Non-volatile data logging for Industry 4.0

With the advent of Industry 4.0, factories are becoming increasingly intelligent and connected. Mechanical equipment in smart factories can use real-time data from connected wireless sensor nodes to predict possible failures in advance and notify the control system to take corrective measures to avoid unexpected system downtime. The accumulated data can be used to improve predictive analysis and achieve better preventive maintenance of machines.

The ability to reduce downtime is an important factor in managing industrial facilities. Currently, downtime is difficult and very expensive to predict. For example, downtime in the average automotive manufacturing plant can cost as much as $22,000 per minute or $1.3 million per hour (Source: Advanced Technology Services, 2006: “Automotive Industry Downtime Cost: $22,000 per Minute: Survey.”).

With the advent of Industry 4.0, factories are becoming increasingly intelligent and connected. Mechanical equipment in smart factories can use real-time data from connected wireless sensor nodes to predict possible failures in advance and notify the control system to take corrective measures to avoid unexpected system downtime. The accumulated data can be used to improve predictive analysis and achieve better preventive maintenance of machines. These advances are designed to improve factory operational efficiency and reduce overall downtime.

Actionable data, the foundation of Industry 4.0, enables everything from real-time sensing to predictive analytics. Therefore, it is critical to continuously and reliably record data, especially in the event of a failure, as this data is often critical information. In addition, the amount of data that needs to be recorded is expected to continue to increase. We will need continuous data collection not only on traditional industrial systems, but also on the thousands of connected sensor nodes that will be spread throughout the smart factories of the future.

To overcome these challenges, next-generation industrial systems require high-performance non-volatile memory to ensure "zero data risk" and reliable data backup during normal operation and system failure. To ensure the reliability of this data record, Industry 4.0 non-volatile memory needs to support fast write, real-time non-volatility, and nearly 100 trillion cycles and nearly unlimited read/write endurance.

Challenges of Industrial Data Logging

Industrial control systems include automation, energy management, process measurement, and test measurement, all of which require high-performance non-volatile data logging memory. In all of these market segments, non-volatile memories are used to continuously record real-time system data. They must also be able to instantly capture real-time system status data in the event of a power outage or system failure. Non-volatile data logging memory for industrial control systems and wireless sensor nodes needs to meet different requirements. We will explore the challenges and unique requirements of industrial control systems and wireless sensor nodes through two examples: programmable logic controllers and IoT sensor nodes.

Figure 1: Programmable Logic Controller (PLC) Block Diagram

Programmable logic controllers (PLCs) used for industrial automation are common in industrial control systems. In PLCs, real-time system data captured by non-volatile data logging memory is used to detect and repair faults and prevent future failures. In addition, non-volatile data logging memory captures the last system state before power was lost. This data is critical to ensure that the PLC and all connected machines restart in a safe operating mode when power is restored. Without this capability, other machines and people in the surrounding environment are exposed to potential risks.

In next-generation industrial control systems such as PLCs, there is a need to reduce the number of microcontroller pins used to interface with external memory. This demand is driving the transition from parallel to low-pin-count serial interface memory. That is why memory manufacturers are developing low-pin-count memory for industrial applications. For example, Cypress' Excelon-Ultra is dedicated to industrial control systems and provides a low-pin-count 108-MHz QSPI interface.

Non-volatile memory is superior to the battery-backed SRAM commonly used in industrial control systems due to the high reliability brought by the removal of the battery. In addition, non-volatile memory reduces the bill of materials cost by replacing the multi-component subsystem (SRAM + battery + power management controller) with a single chip and avoids the maintenance associated with the cost required to replace the battery.

Figure 2: IoT sensor node module diagram

Wireless IoT sensor nodes are the eyes and ears of a smart factory. As mentioned above, sensor nodes can continuously monitor system and environmental parameters and then notify connected machines or control systems to take corrective actions when necessary.

Wireless IoT sensor nodes present different challenges than industrial control systems. Sensor nodes are small in size. In addition, they are often located throughout a smart factory, including in remote or difficult-to-access locations. As a result, they are often powered by batteries or through energy harvesting.

Therefore, sensor nodes require non-volatile memory in a very small form factor to continuously record real-time system data. They must be able to reliably complete this task throughout the life cycle of the sensor node and use as little power as possible. For example, Cypress's Excelon-LP uses a small GQFN package of approximately 10 square millimeters and provides multiple power saving modes, including sleep, deep node and standby, allowing developers to maximize battery life.

To further reduce the size of wireless IoT sensor nodes, a single-chip approach for code storage and data logging can be implemented using non-volatile memory. In sensor nodes, the code size required to measure and collect data is usually small compared to the amount of memory required to store the data. Therefore, having a separate code memory may result in insufficient memory utilization, and a single-chip approach would be more efficient.

The key challenge is to have a nonvolatile data logging memory that is flexible enough to partition the storage of code and data according to the application requirements. The memory used to store the code must ensure that the system cannot accidentally write to the storage area used to execute the code. Therefore, in order to meet the requirements of the single-chip process, the nonvolatile data logging memory needs to have memory protection functions. For example, block protection prevents accidental writes to a range of addresses defined by the developer. This enables a single memory to support real-time data logging while storing and protecting the code.

Ferroelectric Technology

The key to non-volatile data recording memory is ferroelectric technology. Ferroelectric technology combines the high performance and byte addressability of RAM with non-volatile data storage. Ferroelectric technology for non-volatile memory has a memory cell using a lead zirconate titanate (PZT) thin film. When an electric field is applied, the central atom in the PZT crystal changes position. The two positions of the central atom store one bit as a binary state of the memory. When the power is interrupted, the atomic position is retained, thus protecting the data. Its data reliability is also very high, and it can safely store data for up to 100 years without any backup power supply.

The non-volatile data logging memory is very efficient. It consumes 200 times less energy than serial EEPROM and 3000 times less than NOR flash. The technology also provides high data reliability with a read/write endurance of 100 trillion cycles (1014). In contrast, floating gate technologies such as flash and EEPROM wear out in as little as 106 cycles, making them unsuitable for frequent system data capture.

Non-volatile data logging memory also ensures “data-free” operation of industrial systems. The ability to instantly store data protects the system from data loss when power is lost. Technologies such as EEPROM typically have a page write latency of 5-10 milliseconds, putting important last-minute system data at risk.