causing oftentimes an open circuit in the wiring and thus inability of the speakers to produce sound of any kind The system is not put at risk of heat damage at the occurrence of sharp spikes of high amplitudes but only at a continuously high level of sound being producedand for a reasonably long amount of time In our device design our goal is to protect the speaker from both mechanical and heat damage Our protection algorithm does not involve an integration of the power dissipated by the speaker over time however which would be necessary for an absolute guarantee that an over heating issue would not occur Since we are limiting the amplitude of the signal entering the speaker however attenuating the signal appropriately based on both the speaker s power rating and impedance we would expect that under normal operating conditions where the speaker is driven with normal music for a couple hours at a time our device would offer adequate protection from both mechanical and heat damage A Note About Power Ratings and Distortion In our attenuation calculations we must often incorporate the power ratings of both the speaker as input by the user and the power rating of the amplifier we are using We must be aware however that even though these power ratings appear to be on the same scal
13. terminal of a TLC2272 op amp As the potentiometer s resistance goes down the resistance lowers and the signal attenuates The output of the op amp is fed back into the input terminal making it a simple voltage follower The low terminal of the digital potentiometer receives a 2 5 Volt DC signal The high terminal of the potentiometer receives the music with a Vcc 2 2 5 Volt bias To make sure no DC component is present in the original signal coupling capacitors were used on the input to the resistors and on the output of the TLC2272 op amp 24 5 The premise of the hardware for the limiter and speaker protector was using a potentiometer to attenuate the small signal of an iPod or guitar before the signal reaches the input of the guitar amplifier The _____ 3300 ny Wi n TE LM7805CT 82 T 225 HET Pee eee potentiometer would be digitally controlled by an VOC PB5 MOSI atmega32 microprocessor using SPI protocol serial communication The processor made the mem m n mb esso agi pe a decisions of controlling the digital potentiometer io to change the music volume based on a signal we e fed back from the amplifier Figure 9 Digital PCB 5 1 The Digital PCB
14. 44 7 4 Gantt Chart project Preliminary Proposal UN EN _ Research Order components N NM NM NM NM NM NM BN BM BN BM NM NM NM Circuit Design Paul E Nm NH E E NM E Tt EE E EH HN UN WW EH EH NH NH NH E EH NH EH E Breadboard Circuit Paul Circuit Testing Paul Coding Processor Bethany PCB Design Paul Milling PCB Populating PCB Testing and Validation B E Enclosure co Final Reporting Table 5 Gantt Chart 48 BIBLIOGRAPHY 8 bit AVR Microcontroller with 32K Bytes In System Programmable Flash Atmega 32 Atmega 321 2008 Atmel AVR SPI C Snippets Open AVR 2009 Wikipedia 27 Mar 2009 http www openavr org AVR SPI C Snippets Clark Richard A2420 Some Facts About Distortion and Speaker Damage 2000 Autosound 9 Apr 2009 www monstercable com mpc stable tech A2420 Some Facts pdf Dual Audio Log Taper Digital Potentiometers 2005 Maxim 3 Feb 2009 lt http www maxim ic com quick view2 cfm qv pk 2567 Korn Huelsman Wait Introduction to Operational Amplifier Theory and Applications New York McGraw Hill 1992 LSP 1 Concert Series Speaker Protector Owner s Manual 2000 ARX 23 Jan 2009 lt http www arx com au pdf old LSP 1 manual pdf gt McCarthy Bob Sound Systems Transmission Focal Press 2007 28 30
15. Bethany was also the main contributor towards composing and doing research for the final report and she designed and constructed the poster for the final demonstration in the NEB rotunda In addition to deciding what exact functionality was needed in hardware in order to implement our design designing both the attenuation and step down circuits for our system designing and populating the PCBs and debugging the hardware during both the breadboard and PCB stages Paul also was responsible for the finishing touch work on our hardware in order to make it look professional for our last presentation in the NEB rotunda including buying and soldering custom plugs for our device s external wirings adding LED lights to show audio output changing out the LCD screen cleaning up any wire wrappings etc 7 3 Bill of Materials Product Quantity Aluminum Face 1 4 Mono Input Jacks Red LEDs LCD screen 4x20 Winstar Power Supply 12 12 5 19 Rackmount Casing Electrolytic Capacitors Operational Amplifiers TLC2272 THAT electronics PCB standoffs Atmel Atmega32 Toggles Digital Potentiometer MAX5411 0 N Unit 12 45 4 34 0 95 18 97 0 00 0 00 0 20 0 13 1 56 4 56 0 00 0 00 0 00 Total Table 4 Bill of Materials 47 Total 12 45 13 02 19 95 18 97 0 00 0 00 4 00 0 65 18 72 13 68 0 00 0 00 0 00 101
16. Watts would likely be constantly varying between say 50 and 400 Watts of power and thus would require an amplifier of at least 400 Watts of power rating to amplify this signal without clipping Since we know that the speaker is rated based on the varying power changes in music then we could conclude that a 100 Watt speaker would be able to safely run the signal even though it is driven with an amplifier rated at 400 Watts and with a music signal containing 400 Watt peaks We should be aware that the previous example is dependent on the amplifier operating without distortion that is without the clipping of the signal peaks When the input is clipped by the amplifier the transistors saturate and automatically temporarily output their maximum supply DC voltage of the system to produce the square like waveforms characteristic of clipping When this happens the speakers are more prone to over heating damage than they would be if that signal would have been unclipped and vi reached its full amplitude peak For this reason it is often said that some speaker damage can actually be avoided by using a higher rated amplifier since it would avoid this distorting clipping from occurring In our project since we are not detecting if a peak value observed was from a clipped signal or an unclipped signal in order to avoid an increased risk of heat damage due to clipping we would need to assume that the amplifier is operating in a non distorting range
17. commanded to go to pausing for a very short amount of time at each position to allow the gradual smoothing of the attenuation have audible results Originally this stepping code only allowed the wiper to step through 1 3 additional positions depending on how far away the previous wiper position was in relation to the position being currently commanded However this was determined to be both inefficient in both the bulkiness of the code and in its ability to produce a genuinely smooth transition between positions since stepping through one extra position is no where near as smooth as a transition as is stepping through all possible wiper positions between the origin and destination positions Our final code to implement this feature was an implementation of two if loops each containing a for loop to cycle through all the potentiometer positions see the calculateAttenuationBracket function within the digpotControlFinalBethany h header file in code in the Appendix for full implementation details The introduction of this feature into our code was crucial to maintaining the aesthetic integrity of the sound we are processing NOTE In our final code some unexpected practical considerations had to be accounted for based on the results of some of the final phases of the testing our device causing some major changes to the overall nature of our system s functionality See section 4 4 for updates to the attenuation algorithms and the function of the af
18. int adcStatistic LCD INIT initialize the LCD screen SPI MasterInit initialize SPI communication SPI MasterTransmit OxFF analog power up SPI MasterTransmit 0610111111 enable zero crossing detection while 1 take in user impedence if PORTC 0011110000 0511110001 speakerImpedence 2 LEC FORTC 0b11110000 0511110010 speakerImpedence 4 if PORTC 0b11110000 0b11110100 speakerImpedence 8 take in user s speaker power rating if PORTC 0000001111 0b00001111 speakerPowerRating 100 if PORTC 0000001111 0500011111 speakerPowerRating 200 1 0b00001111 0500101111 speakerPowerRating 300 PORTC 0b00001111 0500111111 speakerPowerRating 400 if PORTC 0000001111 0501001111 speakerPowerRating 500 1 PORTC 0b00001111 0501011111 speakerPowerRating 600 if PORTC 0b00001111 0b01101111 speakerPowerRating 700 if PORTC 0b00001111 0b01111111 speakerPowerRating 800 if PORTC 0b00001111 0b10001111 speakerPowerRating 900 if PORTC 0000001111 0b10011111 speakerPowerRating 1000 overloadVoltage sqrt speakerImpedence speakerPowerRating 0 08 1023 5 calculate the overloadVoltage safety threshold adcStatistic adcPeakOrAve previousBracketNumber calculateAttenuationBracket adcStatistic overloadVoltage previousBr
19. of severe distortion of the audio signal as a microprocessor would likely not be able to produce a high quality of sound once the signal had been converted from analog to digital form and then back again The option of using a DSP chip would be beneficial in that not only is it commonly used in manipulating audio signals and reproducing a high quality of sound but also it would be capable of some very complex limiting algorithms and would even give us the option of using frequency distinction in our attenuation scheme However neither of us had very strong knowledge of digital signal processing nor any practical experience in the use of a DSP chip We were both enrolled in the Electrical Engineering Department s Introduction to Digital Signal Processing class at the time but feared we would not learn enough about the practical functionality of such a device in order to successfully incorporate this into our project with only a month or two of lessons Our next idea was to employ the use of a potentiometer to control the attenuation of our input signal in its original analog form avoiding the possible distortions and complications of digitization However a traditional analog potentiometer would need to be controlled by our system s brain and the only way to control such a potentiometer is by physically turning a knob The only way to accomplish this with electrical signals would be through control of a motor which would involve adding
20. overloadVoltage umber 18 lt 9 9 overloadVoltage umber 19 lt 12 2 overloadVoltage umber 20 lt 15 1 overloadVoltage umber 21 lt 18 8 overloadVoltage umber 22 lt 20 overloadVoltage umber 23 lt 22 5 overloadVoltage umber 24 lt 25 overloadVoltage umber 25 gt 25 overloadVoltage 56 displayADCPeakandAttenuation currentVoltage if previousBracketNumber currentBracketNumber currentBracketNumber 26 SPI MasterTransmit potCommandValue currentBracketNumber else if previousBracketNumber gt currentBracketNumber too soft SPI MasterTransmit potCommandValue previousBracketNumber 1 currentBracketNumber previousBracketNumber 1 else if previousBracketNumber currentBracketNumber too loud SPI MasterTransmit potCommandValue previousBracketNumber 1 currentBracketNumber previousBracketNumbert1 return currentBracketNumber overloadVoltage currentBracketNumber int bracketNumber int PotCommand if PotCommand 0b else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand else if PotCommand 000 00000 re
21. so any change in signal attenuation levels would not be perceived until the input voltage reached the value corresponding to 1 8 overloadVoltage Under normal operation though there may be a few peaks in this region and slightly beyond a majority of the time the signal will likely stay below this level and thus the wiper level changes should not lead to an abrupt drop in signal level The other consideration was determining at what level we should begin our attenuation in order to produce a smooth transition up towards the safety limit The trade off here was that if we started the attenuation too early then we would unnecessarily shrink the dynamic range of the audio signal with excessive attenuation even before it approached dangerous levels but if we waited too long before attenuating we would have an abrupt drop in signal level once it reached its overloadVoltage limit that would likely be distracting to the listener We sought a compromise between these two extremes and decided that attenuation beginning 42 at 0 8 overloadVoltage would be adequate for our purposes and the smoothness of our attenuation and the signal s dynamic dB of Signal Attenuation integrity can be seen best from figure 16 Once we had our signal Vout overloadVoltage attenuation levels defined there was Vin overloadVoltage one more issue to consider in order to Figure 17 Circuit Attenuation in dB for a Low Rang
22. the microprocessor where it could easily and safely be measured by the microprocessor in determining the attenuation algorithms necessary The main problem here was that once the small signal input was fed through the amplifier it was at voltage levels that were unsafe to be directly fed into the microprocessor which could only safely handle voltages roughly in the range of 0 5 5 Volts Therefore we needed to design a circuit that would step down the voltages from the amplifier from their current values to lie within the 0 5 5 Volts range We also 11 needed to only submit positive voltage values to the microprocessor so we decided to add step down circuitry that probes the voltage output fed to the speaker Since this output is a 100 volt peak to peak swing an op amp steps the voltage down by 1 10 Then op amp circuitry rectifies the negative voltages and then an active low pass filter smoothes it out This is explained further in the hardware section 3 5 A Temporary Sidetrack Early on in our project s development we encountered an alternate route that seemed like another adequate way we could solve our problem of speaker protection The website for the THAT Corporation s THAT 4301 Analog Engine see figure 5 included an application note on how this device could be used as a Signal Limiter for Power Amplifiers This chip sounded promising and included many components of the design we had already envisioned all on one chip wi
23. where it experiences an amount of voltage gain depending on the position of the Volume knob on the amplifier We will assume for our project that this gain can be varied at any point in time and do not require it to be set at a constant level for any amount of time Here is where the signal branches off we have this amplified signal both connected to the speaker where it will act as a transducer converting electrical 15 waves into pressure waves in the to produce sound and we also run this signal through a step down circuit that will attenuate and manipulate the signal in preparation for its entrance into our Atmega 32 microprocessor The current step down circuit attenuates the voltage signal by 08x and then rectifies it and smoothes it out with a low pass filter The hardware architecture for this circuit is further explained in chapter 5 the hardware section 4 2 Inside the Microcontroller Analog to Digital Conversion This small signal version of the amplifier s output is fed into a pin on our microcontroller from port A see figure 7 for the software architecture of the microcontroller Please note that all the code for our Atmega 32 processor is included in the Appendix This port s pins are linked to dedicated hardware responsible for controlling the microprocessor s analog to digital conversion functionality We utilize the microcontroller s ADC analog to digital conversion registers to configure and car
24. 567 0041524 2 771281292 2 600 801 8649622 3 919183588 4 600 1023 5 8 600 612 4344811 2 993325909 2 700 866 1131492 4 233202098 4 700 1023 5 8 700 654 72 3 2 2 800 925 9139036 4 5254834 4 800 1023 5 8 800 694 4354277 3 39411255 2 900 982 08 4 8 4 900 1023 5 8 900 731 9992131 3 577708764 2 1000 1023 5 4 1000 1023 5 8 1000 Once this voltage limit value has been scaled down by the same proportional amount used on the voltage during the step down process 0 08 in order to put it in the 0 5 Volt range that value is then multiplied by a factor of 1023 5 in order to put the parameter into a value on the same scale as that of the data from the ADC process as a value between 0 and 1023 so that the two values can be compared accurately This software as the variable newly calculated threshold value is saved by the 19 overloadVoltage and is carried along with either the peak or average parameter from each ADC set as an argument into a function establishing attenuation levels based on a comparison between the ADC set statistic and the threshold voltage for safe operation For example if the peak voltage from a set of ADC conversions is equal or less than less the overloadVoltage value the first attenuation bracket bracketNumber 0 would send a command to the digital potentiometer to position the wiper to allow for 0 dB attenuation With increasing ADC peak or average values in relation to the overloadVoltage value th
25. 771 7863 849524 9900 Overall Attenuation Circuit gain V V 2 54174E 04 0 006344658 0 007974009 0 010017459 0 012577757 0 015781718 0 019785044 0 024777644 0 030989194 0 038694486 0 048217748 0 059934712 0 074270619 0 091691784 0 112687809 0 137741483 0 167284115 0 201636184 0 240937014 0 285072579 0 333616218 0 385800493 0 440536686 0 496489492 0 552199867 0 606233744 0 657325434 0 70448641 0 747061825 0 784732911 0 817476546 0 845499799 0 869166975 Overall Attenuation Circuit gain dB 71 89736676 43 95183576 41 96646563 39 98484857 38 00793614 36 03691468 34 07325957 32 11879988 30 1757943 28 24701832 26 3358615 24 44643 16 22 58365911 20 75339154 18 96246128 17 21870491 15 53090594 13 9086306 12 36192953 10 9008911 9 535056912 8 272744426 7 12035839 6 081798781 5 158074045 4 347197873 3 64439126 3 042547604 2 532869108 2 105562662 1 750493971 1 45772983 1 217935669 Table 2 Potentiometer settings and attenuation statistics 39 SPI Command sent to Potentiometer 10011111 00011111 00011110 00011101 00011100 00011011 00011010 00011001 00011000 00010111 00010110 00010101 00010100 00010011 00010010 00010001 00010000 00001111 00001110 00001101 00001100 00001011 00001010 00001001 00001000 00000111 00000110 00000101 00000100 00000011 00000010 00000001 00000000 attenuation circuit throughout the range of operation
26. AUDIO SPEAKER PROTECTION FROM UNSAFE LEVELS OF AMPLIFIER GAIN USING SMOOTH LIMITING ALGORITHMS AND FEEDBACK CONTROL Bethany M Moatts and Paul D Muri Bachelor of Science 1 Electrical Engineering Spring 2009 ABSTRACT The device we have designed and built is an audio limiting device to automatically calculate how much limiting of the small signal is needed to protect a specific speaker cabinet The user of the device inputs their speaker impedance and speaker power rating The audio amplifier s output then can be fed back to the limiter where the limiter can detect when the power amplifier s output wattage surpasses the power the speaker can take The limiter can then lower the sound level of the audio signal using an algorithm to perform smooth attenuation in order to maintain the aesthetic quality of the audio signal TABLE OF CONTENTS ABSTRACT INTRODUCTION A Note About Speaker Damage A Note About Power Ratings and Distortion LIST OF TABLES LIST OF FIGURES KEY TO ABBREVIATIONS CHAPTER 1 PROJECT FEATURES AND OBJECTIVES CHAPTER 2 ANALYSIS OF COMPETITIVE PRODUCTS 2 1 General Approaches 2 2 Specific Examples CHAPTER 3 CONCEPT TECHNOLOGY SELECTION 3 1 The Attenuator and Analog vs Digital Considerations 3 2 The Brain 3 3 Digital Potentiometer Model Selection 3 4 Feedback Signal Incorporation 3 5 A Temporary Sidetrack CHAPTER 4 PROJECT ARCHITECTURE 4 1 From Input to Amplifier 4 2 Inside the Microcontroller Analog to Dig
27. acketNumber The above code goes through one cycle of adc converting a set of 100 values returning either its peak or its average then using that 52 info along with the overloadVoltage the old wiper position the command 1 determines the destination attenuation bracket the wiper needs to reach while defining each level as a ratio comparison of the adc reading to the overloadVoltage 2 determines how far away from that destination it is as currently 3 goes through gradual steps to set the wiper to its new position 4 returns the new bracket number of the wiper s position so that it can be set the new previousBracketNumber in the main function displayADCPeakandAttenuation adcStatistic overloadVoltage previousBracketNumber update the display of the LCD screen to show the peak from the current set of ADC data a value between 0 and 1023 the overloadVoltage value calculated from the user s input data on the 0 to 1023 range and the attenuation level of the digital potentiometer from 0 dB and 90 dB return 0 S RRR KKK k k k k k k k k k k k k k k k k k k k kk k k k k k k k k k k k k k k k k k k k k k k k k k A k k k k k k k File Name adcAveragingAndLCDfunctionsBethany h S EEK KK K k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k k include lt avr io h gt in
28. advantage over this approach to speaker protection in that we allow the audio signal to continue to play uninterrupted even once the input has reached unacceptable levels because of our limiting scheme A low frequency loudspeaker ISC RETURN LINE processor called the LSP 1 see figure 4 is built with a high signal input line INPUT to monitor exactly what the gain of the LOUDSPEAKER COMPONENT power amplifier is However for every 156 POWER PROCESSOR AMPLIFIER ISC SYSTEM PROTECTION different model of loud k Penge Figure 4 Example of feedback limiting Source this devi i PCB card http www arx com au pdf old LSP 1 manual pdf The only pre made cards available are for a few different types of speakers manufactured by their own specific company If a user wanted to use this device for another speaker not manufactured by the Australian company they would have to submit a request for a custom card to be made in accordance with their speaker type With the introduction of user inputs of their speaker specifications our speaker processor allows the user full customization for this device to work with any traditional loudspeaker model without the need of any kind of chip to be custom made for the device to function properly One speaker protection design that is currently on the market called TruPower Limiting by the Meyer Sound Company however appears to be much more sophisticated than our desig
29. al see figure 15 for this trend Finally we set the attenuation ranges or input signals in the normal operating range of 0 to 20 overloadVoltage Our first decision we make is to keep the output at least 10 below the overloadVoltage safety threshold level This will leave a small margin of error in our attenuation to first account for the slight time delay between the time the audio signal is input into the device until our device has calculated the necessary attenuation and has 38 dB 90 62 60 58 56 54 52 50 48 46 44 42 40 38 36 34 2 30 28 26 24 229 20 18 16 14 12 10 8 6 4 2 0 4 3 16228 05 7 94328 04 0 001 0 001258925 0 001584893 0 001995262 0 002511886 0 003162278 0 003981072 0 005011872 0 006309573 0 007943282 0 01 0 012589254 0 015848932 0 019952623 0 025118864 0 031622777 0 039810717 0 050118723 0 063095734 0 079432823 0 1 0 125892541 0 158489319 0 199526231 0 251188643 0 316227766 0 398107171 0 501187234 0 630957344 0 794328235 Potentiometer resistance in Ohms 0 313065488 7 863849524 9 9 12 46336158 15 69044261 19 75309692 24 86767567 31 30654884 39 41260988 49 61753613 62 4647771 78 63849524 99 124 6336158 156 9044261 197 5309692 248 6767567 313 0654884 394 1260988 496 1753613 624 647771 786 3849524 990 1246 336158 1569 044261 1975 309692 2486 767567 3130 654884 3941 260988 4961 753613 6246 47
30. another component to our device that would make matters unnecessarily complicated To circumvent the problems presented us by the traditional analog potentiometer we decided on using a digital potentiometer that could be controlled directly by the brain of our system The digital potentiometer would contain physical wipers and resistors so that the attenuation of our input signal should preserve the quality of our audio signal The drawback to this is that the digital potentiometer unlike the analog potentiometer has discrete positions for the wiper to be located at whereas the analog potentiometer s wiper position can continuously span between the high and low terminals of the potentiometer The benefit of being able to directly control the wiper s position by the same device that will perform the attenuation algorithm calculations and in a very accurate and precise manner however was seen as a benefit well worth the loss of a continuous spectrum of attenuation from our potentiometer and thus the digital potentiometer was chosen over its analog counterpart 3 2 We then needed to select a device that would act as the brain of our device This would need to first take in our user inputs and our feedback loop information from this it would not only determine the attenuation algorithms necessary to protect the speaker in question but then send commands to perform that attenuation Once we had decided that
31. clude lt stdio h gt include lt stdlib h gt include lt string h gt include LCD h include avr32_adc h unsigned int adcPeakOrAve ADC_INIT unsigned int adc val 100 unsigned int adcSum 0 unsigned int adcPeak 0 for int 1 0 lt 100 i adc 1 1 0 adcSum adcSum adc 1 1 if val i gt adcPeak adcPeak adc val i delay ms 10 return adcPeak void displayADCPeakandAttenuation int adcPeak previousBracketNumber char adcPeak 20 char previousBracketNumber string 20 char overloadVoltage string 20 Initialize ADC set current adc value at channel 4 to value in int keep tally of adcSum for this set of 100 adc conversions if new high value set that as peak this is the wait time between samples plus the 8 clock cycles or so it takes between each ad conversion as occurs in the ADC START function int overloadVoltage int sprintf adcPeak string d adcPeak Convert ADC int value to ASCII string 53 sprintf previousBracketNumber string 4 previousBracketNumber 2 Convert ADC binary value to ASCII string sprintf overloadVoltage string d overloadVoltage Convert overloadVoltage binary value to ASCII string LCD COMMAND LCD CLEAR HOME Clear the LCD and send cursor LCD STRING ADC LCD STRING adcPeak string LCD STRING OV
32. command were setting properly and that our code was in general operating correctly Through this process we immediately discovered that the bit in the SPI Control Register SPCR that was responsible to setting the microprocessor to operate as the SPI master the MSTR bit was being set properly but was being unset in the very next completely unrelated command Upon receiving some guidance from a TA teaching assistant we discovered that the active low chip select CS pin on the digital potentiometer worked more like a reset pin than it did like a traditional chip select pin whereby the device is enabled to work for the duration of time that the active low select signal is held low Therefore 32 instead of outputting a continuously low signal from the SS pin of the microprocessor to the CS pin of the potentiometer we would need to start with the SS pin in the high voltage position transition this to a low right before communication with our potentiometer began and pull it back high again once one byte of information was successfully transmitted Prior to this under the impression that we would simply need the CS pin on the digital potentiometer to always be low signifying that we always wanted to talk to that chip we had simplified the connections between the two devices by simply tying the CS pin on the microcontroller to ground and ignoring the SS pin on the microcontroller Right away we saw this new CS pin with reset functionality
33. d our digital potentiometer s chip select pin The details about the functionality of this line in regards to our single device project will be explained below in section 6 2 6 2 Trial and Error Lessons Learned Our first step in creating our device was to ensure the functionality of the digital potentiometer which was not only the pivotal function of our device responsible for ensuring speaker safety but also likely one of the most complex due to the SPI communication protocol intricacies At first we made the mistake of ignoring the sample code provided in the Atmega 32 processor thinking it would not provide fully SPI communication functionality so we attempted to get some bits of sample code from online databases Once we constructed an initial bread board based design in order to facilitate a piece of sample code commanding the digital potentiometer to move through each of its wiper positions and hooked up a multimeter to the wiper to try and verify if it was in fact following the commands sent by the microprocessor we were seeing voltage changes on the wiper and interpreted this to mean at least a partial success of the potentiometer s functionality We learned on multiple occasions however that seeing an apparent movement of the wiper unless it exactly matches the value your code is setting it to and is consistent upon running the program may times does not at all mean that the SPI communication code is working correctly T
34. digital potentiometer header Since the digital potentiometer s power supply is ground the 5 volts the music needs a vcc 2 bias voltage The music into the high input terminal of the pot is then biased by 2 5 volts The input to the low terminal is simply 2 5 volts The wiper 18 then connected to a voltage follower op amp is then plugs into the large personal amplification system 26 5volt p Fe RS 30kQ 5 2 The Analog PCB Ret Bac ps m Sika m The digital PCB also 25 3300 10080 HDR1X2 TAD fb 1Ns001GP t Usp 86 received a feedback input into svolta 10 2 TLC2272ACP its analog to digital conversion ps 114148 184148 port ADC This feedback input e i D7 R13 PER 184148 I 48 was stepped down using the T 10kQ 3080 ground analog board The schematic 5777 U4 5 logic 2 EMINSET olt n for the analog board is shown Sn 2E caius 02 0 1 1 3 EBF TDF Lici 12 pane 5volt n lt ground on the right The board 3 i Tea m volt p consists of two power ls Dus Aen PA nde ast
35. e of Input Signals smooth out the general purpose operation of our device Once we had assigned a specific potentiometer wiper position command to each appropriate range of inputs all we had was code that essentially said Given this level of input move potentiometer wiper to this position What we failed to take into account however was the possibility that the wiper would sometimes need to jump over a significant number of attenuation positions to adjust for a changing input value For example if a signal had been at a level corresponding to 0 7 overloadVoltage which would put it in attenuation bracket number 0 see digpotControlFinalBethany h header file in code in the Appendix for attenuation bracket numerical definitions and lead to a positioning of the wiper at its 0 dB position but then abruptly jumped up to a level of 1 35 overloadVoltage corresponding to a wiper in the 12 dB position under the current direct command method we would command the wiper to make the leap between these two positions instantaneously This abrupt and dramatic change in wiper levels would cause an abrupt and unnatural drop in sound presented to the listener that would likely be very distracting and would take away from the aesthetic sound of the music In order to make up for this we altered our code to step 43 through each wiper position of the potentiometer between the previous wiper position and the position the wiper is currently being
36. e one another they in fact are not The power ratings of an amplifier is based on the amount of power the amplifier is capable of dissipating with a sinusoidal monotone input into the system producing that power This method of determining he power rating is as straightforward as the power rating on a light bulb where the power a light bulb can dissipate is based on a steady ac power input from a wall socket The speaker s power rating however is rated much differently Since we would almost never expect to be using a speaker to output sound that stays at a single frequency these speakers are in stead based on typical trends found in music waveforms where there are constantly varying amplitudes and frequencies of sound and where more importantly even if the signal may have an instantaneous power dissipation at a high value this is usually followed by a period of much lower values which essentially allows the speaker time to cool off Because of this an amplifier rated at a high level can oftentimes be used to drive a speaker with a much lower power rating safely For example if we have a speaker rated at 100 Watts this speaker could not handle 100 Watts worth of a continuous signal such as a monotone sinusoid into the system However an amplifier rated at 400 Watts would be required to carry out the amplification of a music signal that had instantaneous power peaks of up to 400 Watts A given music signal with an average power of 100
37. e wiper position is commanded to position itself to allow for increasing amounts of attenuation Also incorporated into this same function is a calculation of how drastic of a change in position a wiper must undergo in order to move from its current location to the newly calculated location and based on this calculation send the wiper through an appropriate amount of intermediate steps in order to more smoothly transition from one attenuation level to another For example if we had a wiper in a position corresponding to bracket 0 for no attenuation and we wanted to move this wiper to the position corresponding to bracket 12 to reduce the signal to about 1 5 of its current value we would command the wiper to move through two other intermediate positions on its way from the bracket 0 position to the bracket 12 position 4 4 Practical Adjustments One of the lessons learned from this project is that no matter how much thought analysis and calculation you put into a design once you test that design there is a great likelihood that you will encounter some issues that never occurred to you from a theoretical standpoint This was the case with the attenuation algorithms that had been 20 coded in the software portion of our system During the final testing stages of our device our code worked perfectly but that fact ended up being our only problem During testing when the amplifier was causing the feedback signal to fall in the upper ranges
38. ecibels per discrete position of the wiper varying from 0 dB with the wiper connected to the high terminal of the potentiometer all the way down to 62 dB where the wiper has very little resistance between it and the low terminal of the potentiometer The resistor also features a mute function that guarantees a resulting gain of less than 90 dB For our purposes however we must take into account that the stated attenuation of each wiper position is for use of the digital potentiometer alone in developing that amount of attenuation In our case the digital potentiometer is merely one component albeit the most important one in our analog attenuation circuitry see chapter 5 for details This translates into the fact that not only are the actual changes in attenuation we get for each wiper step less than 2 dB but also that even at the potentiometer s 0 dB level we will still have a series of resistors present that will function as attenuators to the incoming signal and thus the actual attenuation we 36 experience will be 1 22 dB see table 2 for discrepancies between wiper positions and actual circuit gain Once we had defined what gain levels were achievable by our circuit we decided that a Microsoft Excel spreadsheet would be most useful in developing a list of the range of input values and then determining the best level of attenuation amongst those values Since we are basing all of our calculations around the overloadVoltage safety t
39. gnal from amplifier output after steppec down A D conversion Form groups of 100 values use peak or average value adcStatistic User defined inputs impedance power of speakers Calculate maximum signal level speakers could handle safely overload Voltage Value of adcStatistic overloacVoltage Assign bracket number 0 26 Compare lt previous gt previous previous bracket equal to previous y Command for Repeat command for Command viper for 2 dB less current attenuation 2 dB more attenaution level attenaution Figure 8 Revised software flowchart means of comparison The function they now serve is to tell how far away the new ADC voltage level has to be from the old ADC level in order to command the 2 dB change In this function then the bracket definitions are still serving a useful purpose so that the wiper s position will not be changed every single cycle of the function Also built into the attenuation bracket calculations is the characteristic of the natural dB potentiometer s wiper so that signals requiring only a low level of attenuation will be more sensitive to voltage changes 23 whereas it will take more drastic changes in ADC voltage level for changes to occur at higher levels of attenuation 4 5 The Analog Attenuation The digital potentiometer changes resistance and is input into the non inverting
40. han was previously set so our system commands that the wiper position be moved 2dB from its previous setting After making only this slight 2dB change in the wiper position within our attenuation function the code loops back to again take a new ADC reading and calculate a new bracket number where it can be determined if the wiper should be moved for more or less attenuation Also it should be noted that if the signal has not changed levels enough to associate it with a different bracket number then the wiper is simply commanded to stay at its current position By employing this new algorithm the safety threshold is reached a bit more slowly but once it is reached it 22 stays within a reasonable limit of the threshold per our lax interpretation of the threshold explained in section 4 3 which is the main point of the speaker protection function of our device See figure 8 for the new software flowchart Under this new algorithm the initial purposes for the bracket definitions are nearly lost Originally these brackets were meant to calculate out where a wiper should be based on the signal level presented to the ADC sampler for that signal to lie within the threshold value However since we are not sending the wiper automatically to this pre calculated level but instead changing the attenuation only step by step and checking each time to see if the level has yet reached safe values these levels are only used as a Si
41. he 2008 04 speaker html necessary sound waves The mechanical motion of this cone 15 directly related to the oscillating voltage level entering the speaker This means that the higher the peak voltage of the incoming signal the greater the spatial displacement the speaker cone will have to undergo in order to achieve a corresponding sound wave a higher peak in voltage signal and mechanical oscillatory motion of the cone results in a louder sound at the output of the speaker However as with any mechanical component the range of motion of this cone is only designed to oscillate with amplitudes corresponding the power rating and impedance of the speaker Any signal that surpasses these limits puts uncharacteristic strain on the cone and the suspension attaching the cone to the speaker mainframe due to the extreme accelerations stretching and prolonged over vibration etc that it experiences under such conditions To prevent this type of damage we need to ensure a safe range of motion for the speaker cone This however does not necessarily entail always limiting any sharp intermittent spike ill encountered in the audio signal s amplitude If the duration of the spike is appropriately short and the audio signal in the surrounding time window is well below dangerous operating amplitudes it is very likely the speaker cone would be able to withstand such a spike Most of the time one such over excursion of the speaker cone will not cause the da
42. he two major mistakes we initially found were that first we did not have the digital potentiometer wired up to power and ground correctly and secondly that our sample code was much too complex and outdated 31 From this realization we discovered that the sample SPI code for master mode operation on the microprocessor s data sheet was all the code you needed in order to carry out SPI communication With our new set of code we were seeing some very perplexing results For example on a couple occasions we would observe the voltage on the wiper to be showing appropriate values most of the time but would skip many values or oftentimes jump between the power or ground voltage levels in the middle of a voltage progression that was supposed to be in between those two values We attempted to find our problem by inserting some debugging code that blinked an LED light emitting diode at crucial points in our code to ensure it was running properly but sometimes even this method was no use when the LED seemed to show that the system was running through each line of our code properly but the wiper position was not changing at all Finally we made our first step towards realization of our problem when we went through a simulation of our code line by line on the AVR Studio 4 program the same program we were using to write compile and program our code into our device with checking that the register values that we were setting in our SPI initialization
43. hreshold value our list of inputs is given in terms of multiples of this value from 0 meaning an input voltage that is zero times the overloadVoltage value and 25 meaning an input voltage that is twenty five times the overloadVoltage value We must make a side note here that since the speaker thresholds in general should be located a reasonable amount higher than the levels of normal operation a signal level that is twenty five times higher than that threshold will likely not ever be encountered in actual operation However since the range of the potentiometer and associated gain function is capable of attenuating even a signal of this abnormally large magnitude back down to within safe levels we decided we would program the device to be able to handle such levels This way we can be sure to account for the case of a speaker with a very low power rating and resistance that could easily be overblown and could potentially though not likely experience levels that are in fact twenty five times that of its safety threshold It should also be noted that in fact the attenuation our device offers could potentially limit input levels even above this 25 overloadVoltage point For example 62 dB is the lowest position offered by the wiper within its normal range of operation and even below that the device features a mute function that guarantees attenuation of 90 dB or below Converted to a gain in units of Volts Volts 62 dB translates into 0 006345
44. in spacing of 0 635 mm SPI serial peripheral interface a synchronous serial data link standard of communication involving a master talking to its slave over four wires called master in slave out MISO master out slave in MOSI clock CLK and slave select SS SS slave select acts as a chip select for SPI communication slave enabling a dedicated pin on port B of the microprocessor active low 1 PROJECT FEATURES AND OBJECTIVES The main distinguishing feature of our device compared to similar devices currently on the market is our capability to accommodate a wide variety of speakers allowing the consumer to use any typical speaker sold commercially today We accomplish this by inserting user interfaces whereby the user can specify the impedance and power specifications of the particular speaker regardless of the speaker s make or model The current industry standards for speaker impedance values are set at discrete values of 2 4 or 8 Ohms and the power consumption levels of a typical speaker ranges between 50 and 1000 Watts The user simply selects the levels of these parameters of their speakers and our device automatically processes these inputs and recalibrates the attenuation scheme of the limiting algorithms in order to allow for protection for a speaker with these specified parameters Another benefit of our device is that it Common Design has a limiter stage employs feedback contr
45. io signals in our brains is calibrated in terms of a logarithmic or decibel dB scale Therefore if we were to use a linear based digital potentiometer and were to start attenuating our signal in linear correspondence with the increasing input signal it would appear at first that there was a huge decrease in the sound being produced by the speaker and then as further attenuation occurred as the signal got louder and louder there would seem to be almost no attenuation occurring at all in the signal We were able to circumvent this problem fortunately by finding digital potentiometers designed specifically to deal with audio signal attenuation with wiper positions at taps positioned along the resistor to allow resistance changes to occur on a decibel scale instead of on a linear scale Our initial approach was to order a wide variety of audio tapered digital potentiometers and to investigate which one we felt would be best suited for use in our project We ordered the following devices from the Maxim Dallas Semiconductor company DS1801 DS1802 DS1808 DS1881 DS1882 MAX5411 and MAX5486 and researched the features of each device Our initial decision was to use the DS1881 model that provided 63 wiper steps of 1 dB per step a power supply range within the 5 V limit of the microprocessor we were using zero crossing detection to eliminate the clicking and popping sounds that tend to accompany wiper position changes in audio chains and an option to st
46. ital Conversion ii ii ii viii viii iv 11 12 15 15 16 4 3 Inside the Microcontroller Attenuation Determination Implementation 4 4 Practical Readjustments 4 5 The Analog Attenuation CHAPTER 5 HARDWARE 5 1 The Digital PCB 5 2 The Analog PCB CHAPTER 6 SOFTWARE 6 1 SPI Communication 6 2 Trial and Error Lessons Learned 6 3 Attenuation Bracket Considerations CHAPTER 7 PROJECT LOGISTICS 7 1 User Manual 7 2 Separation of Work 7 3 Bill of Materials 7 4 Gantt Chart BIBLIOGRAPHY APPENDIX ADDITIONAL PROJECT MATERIAL and 18 20 24 25 25 27 29 29 31 35 45 45 46 47 48 49 50 INTRODUCTION The ever expanding realm of technology and electronics is conquering every field imaginable in this day and the field of music and sound reproduction in general is no exception Unfortunately with the wide range of functions required of electronics in order to manipulate the sound in the desired fashion there has developed a need to incorporate many different individual devices into a single signal chain from input to output In most circumstances the last component accounting for the audible reproduction of the manipulated sound is the speaker As with all electronic devices however there are limits on the operating levels of the speaker that determine its safe operating levels These limits cause a problem when we consider that the other components that have manipu
47. ive counterpart for its many benefits In general this limiting technique can guarantee safer more reliable limiting since it is receiving its feedback directly from the signal line that is being directly fed into the speaker itself see figure 4 in section 2 2 Because of this that signal information can be used directly without having to back calculate the gain of the amplifier hence the limiter does not require recalibration upon each readjustment of amplifier gain Also in this set up knowledge of the parameters regarding the amplifier such as impedance and power rating does not have to be as exact as it does for the predictive set up Since we are presented with the output of the amplifier the main parameters we are concerned with in this situation becomes the parameters of the speakers so that slight variations or discrepancies from ideal amplifier parameters are not as crucial to the functionality of the limiting device in this case 2 2 Specific Examples United States Patent 4173740 is a device that uses feedback control in an attempt to protect speakers from being overdriven but instead of actually attenuating the signal as it approaches dangerous levels this device simply cuts off the supply voltage to a power amplifier circuit in disconnecting a loudspeaker from the output terminal of the power amplifier circuit when an overvoltage is developed at the output of the power amplifier circuit Our device has a significant
48. lated the audio signal up until this point are independent of the speaker and thus unaware of these safety requirements The goal of our device is not only to incorporate a limiting device into a signal chain that would account for these safety limitations of the speaker in the presence of an amplifier but to do so in a manner that would maintain the continuity of the volume s dynamics in an aesthetically pleasing non distracting manner A Note About Speaker Damage With the purpose of our device being to protect speakers we must first understand what damages speakers in the first place When we think of damage to a physical device we first consider the mechanically active parts of the device and how the behavior of these parts could be put in danger For the speaker s case these mechanical parts are located at the point of the transduction of electrical energy in the incoming voltage signal to mechanical energy Here the cone of the speaker and its suspension which is usually attached to the coil containing the incoming electrical signal some designs allow for the permanent magnet 7 cone component to be attached to the cone instead vibrates magnet vibrate magnet 4 and as shown in figure 1 oscillate back and forth attached 7 makes to cone sound in its production of the desired air Figure 1 Speaker components Source http p hardware blogspot com pressure variations in order to create t
49. lue but this threshold value was determined by a relationship of resistance values used in the circuit In order for such values to be extracted from the user s input and then a determination to be made about how the attenuating algorithms operate according to these values we would require the use of a smart devise to take in such inputs and carry out the necessary calculations The only real logical solution for this is to incorporate the use of a microprocessor for there is no functionality in the Analog Engine device that would allow for such calculations to be made An option would then be to use a microprocessor along with the circuitry recommended for the Signal Limiter for Power Amplifiers but then the problem would become both an unnecessary increase in complexity between the microprocessor to Signal Limiter circuitry and the inflexibility to completely control all features of the Signal Limiter circuit due to their dependence on set resistance values that would be hard wired into the circuitry The Signal Limiter circuitry does allow some flexibility in setting the limiting threshold and compression ratio by incorporating 3 different variable resistors but even so many of its parameters 13 are set using a combination of these variable potentiometers and set resistors Even if we could set up our system to use digital potentiometers in place of these variable resistors so that we could control the threshold and limiting parame
50. mage but prolonged over excursions leading to the mechanical fatigue will eventually result in failure of the mechanical parts including ripping or tearing of the cone or its suspension Another thing to note is the different types of speakers that can be used and which types are most prone to different types of speaker damage In the case of a mechanical failure woofer and tweeters are the usual culprits Woofers handle the low frequency spectrum of the input and are oftentimes overdriven at loud parties where a loud bass sound is desired Tweeters are oftentimes overdriven mechanically when they are associated with an inadequate crossover system since these speakers are very small fagile and easily damaged In our project we will have to find a delicate balance between protecting the speaker from high amplitude spikes that are mechanically hazardous to the system while still leaving the sound levels unaffected if it is determined that such a spike in sound will not likely be detrimental to the speaker The other type of damage a speaker could encounter is heat damage If the speaker is producing a high volume of sound for an extended amount of time the amount of power being dissipated rises to very high levels which causes the speaker to heat up to extremely high temperatures and puts the speaker at risk for damage by overheating At these extremely high temperatures either the voice coil adhesive can soften and cause the voice coil to come apart
51. ming the same signal attenuation as our device but very few would be smart enough to automatically and continuously adjust this attenuation towards the goal of protecting a speaker of a specific speaker Within the realm of limiters designed to protect speakers there are in general two different types predictive limiters and negative feedback limiters Predictive limiters are convenient in that they are more portable in general than feedback limiters and aren t necessarily used as dedicated speaker protectors all the time However their drawbacks far outweigh this slight benefit As seen in figure 3 below the predictive limiters merely predict the output of the amplifiers that the speakers will be exposed to and do not have any connection to the actual voltage the speakers are exposed to As such in order for the limiter to adequately calculate the amplifier s output from its input the amplifier must be held at a constant value Once the amplifier s gain has changed the limiter must be L2 OUT RUE Figure 3 Predictive limiter Source http www rane com note127 html recalibrated else the speaker is in danger of being over driven Negative feedback limiting is the concept our design is based off of though we were not aware of the existence of this technology upon the conception of the idea for our project This type of limiting in regards to speaker protection has become more popular than its predict
52. n taking into account other details regarding speaker operations that our design does not consider However we must first take into consideration that this design is merely a technology that has been directly incorporated into a company s speakers they themselves manufacture Since the design is literally embedded within the speaker cabinet itself there leaves no room for this device to be used any speaker amplifier combination the user wishes and therefore does not offer the same freedom of speaker choice to the user as our device does TruPower is literally designed to calculate the true power that is being delivered to the speaker For example though in our device we assume that the impedance of the speaker as specified by the user as an input to our system remains a constant this impedance actually varies slightly both over the frequency range of the signal and over the operation time of the speaker as the speaker coils warm up and experience a slight increase in impedance The impedance increase in the speaker coils causes a slight drop in the overall dynamic range of the speaker system over time and this TruPower technology seeks to correct this problem On the opposite end of the spectrum the frequency dependency of the speaker impedance can cause the speaker impedance to unknowingly decrease at certain frequencies and thus open the speaker up to potential damage occurring at signal levels that under normal impedance wo
53. nge again so the attenuation would be unset This behavior both made the sound output from our system to be distracting unnatural and aesthetically unpleasing and caused the output signal 21 going to our speaker to oftentimes reach very high levels that would very likely allow for damage to the speaker to occur In order to change this we only had to modify a few lines of our code but this drastically changed the behavior of the system in practice We decided that instead of calculating a destination attenuation bracket the potentiometer should reach based on its current level which would lead often to the toggling 0 dB bracket once the attenuation did its job of bringing the signal back to safe levels we should have the system only change its attenuation bracket one at a time To do this we used the framework of the existing attenuation bracket definitions to assign the current level to a bracket number the bracket numbers 0 23 correspond to wiper settings 0 46 dB with a multiplication of 2 factor to convert between the two and then compared this bracket number with the last bracket the wiper was assigned to If the new bracket was lower than the previous one that meant that less attenuation was needed than was previously set so our system commands that the wiper position be moved 2dB from its previous setting If on the other hand the new bracket was higher than the previous one that meant that less attenuation was needed t
54. of attenuation we would see the code step through the steps needed to reach the destination attenuation bracket that had been specified for it in our code and then once it got there as a verification that the attenuation levels had been indeed calculated correctly we saw the ADC readings to go from the high values of the amplified signal to low values that were well within the overloadVoltage safety threshold value The problem then was that the next command in our code was to run the function again to calculate an appropriate attenuation bracket for the feedback signal Since this function had already run once and had done its job of bringing the signal into the range where it would not need further attenuation this second time around the code put the attenuated signal into bracket 0 where no attenuation was required and the potentiometer s wiper would be moved accordingly We must realize here though that by lowering the level of our small signal audio input when we first commanded the attenuation we did not change the high gain setting present on the amplifier so that once we bring this attenuation back to 0 dB the high amplifier gain is still present Therefore essentially the current code was causing the attenuation for a high feedback signal to toggle between a given attenuation level and 0 dB of attenuation once through the attenuation function and the attenuation would be set a second time through and the level would be in the safe ra
55. ol where the feedback Into the amp signal is taken directly from the output of the amplifier which is also the input signal fed Limiter directly into the speakers see figure 2 This provides a greater safety guarantee over another New limiter would look at the amp feedback common design involving predictive control to automatically determine when to limit power since we have direct access to the signal that is to be seen by the speaker and are not merely trying Figure 2 Feedback control to predict this signal The final benefit of our device design is its attenuation scheme Unlike other products intended to protect speakers from unsafe levels of input our product does not simply disconnect our circuit once dangerous levels are experienced thereby abruptly muting the signal Instead our system employs gradual attenuation algorithms that are continuously yet subtly limiting the system input By beginning to slightly attenuate the signal in the upper range of the safe levels of speaker operation once the threshold of the speaker s capability is reached the limiting function of our device has already begun and therefore moves seamlessly over this threshold 2 ANALYSIS OF COMPETITIVE PRODUCTS 2 1 General Approaches Essentially our device is simply a limiter programmed to accomplishing a specific goal There are many different kinds of limiters that would be physically capable of perfor
56. or The full wave rectification circuitry was based on figure X from the book Introduction to Operational Amplifier Theory and Applications Figure 12 ADC step down vs high voltage cosine top and high voltage Huelsman 1975 The top of figure 12 shows music bottom how a simple cosine signal swinging from 50 voltage to 50 volts is rectified and then smoothed out to be about 4 volts The bottom of figure 12 shows how the rectified signal changes with music and jump up and down depending on the music s amplitude The figure below takes one through the step of how the cosine signal voltages look at each stage of the analog board Figure 13 Full wave rectification left diode in series middle low pass filter right 6 SOFTWARE 6 1 SPI Communication The use of Serial Peripheral Interface SPI communication was required for our microprocessor to be able to control the digital potentiometer model we had chosen the MAX5411 As mentioned in chapter 3 the prime reason for using this SPI format of communication was that the Atmega 32 chip contains pins that Figure 14 SPI communication wiring Source http dev emcelettronica com category tags spi protocol are connected to internal hardware designed to carry out such communication Also to correspond to such hardware there are registers featured on the microprocessor that enable the SPI hardware to be configured and utili
57. or the speaker This is due to the lax definition on how the actual speaker rating is calculated as mentioned in the Introduction where the threshold is assigned to an average level over time and allows for slight up and down fluctuations in the music around this level The overloadVoltage equation is overloadVoltage wattage rating x impedance 08 1023 5 This comes from the voltage fact that Power so then Power x resistance voltage The step down resistance circuitry attenuates a signal to 08 of its peaks so voltage Power x resistance x 08 Voltage is the overloadVoltage power is the power a speaker can take and resistance is the speaker s impedance which is commonly 2 4 or 8 ohms Below is a chart of common speaker impedances and power ratings that the user can set with their corresponding overloadVoltage values 18 overloadVoltage overloadVoltage Speaker Impedance Power Table 1 overloadVoltage Calculations given all possible user input combinations Rating decimal 0 1023 volts Ohms Watts 231 4784759 1 13137085 2 100 327 36 1 6 4 100 462 9569518 2 2627417 8 100 327 36 1 6 2 200 462 9569518 2 2627417 4 200 654 72 3 2 8 200 400 9324811 1 959591794 2 300 567 0041524 2 771281292 4 300 801 8649622 3 919183588 8 300 462 9569518 2 2627417 2 400 654 72 3 2 4 400 925 9139036 4 5254834 8 400 517 6016074 2 529822128 2 500 731 9992131 3 577708764 4 500 1023 5 8 500
58. ore wiper position information in an EEPROM upon power off However we were forced to abandon using this potentiometer once we discovered that the difficulty of using the Atmega 32 microprocessor for I2C Inter integrated Circuit 10 communication far surpassed that of using the SPI serial peripheral interface communication required for the MAX5411 potentiometer The functionality for carrying out SPI communication was built right into the hardware configurations of the Atmega 32 microprocessor as accessed by the dedicated pins at port B The Atmega processor already had registers set up in order to allow the parameters of the SPI communication to be adjusted to suit the needs of the given SPI communication application and some simple sample code was already presented in the data sheet to initiate SPI communication and to begin sending information via SPI communication The only major drawback to using this new potentiometer model was the fact that this model only allowed 32 wiper positions of 2 dB per step which would result in less resolution in our attenuation accuracy and a slight loss of smoothness as the wiper transitions took place It was determined that the ease in communication format and microprocessor coding was well worth this slight sacrifice in attenuation quality 3 4 Feedback Signal Incorporation Our next task was to find a way to tap into the signal coming out of the amplifier and feed back a version of this signal into
59. orementioned attenuation bracket calculations 44 CHAPTER 7 PROJECT LOGISTICS 7 1 User Manual Operation of our device is simple and nearly self explanatory The left most jack receives the power amplifier s high voltage signal This signal is captured using the red cable that came with our device The cable splits the output of the power amplifier One end goes to the speaker and the other goes to this left most high signal jack The middle jack receives the small signal output of the your instrument The right most jack is the output into your power amplifier Once the device has been properly connected the user first powers on their amplifier and speaker with the amplifier gain knob at a modestly low setting since the user has not entered the correct parameters for their speaker at this point and proper attenuation to ensure speaker safety is therefore not yet operational and power on our device s power supply by flipping the black power switch at the rear of the device Next the user dials in the speaker impedance and power rating as given by the speaker manufacturer s specifications The impedance is to be dialed in on the knob on the left and has settings of 2 4 or 8 Ohms the values increase with a clockwise turn of the knob and the current value set can be seen on the LCD screen display as it is changed The power rating is to be dialed in using the knob on the right and has settings of 50 Watts to 800 Watts with the current po
60. r attenuation and a significant drop in the dynamic range experienced by the listener at the speaker output and lead to many abrupt and dramatic attenuation leaps between load and soft as any intermittent spike was encountered User defined inputs Signal from amplifier impedance power output after of speakers stepped down Calculate maximum signal level speakers A D could handle safely conversion overload Voltage Form groups of 100 values use peak or average value adcStatistic Value of adcStatistic overload Voltage 0 1 00 1 0011 10 gt 25 Command wiper for Command wiper for Command wiper for postion 0 048 position 1 4 dB T position 24 input is attenuation attenuation muted Figure 7 Software architecture 17 4 3 Inside the Microcontroller Attenuation Determination and Implementation Once we have the peak or average information from the ADC results the user s inputs that were read into the system upon the device s power up come into play These user inputs of speaker resistance and power rating are used to calculate the software variable called the overloadVoltage which is the threshold determining safe operation of the user s speaker As a short side note we are using the term threshold more loosely for our actual attenuation implementation where we allow the signal to float close to and even slightly above this threshold level and consider this safe operation f
61. rac f brac f brac f brac f brac f brac Ct ct ct ct ct ct ct ct ct ct ct ct umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber umber 0 return 0b00000000 1 2 3 4 5 6 Wo gy wg Ww m m m Ww Ww m m m W Ww Ww Wo gw Ww Ww W Ww m W W m W 11 W W mW 1 W W I NON PB PD P pP pP pH pP pP pP pP to 0 i QN IB OO 22 23 24 25 26 ret re ret re Let ret re re re Le re re re re re re re re Le re re re urn curn urn curn urn urn curn urn curn Cur Cur Cur Cur cur cur cur cur Cur Cur Cur Cur cur Cur cur Cur Cur 00500000001 00500000010 00500000011 000000100 0500000101 0500000110 0500000111 000001000 0500001001 n 0500001010 n 0500001011 n 0500001100 n 0500001101 n 0500001110 n 0500001111 n 0500010000 n 0500010001 n 0500010010 n 0500010011 n 0500010100 n 0500010101 n 0500010110 n 0500010111 n 0600011001 n 0000011111 n 0010011111 58
62. re safe in ignoring this pin If this pin is configured as an output pin while the device is in master mode for it cannot be configured as an output in slave mode or it would lose its ability to be selected as such a slave it becomes a general purpose output pin and the SPI communication system is not affected by this pin s value However we were not observant of the fact that in the code from the Atmega 32 datasheet we were using they left this SS pin set as an input expecting the user to desire the microcontroller s option to act as a SPI slave note that inputs are configured to a port value of 0 so that through the lack of mentioning the input output functionality explicitly in the code the SS pin s function as an input was implied The simple fact of the SS pin being configured as an input would not have caused any problems however had this pin not been an active low pin Unfortunately an input pin with no signal to drive it high is sitting at a low value Therefore under the configuration we were using our microcontroller was being informed that it was required to act as a slave and was not actively trying to send information but to receive it Our problem was fixed once we combined these two bits of insight into the functionality of the chip and slave select pins of our two devices we first set the SS pin on the microcontroller to be an output pin and then connected a separate general purpose output pin having no ties to the SPI sy
63. ry out such an operation converting our small 0 5 5 V signal from the feedback loop into a set of discrete integer numbers valued between 0 and 1023 corresponding to a 10 bit digital result Here we have the difficulty of an analog signal at our port A pin that is not only is not sitting at a constant voltage level due to the changes in the amplifier s gain but also fluctuates in a half rectified fashion between 0 Volts and its peak voltage even while the amplifier gain remains constant due to the sinusoidal characteristic of musical frequencies To be able to accurately interpret the signal level we first acquire sets of 100 samples of ADC readings and then perform statistical analyses namely calculating the average value and the maximum value on each set of data in order to be able to make a 16 better decision on how much attenuation is needed based on each sample period The goal here is to in general set up our attenuation to cater to the peaks experienced in order to allow for maximum protection of the speakers but to ignore this peak value and instead use the average value in the case that a very short lived spike in voltage is experienced that is well above the rest of the values of that sample see code in the Appendix for details regarding this selection criteria If we were to only observe the peaks as a rule we would always be responding to every little voltage peak in the system which would lead to both a general sense of ove
64. ssion of this message is determined by the SCLK signal and the relationship between the phase and polarity relationship between the clock signal and the message signal are set up using registers in the microcontroller this phase and polarity must match the phase and polarity required by the SPI slave you are communicating to or else the signal will not be interpreted correctly The MISO line is meant to enable the slave to talk back to the master device but is not used in our project The potentiometer would be reading back to the microprocessor the current position of its wiper via the MISO line but in our case we are merely concerned about telling the digital potentiometer where to position its wiper and based on that command from the master as long as the slave device interprets the commands correctly we should already have knowledge of the current wiper position at any point in time The SS line is an active low line where the master prepares the slave device to receive a transmission This line is most important when more than one slave is present in a system where the master can define many SS output pins to be able to select each slave device to talk to one at a time The master s SS line is usually connected to the slave device s CS chip select active low so that only the chip that is selected by the 30 master will be able to actively interpret the master s commands This same connection is made between our microprocessor an
65. stem itself to act as an output signal to the potentiometer s CS pin that would carryout the reset like function required for communication with our slave device 34 6 3 Attenuation Bracket Considerations The goal of our attenuation is to gradually limit our input signal as it approaches and exceeds levels that would be considered unsafe for the speaker in consideration Since all of our considerations regarding what signal levels to attenuate are made based on this threshold of safe limits for our speaker and the safe limit for our speaker s operation is based on the impedance and power rating parameters as defined by the user inputs into the system we first must establish this safety threshold which is referred to as the variable overloadVoltage in our code As briefly mentioned in section 4 2 we use the relationship power voltage resistance along with our user inputs of power and impedance to obtain a maximum voltage we want to have entering our speakers from the output of the amplifier After this formula is used to find the correct voltage value we must remember that the voltage information we are comparing this threshold value to is from the result of analog to digital conversions of the stepped down feedback signal and being as such will be an integer value between 0 and 1023 due to the 10 bits of resolution available to the ADC hardware Therefore to get this within an equivalent range we must multiply this high le
66. ters with our microprocessor this would seem to be at least three times as complicated as our original design involving communication with only one digital potentiometer and without the worry of having to hard wire some parameters of the limiting into our circuit through the use of set resistor values Therefore we decided that our original design would be the simplest and most efficient way to accomplish our goal of speaker protection 14 4 PROJECT ARCHITECTURE 4 1 From Input to Amplifier A very high level system diagram can be seen in figure 6 The very first function our device serves upon system power up is the Small Signal Gain Control input of the user s User Inputs Impedance speaker parameters This Power Rating Signal Processor is accomplished through the use of toggle he USE Can Figure 6 High level system diagram select speaker impedances 2 4 or 8 ohms and speaker power ratings 100 200 1000 watts With this information the microprocessor calculates what the threshold voltage is to start attenuation the details of such calculation will be explained in section 4 3 The audio signal itself originates from some electronic instrument microphone etc as a very low voltage signal representing the pressure variations in the air produced by the music desired to be amplified out to the general public This signal travels through the power amplifier
67. th only the addition of a couple smaller components such as capacitors resistors and diodes needed Another benefit besides convenience that this new solution had to offer was its use of an analog voltage Figure 5 THAT 4301 Analog Engine Source http www ka electronics com that4301 that4301 htm controlled amplifier which would allow the attenuation to occur through a continuous set of values This would provide smoother sounding volume transitions over the discrete wiper position changes that would be necessary in our use of a digital potentiometer Also the device not only was designed to operate by taking in the feedback signal from the amplifier s 12 output similar to our original design but it additionally contained an RMS detector to enable that feedback signal s strength to be more readily detectable by the device This would eliminate the need of our originally intended step down circuitry However there was one crucial drawback to this design that made it implausible to use it had no real user defined flexibility One of the main distinguishing features of our device is its capability to be used with any speaker the user wishes where the user only has to select two simple parameters of their speakers and our device will automatically recalibrate itself to fit their speaker s needs This new design was meant to operate around a threshold value and automatically attenuate the audio signal past that va
68. tistic instead Input vs Output from Attenuation Circuitry Low Range 1 Now that we have justified our use of imposing a 0 9 overloadVoltage limit on our output signals 0 0 2 0 4 0 6 0 8 1 1 2 14 1 6 1 8 2 we then are left with simply Vin overloadVoltage calculating the definitions Figure 16 Input vs output trend of attenuation circuit low range only between where one attenuation level potentiometer wiper position stops and another starts based on when the output 41 signal approaches that limit while using the first attenuation bracket Such transitions can be seen clearly in Figure 16 As the signal approaches the 0 9 overloadVoltage output line the signal shows a sharp drop signifying a change in the potentiometer s wiper position Note however that even though these drops appear to be abrupt and of a significant size from figure 17 it can be seen that these drops between attenuation levels are well less than 1 dB in magnitude for reasonable levels of signal input magnitude This graph does alert us to the fact that as the input levels and thus attenuation levels increase this drop becomes larger and thus more in danger of being heard As mentioned earlier the just noticeable difference JND of a change in volume that a human ear can detect is 1 dB this was how the system of decibels was created to define a unit that corresponded to the capacities of human hearing
69. turn 0 0b00000001 0b00000010 0b00000011 0b00000100 0b00000101 0b00000110 0b00000111 0b00001000 0b00001001 0500001010 0500001011 0500001100 0500001101 0500001110 0000001111 0500010000 re re re re re re re re re re re re re re re re curn curn curn curn curn curn curn curn curn curn curn curn curn curn curn 57 4 TRA tay 14 1 55 1657 15 15 else else else else else else else else H H H H H H H H H H PotCommand PotCommand PotCommand PotCommand f PotCommand PotCommand PotCommand PotCommand PotCommand PotCommand 0b00010001 0500010010 0000010011 0500010100 0500010101 0000010110 0000010111 0000011001 0000011111 0010011111 return 177 return 18 return 19 return 20 return 21 return 22 return 23 return 24 return 25 return 26 int potCommandValue int bracketNumber if bracketNumber else else else else else else else else else else else else else else else else else else else else else else else else else else i H H H H H H H H H H H H H H H H f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f brac f b
70. uld have passed as harmless Though this frequency vs impedance relationship is likely hard to determine directly this technology seeks to prevent any speaker overload due to this phenomenon by automatically attenuating the signal by a couple extra dB after a long period of operation time to prevent overheating of the speaker 3 CONCEPT TECHNOLOGY SELECTION 3 1 The Attenuator and Analog vs Digital Considerations We knew from the start that the three major functions that needed to be performed by our device were the collection of user input and feedback loop data the processing of that data to form the appropriate attenuation algorithms and finally the actual performance of the attenuation on our input signal We knew that in order to interpret and process information regarding analog data we would likely find it best to perform our processing in digital form Our initial idea was to first perform analog to digital conversion from our feedback signal to determine the amount of attenuation needed and then to go back and attenuate a digitized form of the input signal by that amount before sending the analog version of the newly attenuated input signal through the amplifier Under this design we thought it would be best to use either a digital signal processor DSP or a microprocessor to carry out the attenuation of the signal in its digital form However the problem with using the microprocessor for this purpose was a fear
71. using a digital potentiometer to perform the attenuation would be best it was an obvious choice to use a microprocessor for this purpose Once this was decided the two main considerations for settling on a final device were the type of microcontroller to use and what size of a microcontroller was needed Our choice of an Atmel brand microcontroller over a PIC Peripheral Interface Controller controller was made on the basis of code familiarity Bethany was in charge of the software programming of the microprocessor so she chose the c programming coding language of the Atmel controller over the basic coding language of the PIC due to her experience with the related C programming language Regarding size considerations with our experience with the Atmega 32 microcontroller and our general estimation of the number of pins and the types on functions that would be needed analog to digital conversion communication with the digital potentiometer dedicated user inputs and potential LCD output pins we decided that this device would be sufficient to meet the needs of our project 3 3 Digital Potentiometer Model Selection To begin with we had very little knowledge regarding not only the types and different features of digital potentiometers but also regarding which of these features would be most suitable and beneficial to our device What we did know however was that we were to be attenuating an audio signal and that detecting changes in aud
72. vel voltage value by a factor of 1023 08 5 16 4 This factor comes from the step down circuitry Since we know that this threshold value will be different for every given set of speaker parameters we need a way of defining bracket levels where each bracket corresponds to a range of incoming data from the ADC that translates into a certain command for the potentiometer s wiper position and hence a certain amount of attenuation of the input audio signal Since this attenuation depends on the overloadVoltage value we define each bracket based upon the division of the statistic 35 whether it is the peak or the average value describing the set of incoming ADC values by the overloadVoltage value For example 1 0 would mean that either the peak or average value from the set of values that were sampled using the microprocessor s ADC function is exactly equal to the corresponding voltage threshold value that would likely blow the speakers and 25 0 would mean that the peak or average ADC value is twenty five times the value of the safety threshold for the speaker the point at which our device essentially begins to mute the incoming audio signal Next actual levels must be defined to both keep our signal at safe levels and to avoid any abrupt changes in attenuation so that the sound coming out of the speaker retains as much of its original volume continuity as possible The resistance positions of the potentiometer s wiper change by two d
73. wer rating also being displayed on the LCD screen Once these two values have been dialed into the system the amplifier s gain can be changed by the user at will and our system will protect the speaker from being damaged Also displayed on the LCD screen is the overloadVoltage value of the safety threshold corresponding to the user specified speaker ratings and the amount of attenuation currently affect on the device Once the use of the amplifier and speaker is no 45 longer needed cut off the power to the speaker and amplifier first before cutting off power to our device by flipping the black power switch again at the rear of the device to ensure that speaker safety is maintained for the entire duration of speaker operation 7 2 Separation of Work Bethany Processor breakout Atmel coding design uP code testing UI coding and testing Final board population Test debug Paul Processor breakout Design of PCB Hardware testing Power sensing testing Final board population Test debug Table 3 Division of Labor In general Bethany was in charge of the software of our device and Paul was in charge of the hardware of our device In addition to designing the attenuation scheme we 46 intended to use in our limiting algorithms writing the software code debugging the code and testing the code s functionality with various hardware components at different stages during the design process
74. y all the spikes within the human range of hearing We should also near in mind that when we talk about spikes we are implying that these high amplitude signals are very short lived and will likely not cause any damage to the speakers for if they are of a 40 longer and more threatening duration our system should be fully capable of detecting them and setting the attenuation accordingly As for a discrepancy between the statistic values we choose from our ADC sampling sets as briefly mentioned in section 4 2 and as shown in our code in the ADCAveragingandLCDFunctionsBethany h header file under most circumstances we base all of our attenuation off of the incoming voltage s peak value from the set of ADC samples we have taken However we must take into account that sometimes we may measure a very short lived and therefore essentially harmless spike to our system In that case we have decided that it would be more appropriate to use the average value of the incoming ADC sample data in order to get a better characterization of what the incoming signal is like without consideration of the spike Therefore our selection criteria becomes 1 calculate both the peak and the average of the incoming voltage signal 2 use the peak as the determining ADC statistic to characterize that data unless that peak is greater than 3 5 times the average value 3 if the peak is in fact greater than 3 5 times the average value use the average value as the sta
75. zed easily As shown in figure 14 SPI communication is carried out between two devices the devices can be daisy chained so that one master can communicate with multiple slaves be we only require two device communication in our project one being specified as the master device and the other as a slave device In our case we configure the internal SPI Control Register SPCR s bit corresponding to master mode MSTR to have our microprocessor act as the master who is then responsible for sending data to the digital potentiometer slave which acts as a slave by default since it lacks the capabilities to perform the duties of a master device such as generating its own clock signal SPI communication requires 4 different wires running between the two devices as shown in 29 figure 14 labeled SCLK serial clock MOSI master slave in MISO master in slave out and SS slave select The SCLK line is the clock that synchronizes the timing between the two devices This clock signal is generated by our microprocessor master and its frequency is determined by setting registers corresponding to certain division factors from the internally generated clock signal always present in the microprocessor The MOSI line is the most important connection in our SPI communication system This is where the transmitted message is sent as a series of ones and zeros from the microcontroller to the digital potentiometer The timing of the transmi

AUDIO SPEAKER PROTECTION FROM UNSAFE LEVELS

Contents

Download Pdf Manuals

Related Search

Related Contents