Optical Theremin Critical Design Review Yanzhe Zhao, Mason Story, Nicholas Czesak March-07-2015
Abstract A theremin is a musical instrument whose tone and pitch can be controlled without physical contact. A traditional theremin makes use of antennas and oscillators in order to allow the user to control the volume and pitch. However, there are a number of ways to implement the overall design for a theremin without the use of antennas and oscillators. One of them takes the form of what is known as an optical theremin, which uses the intensity of incoming light rays to control the characteristics of the theremin. A useful method of implementing an optical theremin is the use of two OP906 photodiodes. Passing each current generated through two 1 MΩ feedback resistors implemented with two Texas instrument s TL074 lownoise JFET-input Op-Amps produces two output voltages that can be thought of as the volume and pitch, respectively, of the optical theremin. The magnitude of the output voltage is therefore controlled by the intensity of light shined of the photodiodes, which can be controlled through how the user moves their hand relative to the diodes themselves. This data can then be input into an NI mydaq and processed through an NI LabVIEW program. The LabVIEW program can normalize and scale the data to fit the requirements of the optical theremin, and afterwhich include settings for auto-tuning and distortion. The final waveform should be able to be output by the mydaq and processed by a set of speakers to produce an audio signal. Introduction In order to prototype a minimalistic, functional optical theremin our team will first have to decide how best to obtain a signal from the intensity of ambient lighting. With minimal parts provided we had to come up with a simple design with most of the manipulation occurring within LabVIEW. The final design our team constructed provides the user with complete manipulation of the audio signal through LabVIEW and the intensity of the ambient lighting. With a fully customizable front panel our solution provides further customization in the future. Our team designed a solution assuming that the user is in a well lit room with basic knowledge of LabVIEW. Our vision is to develop a fully customizable LabVIEW VI in conjunction with a simple integrated circuit design. This project will provide the ability to operate a optical theremin with complete signal manipulation capabilities including scalability, auto-tune ability, and distortion. The specific problems we encountered within our design were the normalization and scaling of the signal, auto-tuning of the frequency, and distortion of the amplitude. The signal voltage obtained through the MyDAQ will vary with the intensity of the ambient lighting of the room. In order to provide a consistent signal it first needs to be normalized between 0 and 1 volt. With normalization this allows the user to take the optical theremin into any room and quickly calibrate it; our design allows this to happen. With a normalized frequency and amplitude the user can then determine the limits of the frequency via controls on the front panel. Our solution provides the user the option to limit the theremin to a single octaves or a multitude of octaves. The ambient lighting in the room will provide noise within the signal, so in order to account for the frequencies our solution provides a auto-tuning function to coerce the raw pitch to a whole note pitch. Finally our solution allows the user to distort his signal in two distinct ways providing variety from the theremin. The theory behind our solution is provided within the rationale section and the specific setup is described within the implementation section of this paper.
Rationale In order to make a functional optical theremin with photodiodes we decided on using a transimpedance amplifier. With the photodiodes acting as current sources, the transimpedance amplifier will produce a gain voltage dependant on the feedback resistor and the current from the diodes. With the signal produced by the photodiodes in a voltage, the mydaq can transfer the signal into LabVIEW for manipulation. Once in LabVIEW the signal will first need to be normalized. This will make any signal manipulations easier because the signal is now between zero and one volt. Normalization was achieved by implementing a calibration scale on the front panel in LabVIEW. With the given requirements that the final audio signal be contained within 20Hz to 20KHz we needed to scale that the signal and offset it accordingly. We multiplied the normalized waveform with the difference of the maximum and minimum desired frequencies. This in conjuncture of the minimum frequency offset will produce a signal within the specified frequencies. The last requirement for the first portion of this project was to implement a normalized, scaled, volume. Similar to the frequency, the voltage for the amplitude was normalized by implementing a calibration control on the front panel in LabVIEW. The next portion of the project required a user controlled switch to select the frequency used to generate the audio waveform. The two options were the unmanipulated frequency generated from the scaled normalization waveform or a auto-tuned frequency. In order to auto-tune the frequency, our signal was compared to an array of scientifically established note frequency; we decided on eight octaves ranging from 20Hz to 20KHz. Our signal was then corrected to the scientific frequency by thresholding the value to the closest determined pitch. The final portion of our lab was to add distortion to our signal. There were three options to implement no clipping, hard clipping, and soft clipping. No clipping had no effect on the original waveform. Hard and soft clipping altered the peak voltage based on a control on the front panel in LabVIEW. Hard clipping coerced the signal to stay within the values decided by the control. This results in a straight cutoff with sharp edges on the waveform. Soft clipping required the signal generated to be divided by the percentage of clipping this provides the exponent we can apply to our signal. This allows us to distort our signal while maintaining our curved peak on the waveform. Implementation We used op906 photodiodes as a negative current source leading from the LT074CN operational amplifier with feedback resistors of 1 MΩ to implement two transimpedence amplifiers using two of the opamps on the integrated circuit; specifically pins 1 through 7 see Appendix A. We sampled both analog input channels on our mydaq to create two channels corresponding to frequency and amplitude for our optical theremin which we separated in LabVIEW more information can be found in Appendix B. To eliminate noise within the waveforms we averaged out the signals using LabVIEW s mean function. For frequency we calibrated the normalization, scaled the waveform, and implemented a frequency offset within a formula node as seen in figure A. The calibration controls to normalize the frequency are MaxInt and MinInt with range of 0 to 10. Once normalized between 0 and 1 Volt the controller can set the range that the frequency can vary using the MinFreq and MaxFreq controls with the limits being 20Hz and 20KHz. The difference in this range is multiplied by the normalization in order to scale the signal to the desired level so that when the MinFreq is added to the waveform the signal is meets the specifications. Another control is used to calibrate the normalization of the amplitude. This control is split into a Max and a Min used in the normalization equation with the signal values X.
We then implemented a SubVI in order to auto-tune the frequency. Within this VI we use LabVIEW s threshold function, where our scaled frequency is one input and an array of scientifically accepted note frequencies is the other input. When the user selects the auto-tune function, the raw frequency is compared with the note frequency array and the thresholding function returns a fractional index. We round the index down to the nearest whole number and extract the corresponding scientific pitch to replace in our waveform. With a new frequency and amplitude we use the simulate signal function to produce an audio signal within the specifications of the mydaq audio out limits of 土 2 volts see Appendix B. To simplify the distortion portion of this project we used a case structure with enumerated states of no clipping, hard clipping, and soft clipping. We also used a slider on the front panel with a scale of.3 to.9 or the range of clipping required. The no clipping state required no manipulation of the signal. Hard clipping uses the coerce function with the input of our signal and the limits of the positive and negative value obtained by the clipping slider control. The soft clipping required our clipping control divided by the waveform value to produce a exponent to linearly compress our original waveform without hard clipping. The entirety of the LabVIEW schematic can be found in Appendix F. Conclusion By the end of the first week of March, we were able to develop a final design for the overall circuit. This included the transimpedence amplifier circuit and the virtual circuit that was developed through LabVIEW. The schematic for the amplifier, its assembly, and measurements taken presented little problems to us. The real challenge from the project came with the realization of the LabVIEW code. However, in the end, our final design met the needs for the project and was able to communicate with the signal produced by the transimpedance amplifier circuit. We were able to generate an audio signal from the mydaq that could be processed and output by a set of speakers. We were able to normalize the data coming from the amplifier circuit by making use of the function node in our Block Diagram. This helped to condense our code to a smaller section of the Block Diagram than when we tried to implement arithmetic VI s. This method of implementation would also be significantly more readable for anyone analyzing our Block Diagram and trying to comprehend what the code was trying to accomplish. In regards to the auto-tuning of the theremin, the SubVI we utilized was successful with our method of programming it, which involved setting up an array containing all the specific frequencies contained within each octave. We realize now this may not have been the most efficient way to accomplish this task. This influenced our decision to generate a SubVI for this section of LabVIEW code so it wouldn t take up significant space in our Block Diagram. After passing through the AutoTune SubVI, the new data then passed through a simulate signal function producing the audio signal. This new audio signal section of code dealt with the clipping of the audio signal to produce distortion. The method to accomplish this, which involved the use of a case structure, was deemed pretty obvious by everyone on the team after reading through the requirements. The real challenge revealed itself when we were discussing how to implement each choice for clipping provided to the user. In the end, the final result seemed to mesh well with the remaining portions of the block diagram. Overall, the LabVIEW code utilized in this project proved to be more than enough to accomplish all of the required tasks. The only problem that was encountered in our overall design was the use of a diode to change the volume of the theremin. Covering the diode actually caused the volume to rise instead of
decrease. This is not what we were expecting to occur, although it had no effect on any of the other controls used in the final design, especially the other diode that was used to control frequency. Appendices Appendix A: Data Sheet for TL074 low-noise JFET-input Op-Amp:
:
Appendix B: Data Sheet for NI mydaq:
Appendix C: OP906 Diode Datasheet:
Appendix D: Block Diagram for the Optical Theremin:
Appendix E: Gantt Chart Appendix F: LabVIEW code
Theremin Main VI: AutoTune SubVi: Pitch array containing 8 octaves of frequencies