ANALYSIS OF MECHANICAL AND ELECTRICAL MODIFICATIONS THAT ALTER FREQUENCY RESPONSE

Transcription

1 UNIVERSITY OF WATERLOO Faculty of Engineering ANALYSIS OF MECHANICAL AND ELECTRICAL MODIFICATIONS THAT ALTER FREQUENCY RESPONSE Aastra Telecom 155 Snow Bl, Concord, ON Prepared by Guanhua Wen ID # b Electrical Engineering May 13, 2013

2 Guanhua Wen 455 Laurel Gate Dr., Waterloo, ON N2T 2S1 May 13, 2013 Manoj Sachdevi, Chair E&CE Department, University of Waterloo, Waterloo, ON N2L 3G1 Dear Manoj Sachdevi: Re: Submission of my work term report. This report, entitled Analysis of Mechanical and Electrical Modifications That Alter Frequency Response, was prepared as my 3b Work Report for the University of Waterloo. The purpose of this report is to analyze desk phone attributes that affect the frequency response of the hands free speaker. Aastra Telecom is a global company that provides enterprise communication solutions in a wide and extensive portfolio. These solutions include a wide variety of Analog and Digital Telephone, Internet Protocol Telephones and Video Conferencing platforms. I would like to thank Manjinder Mann for providing me with opportunities to develop new skills outside of my regular duties. I would like to thank Stanley Kwok for providing guidance on how to use Microtronix to test for frequency response. I would also like to thank Tahir Ahmed for providing guidance on how mechanical changes to the chassis can affect the frequency response. I hereby confirm that I have received no further help other than what is mentioned above in writing this report. I also confirm that this report has not been previously submitted for academic credit at this or any other academic institution. Yours sincerely, Guanhua Wen, Encl.

3 Contributions I worked in the product verification team, which consisted of 11 full time employees and 2 coops. The employees manually test to verify phone the software and hardware of Aastra s product line. These tests include basic calling/receiving, special features, stability, speaker/microphone frequency response, etc. The coop s main duties were to automate test cases that would have otherwise required manual testing. Over the course of the work term, I have written and automated over 100 test cases. I look at manual test cases and write code to automatically test and catch failures. Side projects include testing and modifying the frequency response of the speakers and microphones to meet TIA standards. From the experiences in coding my test cases, I acquired expertise in the programming languages TCL and XML. By testing the frequency response of the phones I was able to put my knowledge learned from school to use in the workplace. In order to alter the frequency response mechanically, I was able to modify the new prototype Voice Over IP desk phone via soldering and drilling. By writing this report I now have working knowledge of L A TEX, which proved to be very useful after the initial learning curve. iii

4 Summary The main purpose of the report is to explore possible mechanical and electrical modifications that will alter the frequency response in a favourable fashion. This report can then assist engineers who are looking to correct the frequency response on future prototypes. There are two sections to this report. The first part will describe the test methodology and calculations to be used for comparison. The second will explore the use of sound absorbing foam, changing the shape of the speaker grill and application of a LPF to see their effect on the frequency response. Each of the three methods will include a comparison table of the intensity of the before and after responses. A conclusion will be drawn to determine which methods can be used and which frequencies they alter, followed by a recommendation on which methods the senior engineers can use. The two major points documented in the mechanical, hardware and software characteristics that can mold the frequency response to meet IEEE standards. Mechanical characteristics are the size and shape of the speaker cavity. The major conclusions is both the modification of handset and the addiction of a Low Pass Filter has desirable results. Modifying the speaker grille of the handset to address sound diffraction by sealing the outer holes, the intensity increase 2.39x for frequencies from 2000Hz Hz. LPF soldered on the circuit board of the phone can decrease the intensity by 2.0x for frequencies from 2000Hz Hz. The major recommendations in this report is Aastra proceeding to modify the Handset to increase intensity in the high frequencies while adding LPF to decrease intensity at high frequencies. iv

5 Table of Contents Contributions iii Summary iv List of Figures vi List of Tables vii 1 Introduction Importance of Frequency Response Scope Outline Test Setup Requirements Methodology & Equipment Used Telephone Modifications Sound Absorbing Foam Speaker Grill Modification RC Low Pass Filter Conclusions Recommendations Glossary References Appendix A Frequency Response Data Points v

6 List of Figures 1 Wide Band Hands Free Speaker Standard Anechoic Chamber at Aastra Telecom Frequency Response - Speaker Cavity Without Foam Frequency Response - Speaker Cavity With Foam Frequency Response - Handset Without Modification Frequency Response - Handset With Modification No Foam No Low Pass Filter MicWithNoCap MicWith22nFCap vi

7 List of Tables 1 Sound Absorbing Foam Response - Comparison Speaker Grill Diffraction Response Comparison RC Low Pass Filter Response - Comparison Summary Comparison Sound Absorbing Foam Response - Before Sound Absorbing Foam Response - After Speaker Grill Diffraction Response - Before Speaker Grill Diffraction Response - After RC Low Pass Filter Response - Before RC Low Pass Filter Response - After vii

8 1 Introduction Aastra Telecom designs both hardware and software components of their VOIP desk phone line up. The speakers housed internally in the chassis is purchased from an outside vendor. A prototype of next generation desk phones is designed to have a thinner chassis but also deliver superior sound quality similar to its previous generations. Testing for the frequency response is a useful method to verify the accuracy of the sound reproduced. During the design stages, Aastra ensures certain standards are met by its products. 1.1 Importance of Frequency Response The functionality of a telephone is defined by Websters dictionary as: An instrument for reproducing sounds at a distance; specifically : one in which sound is converted into electrical impulses for transmission [1]. The reproduction of sound has to be accurate for the users to understand. The range of frequencies used for speech is dependent on the user s physical traits. Human conversations typically fluctuate between 300Hz to 3kHz. The human ear however can detect sound in the frequency range of 20Hz - 20KHz [2]. Depending on the design, the intensities of frequencies can vary. This creates undesirable spikes and dips in volume at different parts of a conversation. Accurate sound reproduction is critical. 1.2 Scope This report will analyze possible mechanical and electrical modifications to the prototype that alter frequency response. The senior engineers can then use the results to pick and choose any methods explored to be implemented in the next generation Aastra desk phone. This will contribute to passing with Telecommunications Industry Association standards thus delivering a better product experience to the consumer. The methods explored are the use of sound absorbing foams, reduction of speaker grill diffraction, and the use of a Resistor-Capacitor (RC) Low Pass Filter (LPF). 1

9 1.3 Outline There are two sections to this report. The first part will describe the test methodology and calculations to be used for comparison. The second will explore the use of sound absorbing foam, changing the shape of the speaker grill and application of a LPF to see their effect on the frequency response. Each of the three methods will include a comparison table for before and after changes in intensity. A conclusion will be drawn to determine which methods can be used and which frequencies they alter. Followed by a recommendation on which methods the senior engineers can use. 2

10 2 Test Setup 2.1 Requirements The desk phone prototype has to pass many frequency response requirements. There are two ways a user can interact with the desk phone. One way is to pick up and talk in hand set (HS) mode. Another is to talk from a distance using the hands free (HF) mode. Different speakers are used for HS and HF mode. To further complicate things, wide band (WB) and narrow band (NB) audio can be used for the individual components. Thus all test cases need to be tested individually. The TIA standard for WB HS response requires the prototype to deliver a relatively constant intensity at frequencies from 100Hz Hz [2]. The extreme lows and highs can be negligible as the stated in section 1.1, human conversation fluctuate between 300Hz to 3000Hz. This report will only deal with WB as the purpose is not to document effects for all possible scenarios but to analyze whether certain modifications will result in a significant change in the frequency response that can be applied to both WB and NB. Any modifications applied that causes a frequency change in WB will also cause the same change in NB. Figure 1 visually represents the upper and lower bounds the prototype must obey. The range between approximately 300Hz to 5000Hz allows to only a relative -12dB difference between the highest and lowest intensities. Bounds outside of that range is more lenient due to reasons stated in section 1.1. Figure 1: Wide Band Hands Free Speaker Standard. [2] 3

11 2.2 Methodology & Equipment Used This section describes the equipment and methodology used to test for the frequency response. Equipments used include the Anechoic room, Microtronix Measurement, Artificial Ear Type 3 and Artificial Mouth. Measurement of frequency response is very delicate process. It is vital to ensure the placement of the phone, artificial ear and speakers be placed in the exact specified during every test set by the TIA. As with any standardized testing, having the same test environment is necessary to ensures the data can be compared to other different organizations. Figure 2 shows the telephone placed in the anechoic chamber 40cm adjacent and 50cm hypotenuse from the artificial mouth or ear [3]. The telephone is then connected with the Microtronix Measurement Unit which in essence sends out a pink noise signal to the artificial mouth and listens to the phone microphone response. In vice versa, the Microtronix unit will send a pink noise to the phone s speaker and listen to the response on the artificial ear. Figure 2: Anechoic Chamber at Aastra Telecom. [3] After the frequency response graphs have been generated, the equation used to find in loudness for before and after modification is = 10 x y 10 X and Y are both mean average intensities of frequencies of a certain range that will be specified. X is the intensity before the modification and Y is the intensity after the modification 4

12 3 Telephone Modifications This section will evaluate the different modifications that affect frequency response. The results of this report will be a recommendation of what changes can be used to apply to the prototype. 3.1 Sound Absorbing Foam Speaker cavities can cause distortions due to sound waves bouncing off the walls and back to the speaker. This can alter the frequency response of what the manufacture intended the speaker to be. Larger speaker cavities have more room for the sound waves to dissolve [4]. Obtaining a good response with a modern desk phone is difficult as current trends focus on slimmer and lighter chassis. The chassis of the prototype has been finalized thus changing the dimensions of the cavity is not possible. There are however certain modifications that can improve the low frequency response. Low frequency sound tend to be produced in front and behind the speaker. One possible solution to this is to fill the cavity with sound absorbing material such as fiber glass. The sound absorbing material will act to absorb the lower frequencies in cavity. The speaker should produce louder low frequency sounds. The frequency response of the original set up is shown in figure 3. at 5000Hz in which it dips below the requirement. absorbing foam increases the bass in this experiment. The speaker is generally passing with the exception It is expected that the sound Figure 4 displays a similar response to Figure 3 with a slight increase in the lower end response. There does not appear to be any significant changes that merit this modification. This method is ineffective. Table 1 shows the overall intensity to have decreased by only 1.08 times. This method is ineffective thus should be avoided. Table 1: Sound Absorbing Foam Response - Comparison 200Hz Hz Intensity (db) Before After x 5

13 generated by JPG to PDF Figure 3: Frequency Response of Speaker - Without Foam. [?] generated by JPG to PDF Figure 4: Frequency Response of Speaker - With Foam. [?] 3.2 Speaker Grill Modification Diffraction in such cases helps the sound to bend around the obstacle [5]. When sound passes around an object, it then acts as a new sound source at that location. Any waves emitted from that point can then cause constructive or destructive interference. Ideally, it is best to not have a grill in front of the speaker and it alters the response the manufacturer intended it to be. This is unrealistic as there needs to be a grill to protect the speaker from damages. The speaker from the manufacturer was designed to pass the TIA standard on its own but when inserted in the chassis, the plastic grille in front caused it to fail. The current design of the HS has a bowl 6

14 shape for ergonomic reasons. This results in destructive interference as the outer openings are at a different elevation compared to the center openings. The current design of the speaker grill may have contributed to sound diffraction. Figure 5 shows the original response of the handset. The Y before and after 2000Hz is too large thus causing the miss the targeted area. It is desirable to increase the intensity of frequencies from 2000Hz to 5000Hz. Figure 5: Unplugged Handset Speaker Frequency Response. [?] In this test, the outer holes of ths grill will be airtight sealed with rubber putty as shown in Figure 6. The plugged holes are on a different elevation to the unplugged openings. This should reduce the destructive interference as there are less localized sources at different elevations. The Handset s speaker is then tested for its frequency response. Figure 7 shows an increase in intensity from 2000Hz to 4000Hz when compared to 7. There is however a dramatic decrease in frequencies of 5000Hz and up. This experiment shows the speaker grill design can be changed to address the validity of the frequency response. Table 2 summarizes the intensity of the frequencies from 2000Hz Hz. There modified handset shows an increase of 2.39x in that range. This exact modification is not recommended however more research should be done in this area. 7

15 Figure 6: Modified Handset With Reduced Opening. [6] Figure 7: Handset With Reduced Opening. [?] 3.3 RC Low Pass Filter Human conversation is comprised of sinusoidal waves; these analog waveforms are captured by the microphone. The signal is then encoded via Analog-to-Digital encoders. Transmission of data through Voice Over Internet Protocol is in its digital form. Before the data is transmitted, it sometimes gets passed through a variety of filters to remove certain frequencies that are undesirable such as noise. This experiment will evaluate the effects of adding a Capacitor-Resistor Low Pass filter. Figure 8 shows schematic of a similar LPF the prototype uses with a capacitor value of 22nF. Vin contains the signal from the Microtronix test unit goes through 8

16 Table 2: Speaker Grill Diffraction Response Comparison 2KHz - 5KHz Intensity (db) Before After x Vin via a low pass filter. High frequencies should be reduced before passing through V out to the speaker. The filter was soldered on by the hardware team. Figure 8: RC Low Pass Filter. [?] Figure 9 is the HF frequency response before the LPF was added. It is currently passing the TIA standard however the higher frequencies are too close to the upper bounds. This is undesirable as uncertainties from manufacturing imperfections can raise it. Figure 9: Speaker Without Filter. [?] Figure 10 shows the result of an added RC filter; frequencies above 3000Hz appear 9

17 to be dampened. This is favorable as it creates an overall flatter response. During conversations, frequencies above 3000Hz will not spike in intensity which results in discomfort to the listener. Figure 10: Speaker With 22nF RC Filter. [?] Table 3 provides a summary of the reducetion of intensity from 2000Hz Hz. This is favourable as it is 2 times lower in volume. This modification is recommended. Table 3: RC Low Pass Filter Response - Comparison 2KHz - 5KHz Intensity (db) Before After x 10

18 4 Conclusions From the analysis in the report body, it concludes the addition of the 22nF capacitor can be used to increase or dampen frequencies from 2000Hz Hz. The modification of speaker grille stated in the report cannot be used directly. It requires further research before it can be implemented on the prototype. The sound absorbing foam modification should be avoided as it does not significantly alter frequency response. Table 4 shows a summary of the modifications and their respective changes in volume. Section 3.1 evaluated the use of sound absorbing foam in the speaker cavity. The increase of only 1.08 time the overall sound intensity deems this method ineffective. Section 3.2 evaluated the effects on frequency response by modifying the speaker grille of the handset. The modified handset resulted in a 2.39x increase in loudness for frequencies. This modification proves viable in increasing the intensity from 2KHz - 5KHz. This modification does not improve the frequency response but alters it in a unfavorable way. More research is needed before this modification can be recommended. Section 3.3 evaluated the drop in intensity between 2KHz and 5KHz to be 2x decrease in loudness. This modification proves viable in decreasing the intensity from 2KHz - 5KHz. This modification is recommended. Table 4: Summary Comparison Sound Absorbing Foam Handset Modification LPF Intensity -1.08x +2.39x -2.0x 11

19 5 Recommendations The experiments has shown the handset modification and adding a LPF has desirable effects on frequency response. It flattens out the frequency response between 2000Hz Hz. If future iterations of the prototype exhibit the same problems then it is recommended the senior engineers add a RC LPF to address it. It is also recommended the modification of the speaker grille be researched further. This report has proven it to alter the frequency response, although the results are currently unfavorable. This is still a problem that needs further attention. 12

20 6 Glossary TIA Telecommunications Industry Association WB Wide Band NB Narrow Band RLR Receive Loudness Rating SLR Send Loudness Rating RC Resistor-Capacitor LPF Low Pass Filter IP Internet Protocol VOIP Voice Over IP 13

21 References [1] Definition of telephone : Merriam-webster. dictionary/telephone?show=0&t= (current Mar. 2013). [2] N. N. Corporation, Voice Fundamentals /Voice-Fundamentals (current Aug. 2002). [3] T. I. Association, TIA [4] J. L. Murphy. [5] R. Nave, Diffraction of sound. sound/diffrac.html#c1. [6] G. Wen, Modified handset. 14

22 Appendix A Frequency Response Data Points Table 5: Sound Absorbing Foam Response - Before Freq (Hz). Intensity (db)

23 Table 6: Sound Absorbing Foam Response - After Freq (Hz). Intensity (db)

24 Table 7: Speaker Grill Diffraction Response - Before Freq (Hz). Intensity (db) Freq(Hz). Intensity (db) Freq(Hz). Intensity (db)

25 Table 8: Speaker Grill Diffraction Response - After Freq (Hz). Intensity (db) Freq(Hz). Intensity (db) Freq(Hz). Intensity (db)

26 Table 9: RC Low Pass Filter Response - Before Freq (Hz). Intensity (db)

27 Table 10: RC Low Pass Filter Response - After Freq (Hz). Intensity (db)