Temperature sensing and control represent an integral part of a wide range of applications, ranging from a nuclear reactor to an everyday toaster. The control signal based on the temperature measurement can be as simple as an ON/OFF switch. Slightly advanced control systems might involve a temperature feedback mechanism for fan speed or valve flow rate control. Even in applications where the temperature sensor is not the primary sensor, it can be used as a secondary sensor to compensate such primary sensor output as pressure, load cell, or pH.
Currently, these systems are designed using multiple chips. This creates variety of challenges for system designers such as finding optimum devices (which requires evaluating different price/performance tradeoffs), sourcing from multiple vendors, and combining their collateral to design complete system. In this article, we’ll focus on how the efficient use of a System on Chip (SoC) can simplify these challenges and enable the development of high-performance and scalable product platforms.
For example, consider a traditional implementation of temperature sensing with a thermocouple that has a thermistor for cold junction measurement and 4-wire fan control as shown in Figure 1.
Traditional implementation involves multiple devices to bias and measure the sensor signal, process the measured signal and generate a control signal. This involves multiple decisions, designs and evaluations, which can be observed in a generalized signal chain block diagram shown in Figure 2.
The block diagram shows example devices for each block and the decisions involved in choosing each of the devices. For example, temperature measurement can be performed with sensors such as, but not limited to, thermistors, thermocouples, RTDs and diodes. The analog interface can be a combination of voltage reference, Op Amp and ADC or an IDAC and ADC. One or more of the sensors are chosen based on the temperature range, accuracy and cost requirements. A combination of the analog interface devices are chosen based on requirements like sensor of choice, accuracy, cost and other requirements.
This isn’t a simple task. As an example, one of the large semiconductor companies offers more than 900 part numbers just for op amp. You need to go through this type of extensive part selection process for every element, which is a daunting task, and you sure will appreciate some help in this area. It is very likely that you will be designing multiple products to have a good portfolio to meet different needs of your customers, as shown in Figure 4. To do this, you need to repeat part selection process for different price/performance tradeoffs. It is highly likely that devices you identified to best serve your portfolio aren’t pin compatible. It means you will need multiple PCB layouts to cover your portfolio.
Once part selection is finished, the next step is to build prototypes to validate the design. As you are sourcing parts from different suppliers, you need to work with collateral that deals with different part of your system design challenges. As an example, evaluation boards and application notes from analog supplier usually focus on analog design challenges and don’t provide code for MCU, which deals with calibration and linearization algorithms. This tends to be true even if analog and MCU are being sources from same supplier.
You are well aware of these design challenges but the natural question is what is the solution? Ideally what you want is a single chip solution that can interface to variety of temperature sensors you are using. You want it to be configurable through software and want multiple pin compatible options for price/performance tradeoffs. If above features are available, you can design a single PCB layout that can support multiple end product variants for you. It will be even better if same supplier provides collateral that covers complete end-to-end system design for these sensors. This way, you don’t need to combine collateral from multiple sources to create your end system.