How to choose the right sensor

As any engineer knows, working at scale brings a special set of additional development challenges. One key area to consider is what to do when a sensor fails in the field.

There are a number of common pitfalls that can be behind sensor failure and in this article we’ll explore some of the most common issues that can be encountered, as well as broadly applicable ways in which a standardized testing process can be undertaken.

Harsh environments and sensor characteristics

As the famous truism goes, sometimes component failures are actually down to a good component operating in a bad environment, which is perhaps the most obvious – and common – reason for failure. Modern components and sensors have to operate in a huge range of conditions, which in some cases can vary from one extreme to another rapidly, or intensify over a long period of use.

A sensor mounted to a modern car in Canada, for example, might experience ambient temperatures of more than 40 degrees C with high dust levels and low humidity, and within months temperatures below zero, coupled with highly corrosive road salt.

In such harsh environments, sensors become both the chief enablers as well as the weakest links in the control system, they are the most important, yet most exposed part of the system. Therefore, making the correct choice by identifying all the significant characteristics required is an essential step.

Characteristics to consider include: Range of measurement, environmental conditions of operation, repeatability, form factor, resolution, control interface, and any special requirements.

It’s important to remember that sensors are just one part of the system, so connectors and cables like Class 1 Division 1 must also be considered and allowed for.

Sensor modelling

One technique that can be used here is sensor modelling, which uses a mathematical model to characterize the sensor behaviour in a wide range of circumstances. Determining the transfer function of a sensor is an important step, and requires observing the standard signal response of the sensor under a variety of conditions.

Once determined, the transfer function of the sensor can be used as a reference model by the data analysis module, which compares the sensor readings to the predicted readings using the reference model obtained.

Field-failures of any component, whether a pressure sensor or a capacitor in a dedicated circuit tend to be less common for one major reason, the burn-in period. This simple strategy lowers failure rates significantly, and therefore is a vital part of the testing process.

The logic behind it is that components that have packaging, soldering or manufacturing defects regularly fail within minutes or hours of first powering on the device. After this initial few hours of burn-in, component failures typically bottom out and happen at random intervals, until component age begins to take a hand.

Manage change

Another classic generator of component failures is change, and not just environmental change, but most commonly cost saving–generated change. Cost reduction is often introduced once a well-designed product is in production, which can mean that verification plans to quantify the changes are incomplete, which generally leads to disaster.

Although a better strategy would be to iterate a new, next-generation design rather than change the existing one, it is vital that a full verification plan for any changes be constructed, especially taking account of downstream assembly and test operations.

In many ways, choosing the right sensor is actually about eliminating all the wrong choices. Once the application requirements are set down, eliminating technologies and products that don’t fit the requirements may be laborious, but ultimately successful.

Happy hunting!

Martin Keenan is the technical director at Avnet Abacus, which assists and informs design engineers in technological advances including developing applications of pressure sensors, to bring new products to life.