We live in an analogue world. To give the electronic systems we design the ability to detect changes to what is around them we need analogue sensors. But the processing that sits behind those sensors is in almost every case going to be in the digital domain.
Although there are some processing techniques that may be more energy efficient in the analogue domain, conversion to digital provides the greatest flexibility and opens up the true potential of the IoT.
It is not just the conversion to digital numbers that unlocks the power of the IoT; it is the ability to tag those values with additional metadata. The value of a pressure sensor, for example, is only useful if we know when and where that reading was taken so we can correlate it with information from other sensors. By working with the data in combination with the metadata we can gain a better understanding of what is happening around us and then take action. For example, pressure sensors are now being incorporated into smartphones. One of the applications is to use changes in environmental air pressure to confirm the position of the device in the world.
There are multiple markets that need to collect, interpret, process and deliver digitised data from the real world. Connected home appliances can combine information on air pressure with temperature and other environmental signals to optimise the pumping action of heating and air-conditioning systems. In the industrial environment, time-series pressure data can reveal important information about the operation of the system. For example, in oil drilling the changes in pressure over time can reveal leaks and other problems. The information the sensors provide can be used by remote servers to update preventive maintenance applications, which are increasingly making use of machine-learning techniques to identify problems before they become critical. The ability to provide both data and metadata, such as the time and pump conditions during each reading, is vital to ensure each of the target systems understands the proper context.
There are other more subtle advantages to digitising data. Through Moore’s Law scaling, we have seen the energy required per calculation fall year on year. And we can look forward to another decade’s worth of digital scaling at least. We can also take advantage of improvements in circuit innovation and digital design to reduce the voltages at which the processing takes place. As power consumption has a quadratic relationship to supply voltage, every fraction of a volt our circuits are able to shave off leads to a much greater benefit in energy saved.
The question then is where does the analogue-to-digital conversion take place? Traditionally, it has been the job of the system designer to take an analogue-output sensor and add the various signal-conditioning, amplification and conversion components to their board design. This provides them with the flexibility to match the sensor’s operation to the needs of their application. But many designers lack the time to optimise the interface fully, particularly in terms of energy consumption.
Very often, there is a complex relationship between the signal conditioning op-amps and the front-end amplifiers that feed into the analogue-to-digital converter (ADC). Sensor designers have an intimate understanding of the signals their products are designed to handle and can provide great insight into the most efficient way to convert the output into the digital domain. By bringing the digital conversion into the sensor module itself, they can provide this skill and understanding to the customer. The result is faster time-to-market for the customer.
There are other advantages in using a sensor with built-in digital output. The close integration of sensor and ADC reduces the risk of electrical noise coupling into the interface, ensuring a higher-quality output. A less obvious advantage is the impact on microcontroller performance. A sensor with integrated digital output is simpler to deal with at the firmware level. With an ADC-based interface, the firmware needs to be able to set up conversions and move the resulting data into a memory buffer within a narrow time window to minimise the adverse effects of timing jitter. A digital sensor may use a standard communications bus such as I2C to communicate with the host microcontroller. The sensor module will buffer the recorded data until the host microcontroller is ready to receive the data, resulting in less stringent timing requirements. That can translate into lower system cost and faster time to market.
So, despite the world being an analogue place, getting signals into digital form as close to the source as possible makes a lot of sense.
Are you developing a new device or application that requires pressure measurement? Read our design guide on choosing pressure sensors and explore a range of additional pressure sensor resources to help you find the right solution for your design. Alternatively, if you’re ready to take the next step, contact our team of technical specialists who are on hand to discuss your design and assist with product selection.