Design Like a Physicist


Physics 398DLP, Spring 2021

Online via Zoom, Friday afternoons, 1 pm - 5 pm

(I'll email you instructions for how to connect.)

3 credit hours


Dean Bashir has asked faculty to distribute to students the information linked here.

Introduction

Those of us whose careers have included both teaching and research have long found that our students undergo a dramatic transition in ability between their undergraduate years and the end of their first year of graduate school. As undergraduates they would attend lecture-based classes and master course content by listening to their professors and slogging through weekly problem sets. (You know what this is like!) By the end of the semester, most of the class would understand most of the material, but would find it difficult to integrate it into a coherent picture of, say, classical electrodynamics. And a semester after a course had ended, most students would not have retained their mastery of the topic.

But after a year of graduate school—during which students would work on difficult material without the distracting edge effects of 50-minute class periods—their competence at navigating confusing subjects and difficult problems would increase enormously.

Project physics

Many of us suspect that teaching physics to undergraduates in a manner that more closely resembles graduate education might be beneficial. One aspect of this is to offer project-based courses, in which students would learn physics by mastering what they needed to complete tasks that were more like research projects than was usually true in undergraduate instruction.

You've already had some experience with this instructional mode if you've taken Physics 298owl from me. It's different from fighting to stay awake for an hour in lecture, then sifting through the wreckage to extract what you need to do the homework assignment!

This course

In Physics 398DLP you will be performing the one-semester analog of a PhD research thesis: defining a measurement to be performed, designing and building an instrument that might be capable of recording data necessary for the measurement, testing your device, doing the field work to record valid data, then analyzing the data to form supportable, reproducible conclusions. Along the way you will report on your progress, both through informal presentations to the class and, at the end of the term, in a concise, more formal paper intended for an external audience.

If all goes well, you'll find this so captivating that it will be hard to put your work aside to attend to your other academic obligations. I suspect it is this strong engagement with a project that drives the transition from an undergraduate level of skill to the expert mastery typical of graduate researchers.

Prerequisites

You must already know how to program. If you've learned to code in python or C/C++, or Java, or some other language, you'll be fine. A B- or better in CS 101, CS 125, or Physics 298owl are suitable prerequisites. It's also fine if you've learned on your own. But if you've never programmed before, or did poorly in an intro CS course, you should delay enrollment in Physics 398DLP until after you've done some coding.

You must have a basic working knowledge of introductory physics at the level of Physics 211 and Physics 212. More is better, though not necessary.

We are not building robots

Physics 398DLP is not a course in building cute robots for the sake of learning to build robots. That would be an engineer thing, and we are physicists, not engineers. We are going to tackle measurements that—if they prove feasible—might make our corner of the world a little bit better. If we did build a cute robot, it would be to accomplish a significant end, for example recognizing the onset of a potentially catastrophic fall by an elderly person.

In Physics 398DLP you'll construct a hand-held device loaded with inexpensive sensors that are interrogated by an onboard microcontroller—a small computer larded with additional features such as timers and analog-to-digital converters—and write the data acquisition software necessary to perform the measurements associated with your project.

You'll assemble a prototype on a breadboard, construct a final (electrically equivalent) version on a printed circuit board, use a 3D printer to build a case for it, do field work, then write analysis code to understand what conclusions can be drawn from your data. You'll write a report presenting your results and justifying your conclusions, publish it to the web, and perhaps send it to the appropriate recipient—the Illinois Department of Transportation, for example—and request a meeting to discuss your findings.

Some possible projects

The intellectual tradition in physics is for researchers to build their own instruments (buying off-the-shelf parts when available), ultimately creating sophisticated devices to perform the measurements that will tell us about the physics we are researching. It is not like this in all fields; my wife's background is in bio-inorganic chemistry, and she would assemble reactors from stock components, then run reaction products through spectrometers built by vendors like Varian and Hitachi. So you'll be following the physics tradition, and you will be working in close (intellectual, though socially distanced) collaboration with three other students.

Some of the projects are probably best imagined as feasibility studies that might inform the design of a more definitive future measurement. We will see how it goes! Here are some that I have in mind. You are free to suggest other possibilities, though I reserve the right to veto anything that I feel is too difficult or too expensive. We'll also need to discuss how to do these while avoiding direct face-to-face interactions among group members.

  • • Zoom studies: degree of timing mismatch between video and audio; variations in source-destination latency. Can a hearing-impaired person lip-read during a Zoom conference? Can a chamber orchestra use Zoom for rehearsals?
  • • Multiple-head IR non-contact thermometer, interrogated by a single processor; consider issues with expansion to a 100-unit system, and skin-tone-related systematic effects.
  • • How well does an N95 mask fit? Sensitive differential measurements of pressure inside and outside an N95 mask during the wearer's respiration cycle.
  • • Dried blood spot analysis: how might one do a non-destructive determination of the amount of blood in the sample? How do the results depend on temperature and humidity? (See the Wikipedia article "Dried blood spot" as well as material on the NIH web site.)
  • • Radio-linked environmental monitoring backpacks: T, P, RH, VOC, PM2.5, CO, methane, ambient acoustic environment; GPS, RTC, microphones microSD; WiFi-capable base stations; forward-facing camera imagery. (Ask me about the "informal electronics recycling industry" in Ghana!)
  • • Airborne particulate concentrations in agricultural settings (outside/inside tractor and/or combine cabins)
  • • Precision real-time cross-timing of multiple devices for use in real-time acoustic adjustments in outdoor music venues
  • • Four-microphone doppler vector anemometer for outdoor performance venue acoustic tuning
  • • Effectiveness of particulate filtration by homemade masks (styrofoam heads, fishtanks pumps, PM2.5 units? Human air consumption is about 8 liters per minute. Fishtanks pumps cost about $30 and can do 40 liters per minute.)
  • • Multiple radon detectors in a single house. (Perhaps: use Radio Feathers, each with a TTL camera and a BME680, and also a WiFi module on one to connect to a home WiFi network. Possible detectors: Tjernlund 9873739 Radon Inspector 3 Detector, $100.)
  • • Airborne particulate concentrations in institutional kitchens
  • • Riding the rails II: using sensor data from the first car in a passenger train to predict anomalous accelerations inflicted on subsequent cars
  • • Measurement of a runner's tendencies towards pronation or supination
  • • Radio-linked bovine methanogenesis monitors
  • • Large amplitude, high frequency vibrations inflicted on MTD bus passengers.
  • • Distributed sensor network studies: latency and stability of 915 MHz LoRa radio-linked data communications vs. distance, broadcast power, and environmental clutter.
  • • Very short term stability of propagation time of signals across cell phone networks. Could we have used cell phones reporting to a base station to triangulate the location of the chemical explosion in Beirut? Could we make a widely distributed network of cheap cosmic ray detectors to construct an "extended air shower array" with milliradian pointing accuracy?
  • • Hospital ventilator insertions: could we use basic fluid dynamics to allow determination of the gas flow rate to a ventilated patient purely on the basis of sensitive pressure measurements? We are well stocked with BME680 T/P/RH/VOC devices.
  • • Digital Agriculture 1: studies of (inexpensive) drone-borne thermal imaging methods in agricultural settings. We have a cool DJI drone and a low-resolution IR camera. (one or more members of your group will need to apply for an FAA drone pilot's license.)
  • • Digital auscultation: propagation speeds of cardiac and other human-body-produced sounds. (You'll need to contact the Institutional Review Board for permission to measure human stuff.)
  • • Porosity of different dust mask materials to airborne particulates. We have a handful of laser-based airborne particulate sensors. With some tubing and a fish tank air pump we might be able to measure the effectiveness of different materials at blocking airborne particulates.
  • • Directional selectivity of a two-electret-microphone-per-ear hearing aid.
  • • Noise and pressure profiles in the vicinity of wind turbines
  • • Digital Agriculture 2: an initial feasibility study of image processing algorithm(s) to distinguish corn rootworm beetles from other agricultural grit caught on glue traps, using JPEG images captured by an inexpensive camera

The style in which we will work

"DLP" stands for "Design Like a Physicist." That's a reasonably descriptive term for how we will go about things. Here is what I mean. If you took Physics 298owl from me you'll remember that I had you hand-code a lot of algorithms—integrators, Fourier transforms—that could also be found in professionally produced libraries. For pedagogical purposes, I had you reinventing a lot of wheels.

That's not how I've gone about my own research. If there's a pre-coded numerical algorithm that I can use, I'll appropriate it, generally putting proper attribution to its source in comments in my own code. If there's a circuit I need that's described in an engineering web site, I'll use it, identifying its source on my schematic diagram.

You will keep track of your efforts in an electronic diary in which you describe your work, useful revelations, and calculations. You should put notes about techniques you find (or invent) into your diary so you can find them later. The diary should be cumulative, rather than something you close off at the end of each class meeting. You are required to upload a PDF version of the file to the course directory before each class.

Real physicists are fearless, and sometimes confused

You will be working with things for which your understanding will often be really blurry. That's OK, and in fact that's the usual state of things in research. Taking the time to understand every last detail about an IDE is a waste of time: it is better to focus your efforts on getting by, on muddling through. You will get more done this way than you would if you spent the time to understand everything completely. There is too much to do, and far more interesting things to consider than the arcane details of SPI and I2C interfaces. You will want to understand them well enough to work with them, that's all.



Photo credits: Cockroft-Walton linked to the course title is from https://www.sciencefriday.com/articles/chasing-time-machines-a-photographer-turns-high-energy-physics-into-art/. The image below the page header is a breadboarded version of a Physics 398DLP data logger. About the rose theme: ask me about it in class some day.