Get Help On Your Device Design

Device Design: Strategies to Improve Sensor Accuracy and Reliability

post_banner

Wearables and IoT devices have become significantly more advanced in the last decade, and a big part of that is improved sensor design and accuracy. It wasn't long ago when these devices were niche, and consumers were hesitant to buy them due to simple technology and inaccurate readings.

Now, we have consumer-level wearables that are sensitive enough to detect the presence of atrial fibrillation and other potentially life-threatening health conditions.

While well-designed software does much of the heavy lifting in alerting users of such issues, none of that would be possible without accurate sensors. Improving sensor accuracy is an ongoing effort, but there's no denying that better accuracy paved the way for the widespread wearable and IoT adoption we have now.

For some perspective, consumers bought a little less than 9 million wearables in 2012. That's a far cry from the 320 million wearables experts estimate will ship worldwide in 2022.

Accurate sensors are crucial in modern devices, and they will continue to drive innovation in the future.

Understanding Sensor Accuracy


There are a few different measurements to consider when designing devices with sensors. While many use the terms "accuracy" and "precision" interchangeably, they're unique factors contributing to a sensor's overall performance.

Accuracy is how close the measured data is to the actual value.

For example, a thermometer that measures 74 degrees when the correct temperature is 74.4 degrees is fairly accurate. While not perfect, it's close to the real-world value.

The goal for most sensors is to get as close to 100 percent accuracy as possible.

A sensor's accuracy is one of the most critical details to test. That's especially true when developing medical devices or anything involved with healthcare monitoring; however, in some applications measuring change is more important than accuracy.

Precision would then be important.

Precision refers to how close multiple sensors or readings are to each other. A good example here would be having numerous thermometers picking up measurements. One might be 73.9, another could be 74, and another could be 74.2. Those sensors are relatively precise because the readings are tight.

You could also view precision as the degree of repeatability. A precise sensor would have nearly identical readings for multiple measurements, proving that it can reproduce reliable results.

Precision and accuracy are similar, but they're not exclusively dependent on one another. That's why it's crucial to understand the difference and test for both.

A sensor can be accurate but not precise, meaning it can't reproduce the results enough to be reliable. On the opposite side of the coin, a sensor could be precise but nowhere near accurate to the actual value.

Highly Successful Engineering Design Projects

Common Sensor Errors


Sensors can encounter many different issues throughout the life of the product. The quality of the sensor itself matters, and components that cost more tend to be more reliable over time. However, even premium sensors can produce errors affecting data acquisition accuracy.

🚫 Linearity and Drift


Linearity refers to the response of a sensor's output over the entire range of possible measurements. Ideally, a sensor will change the same amount throughout its range, such as a temperature sensor changing 10 millivolts for a 1-degree change.

Some sensors are highly nonlinear, such as a thermistor temperature sensor. Typically, a thermistor will change rapidly at cold temperatures and much less at high temperatures. It needs to be made linear with a calculation, typically done in software.

Many sensors are approximately linear, but none are perfect. The linearity error is one factor in accuracy and precision.

Some sensors gradually change their output over time or temperature for the same input. Also known as "drift," this phenomenon can be a product of sensor degradation or irregular calibration.

Temporary change due to variations in temperature, humidity, or other environmental factors is also referred to as drift. Drift is a limitation to any sensor and is another factor in accuracy and precision.

🚫 Offset Errors


Offset errors occur when the sensor's output at zero input is not zero. For example, a pressure sensor should read zero Volts when the pressure is zero.

Typically, an offset needs to be corrected, usually in software. The offset is not identical in each device, and the offset can drift with time, temperature, or other environmental factors. Offset is also a factor in accuracy and precision.

🚫 Noise, Artifacts, and Aliasing


Of course, sensors can also encounter errors due to noise and issues with signal processing. Sensors are responsible for data acquisition.

Other processing components perform some capabilities that people think actually happen on sensors. That's not always the case, but it's true in many consumer-level devices.

For example, activity trackers utilize accelerometers to detect movement. However, they don't calculate how many steps you took or calories you burned. That all happens in software, often on your smartphone or the cloud.

Noise is rapid fluctuations in the signal that are not caused by fluctuations in the input being measured. The noise limits how small a change can be detected.

Data acquisition (DAQ) systems interface with the sensors so that software can use the information and handle the functions users want. Aliasing can ruin data accuracy as it goes through the DAQ (refer to the link for a full discussion of aliasing), which causes distortion and artifacts in the data.

🚫 Signal Conditioning

Signal conditioning is an electronic circuit that corrects some errors, such as offset or linearity. It also includes amplification, and with many sensors, signal conditioning includes proving power to make the sensor operate.

The Importance of Sensor Calibration


There are ways to make sensors more accurate. The most straightforward technique is to implement calibration.

Calibration helps to ensure the accuracy and precision of the sensor output. It is often done at the factory when the device is made.

Drift errors can only be corrected by repeated calibration over time or recalibration. Few wearable devices get recalibration, as it is very inconvenient.

Often calibration is done at several points from the low to the high end of the range of a sensor. This is multi-point calibration. Usually, devices can be made accurate enough with a calibration at a single output or single-point calibration.

Calibration can be crucial to maintaining long-term sensor accuracy.

For some devices, it's a requirement. FDA regulation says that medical devices requiring calibration must go through calibration regularly. Not only that, but the calibration chain has to be traceable to the ultimate reference, usually the National Institute of Standards and Technology (NIST).

Guide to Developing Product Requirements for Medical Device


Gaining Accuracy with Sensor Fusion


Another technique for improving accuracy across the board is to have sensor fusion.

With sensor fusion, you're not just using one sensor per data point. You're using several at once to merge data and get more accurate outputs. Essentially, it's using one sensor's output to validate another's.

For example, when a wearable device measures the temperature of the skin, usually the desired measurement is core body temperature, which is often different.

By measuring the ambient temperature and the activity of a person, it can be determined when the skin temperature is closer to the core body temperature.

Sensor fusion wasn't a realistic option over a decade ago due to the price of individual components. But today, sensors are far more affordable, and smart software can now fit on a wearable device.

As a result, device makers are taking advantage of sensor fusion.

Thanks to increased accuracy and lower sensor costs, we're starting to see wearables and IoT become more prevalent. That's not just in the consumer market but also in industrial and commercial applications.

Design Your Device for Sensor Accuracy


Sensors are complex components that act as the heart of your devices. It doesn't matter what kind of wearable or IoT you're creating. Its success relies on its ability to take accurate measurements throughout its life.

While the sensor's quality impacts accuracy and precision, so does device design. Getting things right from the beginning makes a difference, improving performance and functionality across the board. Sensors are mostly analog, and Voler has highly experienced analog and digital circuit designers.

Do you need help designing a sensor-based device? Turn to Voler Systems. We can take care of the product development cycle from concept to prototype.

Let us use our sensor and electronic design expertise to bring your ideas to life.

Whether your product is destined for medical, industrial, or consumer applications, we're committed to ensuring that it's accurate and reliable.

Contact Voler Systems Today

Share:

TELL US ABOUT YOUR NEXT DESIGN PROJECT

Do you have a question about our services, pricing, samples, resources, or anything else?

Contact Us Now

Related News

Finding some "sense" in 2018 | Voler Systems

Sensors will continue to be a driving force for innovation. Low-cost sensors are...

Read More

Discrete Data Collection: Understanding Common Sensors in Wearables

Wearables are one of the fastest-growing segments in tech. What was once a niche category...

Read More

How to Design Wearable Sensors That Work for Users

The rise and adoption of wearable technology completely changed the healthcare industry...

Read More

Interested in Learning More? Contact Us Today!