1 Operations and Applications of a Laser Microphone 2 Ed Fortin and David Cui 3 University of the Fraser Valley, Department of Physics 4 (Dated: April 28, 2010) Abstract A laser microphone was constructed using a red Helium-Neon laser and a photoresistor circuit as a detector. It was tested using a glass window as the vibrating membrane and a speaker powered by a function generator as a monotone sound source. The detected signal voltage waveform was then analyzed and its frequency compared to the frequency of the sound source. We found that our detected signal frequency was the same as our sound signal frequency within experimental error, allowing us to conclude that we could detect and reproduce sound waves using a window as a vibrating membrane. 1 5 I. INTRODUCTION 6 One method of surveillance which was developed in the mid 1950’s enabled people to 7 listen in on conversations at a distance through the use of an infrared laser. By bouncing 8 the laser off a window and collecting the reflected signal with a detector, they were able 9 to bug any room with a view. Sound waves travel through the air and interact with the 10 window, causing it to vibrate. The laser’s angle of reflection changes slightly with this 11 vibration, resulting in variations of intensity on the detector. By translating the variations 12 of intensity into a change in voltage, one can reproduce the sounds that are being generated 13 from the vibrating window pane. 14 The goal of this experiment was to determine the accuracy of a laser microphone. We 15 will compare the frequency of the signal detected by the apparatus to the frequency of a 16 given sound signal. In the apparatus section the different parts of the apparatus will be 17 explained. The theory section will outline the different equations which will be used to find 18 the output frequencies. The procedure section will go through the steps of setting up the 19 apparatus and measuring test frequencies. In the analysis section, the recorded data will be 20 confirmed using error analysis and a graph comparing the sound signal to the detected signal 21 will be produced. The results section will then display the confirmed fidelity by showing the 22 output versus input graph. The discussion section will go over where errors or difficulties 23 were encountered as well as some possible solutions. Finally, the conclusion will detail the 24 ways this lab may be expanded and improved. 25 II. APPARATUS 26 The apparatus for this lab was spilt up into three parts; the emitter, the membrane, 27 and the collector: as seen in Figure. 1. The emitter part consisted of the laser, including 28 mounting pieces that controlled the incidence angle, and the speaker powered by a frequency 29 generator which provided the sound signals. The membrane consisted of the surface off of 30 which the laser is reflected. The collector gathered the reflected signal, interpreted the 31 change in intensity created by the window vibrations, and converted the intensity signal 32 into a voltage. Using the collector’s electronics, the voltage change as seen on a oscilloscope 33 could be interpreted as sound waves. 2 FIG. 1: Laser Microphone Apparatus 34 A. Emitter 35 The first part of the emitter apparatus consisted of the Helium-Neon 10mW red laser 36 with inventory code 114400733, which was aimed at the window so it reflected back to the 37 collector. The other part of the emitter was the REALISTIC 8 Ω speaker, which created 38 the input signal to be collected by the laser’s vibrations. We controlled the frequency 39 and amplitude of the input signal by connecting the B+K PRECISION 4017A 10MHz 40 SWEEP/FUNCTION GENERATOR to the speaker. 41 B. Membrane 42 We tested several reflective objects prior to using the window. Initially the bottom of a 43 beaker was used; however due to its size it did not oscillate easily and the output signals 44 were full of noise. A drum made from a plastic bag was used next. We observed that only 45 frequencies around 200 Hz seemed to give any coherent output signal to the drum when 46 using the PasPort light detector. This was probably the resonant frequency of the drum, 47 which would explain why only one frequency worked. Higher frequencies were extremely 48 difficult to detect due to the limit of 1000 Hz in the PasPort light detector sampling rate. 3 49 Finally a window approximately 1.1 metres by 1.45 metres was selected. The window was 50 large enough to vibrate easily when a speaker emitted sound waves towards it. The window 51 was also selected because it held true to the concept behind this experiment: that this laser 52 microphone was used as a surveillance device on exterior windows. 53 C. Collector 54 The collector part of the apparatus firstly consisted of the photo resistor voltage divider, 55 which converted the varying intensity of the laser into a voltage signal. A photo resistor is 56 a device that changes resistance with respect to the intensity of light it is exposed to. To 57 minimize amount of background light, a red filter plate was mounted to the end of a small 58 box with the photo resistor at the other end of the box. Closing the box ensured that the 59 only light reaching the photo resistor was the light from the red laser. The photo resistor 60 was followed by a series of high pass filters. These were used to filter out background noise 61 from the window. This background noise was likely noise from outside, ambient noises from 62 inside the room, and was possibly due to the harmonic frequency of the glass. The voltage 63 signal was then sent through an amplifier which boosted the signal and created a readable 64 signal for the oscilloscope. The oscilloscope was used as the display for the output signal. A 65 741 op amp was selected for the amplifier for its simplicity and ease of use.1 Using the scope, 66 the output signal was visible as a waveform and was analyzed using the run/stop function 67 to determine the frequency and magnitude of the output waveform. Fig. 2 shows a circuit 68 diagram of the detector unit. 69 III. THEORY A laser microphone is a device used to detect vibrations due to sound on a distant, reflective object. Sound waves interact with the object, a glass window in our experiment, causing it to oscillate. When the window vibrates, so too does the laser’s mark on the photo resistor. The collector detects these vibrations as minute changes in intensity; these changes are the sound waves from inside the room. The collector amplifies this voltage signal so the small vibrations become an oscillating voltage signal. This output signal is interpreted as sound waves which power a speaker so the eavesdroppers can listen to sound from inside 4 FIG. 2: Circuit Diagram of the Detector Unit. the room.3 The detected signals are found by measuring the period of the oscillations of the window using the oscilloscope. The period of multiple wavelengths are measured and converted into frequency by the Equation: f requency = n Hz, T (1) 70 where n is the number of wavelengths being measured, and T is the time to traverse those 71 wavelengths. The glass membrane can oscillate due to signals other than sound, such as ground vibrations and wind. These appear as relatively low frequency (<50 Hz) signals, but are much stronger than the signals created by sound. In order to remove these background noises, a high pass filter was included in the apparatus. Passive filters placed in series were used in this experiment to quickly attenuate unwanted frequencies. A background frequency of 16 Hz was noticed during the developmental stages of the experiment, so high pass filters were included that attenuated frequencies below approximately 33 Hz. We chose a cut off frequency for the filter above our background noise since the passive filters attenuate frequencies according to a logarithmic scale.1 A high pass filter was constructed by having the input run through a voltage divider consisting of a capacitor then a resistor as seen in Fig. 5 3. The cut off frequency of the filter was determined using the equation: f requencycutof f = 1 , 2πRC (2) 72 where R is the resistance of the resistor in ohms and C is the capacitance of the capacitor 73 in farads. FIG. 3: Schematic for a High Pass Filter 74 IV. PROCEDURE 75 The apparatus was set up as described in the previous section. The laser was turned 76 on, leveled, and trained on the window so the beam reflected into the collector’s housing. 77 Once the laser was properly trained onto the photo resistor, the detector’s housing was 78 closed to minimize the amount of external light that reached the photo resistor. The sound 79 source was then set up approximately 0.05m from the window to simulate someone talking. 80 The frequency generator was connected to the speaker and was set to a test frequency. 81 The oscilloscope was then used to find the waveform detected from the laser. A range of 82 frequencies were tested using the frequency generator and the output waveform’s periods 83 were recorded in Table I to find the output frequency, and compare it to the original input 84 signal. 6 TABLE I: Frequencies (Hz) Source Measured Quality Source Measured Quality 120 120.48 clear 570 566.89 clear 150 150.15 clear 600 603.86 clear 180 179.21 clear 630 628.14 clear 210 210.08 clear 660 652.74 unstable and jagged 240 240.38 clear 690 692.52 clear 270 270.27 clear 720 718.39 clear 300 301.20 clear 750 748.50 jagged 330 322.58 clear 780 783.70 unstable 360 362.32 jagged 810 809.06 clear 390 383.14 jagged 840 841.75 unstable 420 420.17 clear 870 862.07 unstable 450 452.49 clear 900 886.07 unstable 480 478.47 clear 930 925.93 unstable 510 507.61 clear 960 961.54 unstable 540 534.76 jagged 990 988.14 unstable 1020 1028.81 unstable 85 It should be noted that some of the output signals were classified as either jagged, or 86 unstable. A jagged waveform is one that has a measurable period; however the overall 87 waveform is not smooth. An unstable waveform is one in which the waveform is smooth, 88 but not continuous. In other words an unstable waveform would be present for a time, and 89 then it would became unrecognizable noise for a short amount of time. These unstable waves 90 were measured with the oscilloscope’s run/stop toggle, capturing a snapshot of the waveform 91 which could be analyzed. Another point of interest was that all of the signals became smooth 92 and coherent once the output level was maximized on the frequency generator. If the output 93 level was left unchanged, some of the signals did not appear. 7 94 V. ANALYSIS To determine the fidelity of our output signal, we directly compared the frequency of the output signal to the frequency of the input signal at a given input frequency. We compared them directly by finding their ratio: Ratio = OutputFrequency(Hz) , InputFrequency(Hz) (3) where the output frequency is obtained from our detected signal, and the input frequency is obtained from the sound signal. A frequency ratio of 1.0 would indicate that our detected signal and the sound signal are identical. Ratios of output frequency to input frequency were created for each data point. The error present in both input and output frequencies were combined and shown as the error of the ratio. A linear approximation was then constructed: Ratio = (a) Out(Hz) + b, In(Hz) (4) 95 where a = (−0.5 ± 5.0) × 10−6 and b = 0.998 ± 0.003. A linear approximation was chosen 96 because the ratio was not expected to change with the signal, and was expected to be 97 horizontal. The ratios at each point, the error bars, and the linear approximation can be 98 seen in Fig. 4. FIG. 4: Ratio of the frequency of the Input frequency to the Output frequency 8 99 VI. RESULTS 100 The purpose of this experiment was to reproduce a sound signal by reflecting a laser off 101 of a vibrating membrane and measuring the change in intensity of the reflected beam on a 102 detector. We found that we could reproduce an output signal of the same frequency as the 103 input signal. This was confirmed since the error bars shown in Fig. 4 contain the ratio value 104 of 1.0. In the linear approximation, the coefficient a in Equation 4 was extremely small, 105 though it had a large error relative to its value, indicating that the ratio was essentially 106 constant over the range of our experiment. As the coefficient b of Equation 4 included the 107 ratio value of 1.0 within its error, the sound signal frequency and detected signal frequency 108 were found to be the same over the range of our experiment. 109 VII. DISCUSSION 110 Originally the collector circuit made use of a PasPort high sensitivity light sensor. The 111 light sensor was connected by USB to a computer and used software which plots the intensity 112 of light versus time. The sampling rate of the PasPort was unfortunately limited to 1000 113 frames per second. This sample rate was sufficient for detecting footfalls or other shockwaves, 114 but it was far too slow for measuring sound waves. The solution was to use a photo resistor in 115 a voltage divider circuit to maximize the sample rate. This also required further electronics 116 to purify and amplify the detected signal to make it large enough to power a speaker. 117 In order to collect the highest possible output waveform amplitude, the detected signal 118 was sent through an amplifier. Other resistor combinations on the amplifier with lower and 119 higher gain ratios were attempted; however beyond a gain of 100× there was not a drastic 120 change in the output signal size. This might have been due to the pre-loading effect a series 121 of high-pass filters has. Pre-loading occurs when there is unwanted feedback in a series of 122 devices. This feedback causes a cascade of altered signals throughout the circuit, altering 123 the output. Since a high-pass filter works by a relationship between its input voltage and 124 its output voltage, only the first filter would have worked as expected. The remainder of 125 the series would have received differing voltages; which would have drastically reduced our 126 signal strength. Another possibility is the signal collected by the photo resistor was too 127 small to amplify to the desired magnitudes. 9 128 Constant background noise which formed a 16Hz wave was minimized by the series of 129 high-pass filters; persistent background noise remained and was amplified along with the 130 desired signal. One solutions was to raise the selected cut off frequency to approximately 131 100Hz. The high-pass filter causes a drop off of signal amplitude at a logarithmic rate 132 below the cut off frequency.1 Alternatively, we found that raising the cut off frequency also 133 decreases the amplitude of the output signal. Additional amplification could allow us to 134 power a speaker using this low amplitude signal. 135 VIII. CONCLUSION 136 In this experiment, we constructed and tested a laser microphone. From testing a range 137 of input frequencies, we found that it is possible to detect and reproduce, with great fidelity, 138 a sound signal on a glass pane. The range of frequencies we used was within the range 139 of normal speech frequencies for humans; however it did not include the extreme lower or 140 upper limits of human hearing which ranges from 20 Hz to 20 kHz.2 141 We began to test polytonal signals with the intent to progress to recognizable speech but 142 were hampered by time constraints. What little was observed using two sound input signals 143 was promising, as the combining of sound waves in our output coincided with expected 144 wave functions. It may be difficult to determine what waves are present in a polytonal 145 signal using an oscilloscope, so a new measurement method may be necessary. A qualitative 146 test may be useful, simply amplifying the signal and sending it through another speaker to 147 be heard and compared with the original sound signal. Perhaps there is a program available 148 to dissassemble such waves forms. 149 Further amplification of the signal, or increased sensitivity of the detector, would be 150 necessary in order to detect normal speech. The input in this experiment was at maximum 151 volume so lower volume situations would also have to be tested, possibly revealing additional 152 unwanted background noise. 153 154 1 A. S. Sedra, K. C Smith, Microelectronic Circuits (Oxford University Press, 2004), pp. 69,10841099. 10 155 2 Sensitivity of the Human Ear, http://www.hyperphysics.phy- astr.gsu.edu/4base/sound/enrseus.html , date accessed, April 7, 2010. 156 157 R.Nave, 3 “Laser Microphone,” http://williamson-labs.com/laser-mid.htm, date accessed, March 24, 2010. 11