6.101 Final Project Proposal Woojeong Elena Byun Jack Erdozain Farita Tasnim 7 April 2016 Fluxgate Magnetometer Motivation: A fluxgate magnetometer is a highly precise magnetic field sensor. Its typical sensitivity range is fit for measuring earth's magnetic fields, but it is also capable of resolving external magnetic field strengths less than.01% of that range. The core technology behind a fluxgate magnetometer was invented in 1936 before WWII with the goal of easily detecting submarines; upon its invention, its impressively high sensitivity helped prove the theory of plate tectonics by measuring shifts in magnetic patterns on the seafloor. Fluxgate magnetometers are widely used in both industry and academia because they are affordable, rugged, and compact; they have been miniaturized to the point of IC sensor solutions with recent technology. The applications of a fluxgate magnetometer are many and varied. They can be used to observe small changes in the Earth's magnetic field for earthquake detection. Their field detection can be used to detect solar phenomena on earth. Recently, NASA invented a magnetometer technology employing many fluxgates placed around a satellite in order to measure a combination of spacecraft-generated noise and magnetic field measurement data. Using subtractive algorithms, the two were separated for accurate magnetic field data collection. The low-power, compact, and inexpensive fluxgate no-boom solution has since been employed in NASA's CubeSat units. The fluxgate elegantly manipulates a physical concept to create an extremely practical and precise sensor. Our team has decided to design and construct a fluxgate magnetometer that demonstrates toroidal drive, signal processing, and display of the signal with actuation and through MATLAB. Project Overview: Our goal is to develop a sensor that can measure magnetic field with high precision. To do this, we will be creating a one-axis fluxgate magnetometer that 1
can measure changes in the Earth s magnetic field, as in response to solar phenomena and earthquakes, for example. The output of the fluxgate magnetometer will be displayed in two selectable modes: MATLAB to display the subtle changes in the Earth s magnetic field or a servo motor that follows the direction of a test magnet. Our block diagram, as seen in Figure 1, outlines the major components of our system. Figure 1. One-Axis Fluxgate Magnetometer Block Diagram Farita, Jack, and Elena will be dividing the project into four different components: physical toroid and sense coil construction and tuning, phase demodulation and signal extraction, timing and excitation, and signal display and actuation circuitry. If time permits, we will all work together to add active current feedback. First, all three of us will work together on creating the physical toroid and sense coil, as well as creating a proof of concept in the very first week to 2
demonstrate that our project is feasible. Then, Farita and Elena will work on phase demodulation and signal extraction. Jack will make the timing and excitation circuitry. After these two parts are working, Farita and Elena will make the signal display and actuation circuitry. If time permits, we will all work on extending our project from a one-axis magnetometer to a three-axis magnetometer, which would add complexity since we would have to prevent crosstalk between the axes. Stages: Toroid and Coil Construction: A typical fluxgate consists of a drive coil wound around a toroidal magnetic core surrounded by a sense coil. In order to ensure that our drive current will be able to fully saturate the core in both directions, the selected core material must be highly permeable. A ferrite core from EDS will be used in order to conduct the proof of concept experiment; for the rest of our project, we will purchase permalloy (nickel-iron alloy) and amorphous glass ( Metglas, iron, boron, silicon, phosphorous alloy) cores, which have very high magnetic permeabilities. Magnetic wire will be acquired from EDS for the drive and sense coil windings. The magnetic field of the toroidal drive coil will be calculated as B = μni/ (2πr). The number of windings needed will be decided by studying the BH curve for the core of interest and ensuring that we are providing a current that is 10 to 100 times the nominal saturation current. The appropriate number of windings on the sense coil will be determined through experimental data. Timing and Excitation Circuitry: For the system to work, the core must be driven into saturation in both the positive and negative direction. The waveform across the excitation coil is a periodic square wave centered around the x axis. The frequency and duty cycle (which will be set to 50%) can be generated and controlled via a 555 astable oscillator as shown in Figure 2. The Vref in Figure 2 is either produced by an adjustable potentiometer or a fixed voltage divider. The 555 oscillator produces a sawtooth curve with the frequency determined by the RC characteristic of the 555 circuit. This will be designed to operate somewhere between the range of 1 KHz to 10 KHz. This curve is sent into a LM311 comparator to produce a square wave. Adjusting Vref will 3
Figure 2. 555 and 311 circuitry adjust the duty cycle of the signal. We selected to use a comparator in conjunction with the 555, instead of just using the square wave output of the LM311 because it will allow us to more accurately adjust the duty cycle. It is important to ensure that the average voltage across the coils is zero as this will ensure that the average magnetic field is zero, meaning the drive circuitry will not cause offset in the output signal. This output will then be fed into an H bridge (one side input with inverted signal). The H bridge will consist of 4 MOSFETs each driven by gate drivers to ensure that they are being fully pushed into saturation. Dead-band circuitry will be built in to help avoid any possibility of shoot through current. The H-bridge circuitry will be done on a specially fabricated PCB. Signal Extraction with Phase Demodulation: We are saturating the toroidal core at frequency f, so the external magnetic field being sensed appears at twice the core driving frequency. The external magnetic field signal is not symmetric, and thus has an average equivalent to a DC offset that is proportional to the strength of the magnetic field running perpendicular through the sense coil. The output signal is then fed into a bandpass filter to reduce noise at frequencies lower and higher than the desired range of the sensed frequencies, which is between 5 KHz and 20 KHz. This should eliminate 60 Hz noise from power lines as well as noise from radio frequencies in the MHz range. The output of the bandpass pass filter is then fed into the phase demodulation stage. This phase demodulation stage extracts only output signals in phase with the drive frequency. We accomplish this by comparing the output signal with a reference sine wave signal at frequency 2f that is also in phase with the drive frequency. This reference signal will be fed into an adjustable RLC phase 4
shifter that will be used to ensure that any phase shift caused by a delay in the sense coil pickup is accounted for so that our desired signal is properly in phase with the drive signal. The phase demodulation therefore removes any odd harmonics and other noise that are not in phase with our drive frequency, thus giving us clear signal voltages proportional to external magnetic field strength. The phase demodulated output is then fed into an integrator in order to extract the DC average of the output signal, which is proportional to the strength of the magnetic field and which has directionality indicative of the direction of the sensed magnetic field. Since we are not exactly sure of the expected signals from the sense coil, we might replace our integrator block with a rectification block in which we half-wave rectify the positive and negative half cycles of our output signal and feed the differential DC outputs into an instrumentation amplifier to magnify the difference in the positive and negative peaks of our phase demodulated output. The output of the integrator or rectification block will then be amplified to produce the final output signal. Signal Display and Actuation: To prove that the sensor does in fact work it is important to compare it to a confirmed reference. Because data is published on the earth s magnetic field and streamed in real time we can use this to compare against the data we are recording and use it to scale the magnitudes and calculate a conversion from recorded voltage to magnetic fields. Our extracted magnetic field signal can be displayed on an oscilloscope and recorded to the audio port of a computer to display on MATLAB. We also have a demo, which is an important means of showing the accuracy and applications of such a magnetometer system. The goal is to build circuitry to remove the offset of the earth s magnetic field from the signal and specifically measure the orientation of a magnet located a distance away in space. Given a known set of maximum magnitudes it is possible to then determine the direction of the magnet. From here we can interpret that signal in analog and create a PWM signal of variable duty cycle to spin a servo to match the orientation of the magnet. In theory this could even be expanded to three dimensions. Active Current Feedback: In order to achieve even more precision in a fluxgate magnetometer, active current feedback may be employed; the sense coil can be driven in order to 5
counteract and cancel out the external field in what is known as null field mode. In this case, the amount of feedback required to null out the external field is used as the measure of the strength of the field. The voltage output of the phase modulation step is fed into an integrator that continues to create a bigger and bigger DC magnetic field until the external field is cancelled out. At this point, the input to the integrator will be 0, and the output of the integrator will be the DC voltage proportional to the strength of the magnetic field. This can then be fed into an ADC after some amplification. The resulting output will help achieve a low-noise, high precision magnetic field sensor. A block diagram of an example active feedback magnetometer circuit is shown below in Figure 3. Figure 3. Example active feedback block diagram Testing and Verification: As we are building our circuitry, we will test each component of our device separately by providing the correct inputs, or inputs simulated to match our desired input, and determining if our output signal matches what we expect. Therefore, we will have verified each block and group of blocks separately before we string them all together into a final device. A fluxgate magnetometer is inherently sensitive enough to pick up the Earth s magnetic field. To test our final device, we will thus compare our live feed from MATLAB to public data from online magnetometers. To verify our final design, we will show that the real time trends of our sensor match the real time trends of online magnetometers. 6
Timeline: Week of April 11 1) Proof of concept: Demonstrate feasibility of toroid and sense coil physical setup with timing and excitation circuitry as well as basic extraction of signal without phase demodulation 2) Order parts: Order permalloy and amorphous metal cores as well as special op-amps and IC s Week of April 18 1) Signal extraction: Add and incorporate phase demodulation block for signal extraction 2) Signal display: Add and incorporate signal display and actuation block 3) Integrate final versions of subsystems 4) Begin testing device Week of April 25 1) Test and debug device 2) Add extra axes and feedback if time Week of May 2 1) Test and debug device 2) Add extra axes and feedback if time 3) Demonstrate completed project Conclusion: Our completed project will demonstrate proper functionality of a one axis fluxgate magnetometer with toroidal drive, sense coil signal extraction with phase demodulation, signal display through MATLAB, and signal actuation through a servo. There are a couple risks associated with our project. Our project is not purely electrical; it incorporates magnetics, so we must be able to make the physics of the toroid and sense coil work in addition to our circuitry. Phase demodulation will be difficult because it is a new topic we have not covered in class and it is also generally hard to achieve phase sensitive detection and demodulation. Furthermore, achieving multiple axes implementation will be 7
difficult due to crosstalk generation. With the skills we gained in 6.101, however, we are confident we will be able to achieve a working final project. 8