While we have so many sensor options for basic physical parameters, sensing within a target setting often becomes a crucial design challenge.
A large portion of analog circuity and interfacing is devoted to sensors and their front-end electronics. Due to advances in both areas, it’s now possible to have reasonably priced, high-accuracy sensors and their subsystems for many physical parameters such as distance, motion, mass, pressure, temperature…the list goes on and on.
Despite this, there’s a dilemma in many cases, as the sensing challenge is not with the sensor and its interface but instead with the in-place sensing situation. Sensing something as common as temperature, for example, in a room can be compared to doing so at a jet’s exhaust. There are dozens of radically different techniques for temperature sensing, including (but not limited to) using a solid-state sensor, a thermocouple, an infrared sensor, or even measurement of thermal expansion along with countless actual implementations of each one.
I recently came across a relevant and timely example that clearly shows the challenges of sensors versus sensing, and demonstrates that sensing issues don’t apply only to esoteric situations. Consider the important topic of measuring air flow in a car’s passenger cabin with two or more occupants. Especially these days, you want to ensure good air flow when people are in close proximity, for well-known reasons.
The obvious question is this: should you open one window, or several windows, and by how much? The likely intuitive answer of “open all the windows” may be wrong, and even if it’s the best technical solution, it may be impractical or undesirable.
If you want to restrict it to just two windows, for example, which ones you open? Is it a function of where the passenger or passengers are seated? Should it be the two front windows? Maybe use the driver’s window and the rear-right window (diagonally across from the driver’s window) or perhaps just the passenger window and the rear-left window? What’s the impact of having the car’s vent setting in different positions? How about the benefits of not using the fan mode or even air conditioner?
All good questions, and they should be easy to answer in a testbed with a wind tunnel in a car mockup and some passive “dummies” representing the driver and passengers. However, it turns out that it’s not an easy situation to instrument for a variety of reason. Yes, there are many available airflow sensor instruments such as the Center 332 Hot-Wire Anemometer from Center Technology (Figure 1). This handheld unit with extendable separate probe measures air velocity from 0 to 25 meters/second (equivalent to 0 to 5000 feet/minute) and air flow (volume) from 0 to 106 cubic meters/minute (about 8.5 × 108 cubic feet/minute) with accuracy of ±3% reading.
Figure 1 The Center 332 Hot-Wire Anemometer from Center Technology provides measurement of air velocity and air flow volume over a wide range; its extended sensor minimizes disturbances to the air flow being measured. Source: Center Technology
But having a good sensor or instrument alone is only part of the solution. This has been made clear in a pair of related articles—one in AAAS Science Advances titled “Airflows inside passenger cars and implications for airborne disease transmission” and the other in Physics Today titled “The air we breathe in a car”—which discuss the challenges related to assessing airflow in cars. Th authors concluded that given the many variables of the arrangement, and how and where you measure the airflow, it’s a problem that does not lend itself to real-world physical instrumentation as much as it does to modeling and simulation.
I don’t have a problem with that approach, as modern simulation tools can be very good. However, there is a potential problem with nearly all such simulations: they are heavily dependent on the fidelity of the underlying models. In this case, I don’t know how accurately you need to model the surfaces and interior geometry of the car for this project. Will a slight variation in passenger cabin dimensions—after all, every car is a little different—make a big difference in the results? Can you do a meaningful sensitivity analysis on how the simulation results will be affected by simplifications in the model, as shown in Figure 2?
Figure 2 Air flow patterns in a car are complicated and dependent on many factors, including cabin size, geometry, vehicle speed, occupancy and number and location of open windows. Source: Physics Today
Fortunately, not all researchers are committed solely to models and simulation results. There’s a revealing study of air-change per hour (ACH) in cars from The Journal of Exposure Science & Experimental Epidemiology titled “Air change rates of motor vehicles and in-vehicle pollutant concentrations from secondhand smoke.” Here, the authors ran tests under a variety of conditions using four different real cars. These researchers were serious in the endeavor: they went well beyond basic airflow sensors and added an instrument-grade monitor to measure carbon-monoxide (CO) concentrations, as well as an optical-scattering monitor to measure respirable-particle concentrations.
It makes sense that a good model followed by simulations is the best way to go, if there is reasonable agreement between the simulation and real world; in the case of the car modeling versus measured airflow, I’d be very impressed with agreement with even a 10% difference. It’s the same as when I see a circuit simulation: it’s nice to see the performance characterized with such apparent precision and reams of data, but I always feel more confident about those perspectives if an actual prototype in near-final configuration is also tested and comes to within about 5% or 10% of the simulation results.
Don’t be fooled: sometimes, problems with sensing of real-world parameters aren’t due to the application scenario, but the sensors. Think carefully about what sensor arrangement you need, how many sensors, their locations, their impact on the test itself, and other relevant factors. In many cases, a good 3D simulation may be a better option if you have solid initial numbers, but only if you can develop a viable model—and that’s a big “if.”
Have you ever encountered a situation where choosing and interfacing the sensor was the easy part, but the situation made the use of the sensing problematic?
This article was originally published on Planet Analog.
Bill Schweber is an electronics engineer who has written three textbooks on electronic communications systems, as well as hundreds of technical articles, opinion columns, and product features. In past roles, he worked as a technical website manager for multiple EE Times sites and as both Executive Editor and Analog Editor at EDN. At Analog Devices, he was in marketing communications; as a result, he has been on both sides of the technical PR function, presenting company products, stories, and messages to the media and also as the recipient of these. Prior to the marcom role at Analog, Bill was Associate Editor of its respected technical journal, and also worked in its product marketing and applications engineering groups. Before those roles, he was at Instron Corp., doing hands-on analog- and power-circuit design and systems integration for materials-testing machine controls. He has a BSEE from Columbia University and an MSEE from the University of Massachusetts, is a Registered Professional Engineer, and holds an Advanced Class amateur radio license. He has also planned, written, and presented online courses on a variety of engineering topics, including MOSFET basics, ADC selection, and driving LEDs.