Power management integrated circuits (PMICs) are an essential component in the design of any power supply. Their main function is to integrate several complex features, such as switching and linear power regulators, electrical protection circuits, battery monitoring and charging circuits, energy-harvesting systems, and communication interfaces, into a single chip.
Compared with a solution based on discrete components, PMICs greatly simplify the development of the power stage, reducing the number of components required, accelerating validation and therefore the design’s time to market. In addition, PMICs qualified for specific applications, such as automotive or industrial, are commercially available.
In industrial and industrial IoT (IIoT) applications, PMICs address key power challenges such as high efficiency, robustness, scalability, and flexibility. The use of AI techniques is being investigated to improve PMIC performance, with the aim of reducing power losses, increasing energy efficiency, and reducing heat dissipation.
Achieving high efficiency
Industrial and IIoT applications require multiple power lines with different voltage and current requirements. Logic processing components, such as microcontrollers (MCUs) and FPGAs, require very low voltages, while peripherals, such as GPIOs and communication interfaces, require voltages of 3.3 V, 5 V, or higher.
These requirements are now met by multichannel PMICs, which integrate switching buck, boost, or buck-boost regulators, as well as one or more linear regulators, typically of the low-dropout (LDO) type, and power switches, very useful for motor control. Switching regulators offer very high efficiency but generate electromagnetic noise related to the charging and discharging process of the inductor.
LDO regulators, which achieve high efficiency only when the output voltage differs slightly from the input voltage to the converter, are instead suitable for low-noise applications such as sensors and, more generally, where analog voltages with very low amplitude need to be managed.
Besides multiple power rails, industrial and IIoT applications require solutions with high efficiency. This requirement is essential for prolonging battery life, reducing heat dissipation, and saving space on the printed-circuit board (PCB) using fewer components.
To achieve high efficiency, one of the first parameters to consider is the quiescent current (IQ), which is the current that the PMIC draws when it is not supplying any load, while keeping the regulators and other internal functions active. A low IQ value reduces power losses and is essential for battery-powered applications, enabling longer battery operation.
PMICs are now commercially available that integrate regulators with very low IQ values, in the order of microseconds or less. However, a low IQ value should not compromise transient response, another parameter to consider for efficiency. Transient response, or response time, indicates the time required by the PMIC to adapt to sudden load changes, such as when switching from no load to active load. In general, depending on the specific application, it is advisable to find the right compromise between these two parameters.
Nordic Semiconductor’s nPM2100 (Figure 1) is an example of a low-power PMIC. Integrating an ultra-efficient boost regulator, the nPM2100 provides a very low IQ, addressing the needs of various battery-powered applications, including Bluetooth asset tracking, remote controls, and smart sensors.
The boost regulator can be powered from an input range of 0.7 to 3.4 V and provides an output voltage in the range of 1.8 V to 3.3 V, with a maximum output current of 150 mA. It also integrates an LDO/load switch that provides up to 50-mA output current with an output voltage in the range of 0.8 V to 3.0 V.
The nPM2100’s regulator offers an IQ of 150 nA and achieves up to 95% power conversion efficiency at 50 mA and 90.5% efficiency at 10 µA. The device also has a low-current ship mode of 35 nA that allows it to be transported without removing the battery inserted. Multiple options are available for waking up the device from this low-power state.
An ultra-low-power wakeup timer is also available. This is suitable for timed wakeups, such as Bluetooth LE advertising performed by a sensor that remains in an idle state for most of the time. In this hibernate state, the maximum current absorbed by the device is 200 nA.
Another relevant parameter that helps to increase efficiency is dynamic voltage and frequency scaling (DVFS).
When powering logic devices built with CMOS technology, such as common MCUs, processors, and FPGAs, a distinction can be made between static and dynamic power consumption. While the former is simply the product of the supply voltage by the current in idle conditions, dynamic power is expressed by the following formula:
Pdynamic = C × Vcc2 × fsw
where C is the load capacity, VCC is the voltage applied to the device, and fSW is the switching frequency. This formula shows that the power dissipated has a quadratic relationship with voltage and a linear relationship with frequency. The DVFS technique works by reducing these two electrical parameters and adapting them to the dynamic requirements of the load.
Consider now a sensor that transmits data sporadically and for short intervals, or an industrial application, such as a data center’s board running AI models. By reducing both voltage and frequency when they are not needed, DVFS can optimize power management, enabling significant improvements in energy efficiency.
NXP Semiconductors’ PCA9460 is a 13-channel PMIC specifically designed for low-power applications. It supports the i.MX 8ULP ultra-low-power family processor, providing four high-efficiency 1-A step-down regulators, four VLDOs, one SVVS LDO, and four 150-mΩ load switches, all enclosed in a 7 × 6-bump-array, 0.4-mm-pitch WSCSP42 package.
The four buck regulators offer an ultra-low IQ of 1.5 μA at low-power mode and 5.5 μA at normal mode, while the four LDOs achieve an IQ of 300 nA. Two buck regulators support smart DVFS, enabling the PMIC to always set the right voltage on the processors it is powering. This feature, enabled through specific pins of the PMIC, minimizes the overall power consumption and increases energy efficiency.
Energy harvesting
The latest generation of PMICs has introduced the possibility of obtaining energy from various sources such as light, heat, vibrations, and radio waves, opening up new scenarios for systems used in IIoT and industrial environments. This feature is particularly important in IIoT and wireless devices, where maintaining a continuous power source for long periods of time is a significant challenge.
Nexperia’s NEH71x0 low-power PMIC (Figure 2) is a full power management solution integrating advanced energy-harvesting features. Harvesting energy from ambient power sources, such as indoor and outdoor PV cells, kinetic (movement and vibrations), piezo, or a temperature gradient, this device allows designers to extend battery life or recharge batteries and supercapacitors.
With an input power range from 15 μW to 100 mW, the PMIC achieves an efficiency up to 95%, features an advanced maximum power-point tracking block that uses a proprietary algorithm to deliver the highest output to the storage element, and integrates an LDO/load switch with a configurable output voltage from 1.2 V to 3.6 V.
Reducing the bill of materials and PCB space, the NEH71x0 eliminates the need for an external inductor, offering a compact footprint in a 4 × 4-mm QFN28 package. Typical applications include remote controls, smart tags, asset trackers, industrial sensors, environmental monitors, tire pressure monitors, and any other IIoT application.
PMICs for AI and AI in PMICs
To meet the growing demand for power in the industrial sector and data centers, Microchip Technology Inc. has introduced the MCP16701, a PMIC specifically designed to power high-performance logic devices, such as Microchip’s PIC64GX microprocessors and PolarFire FPGAs. The device integrates eight 1.5-A buck converters that can be connected in parallel, four 300-mA LDOs, and a controller for driving external MOSFETs.
The MCP16701 offers a small footprint of 8 × 8 mm in a VQFN package (Figure 3), enabling a 48% reduction in PCB area and a 60% reduction in the number of components compared with a discrete solution. All converters, which can be connected in parallel to achieve a higher output current, share the same inductor.
A unique feature of this PMIC is its ability to dynamically adjust the output voltage on all converters in steps of 12.5 mV or 25 mV, with an accuracy of ±0.8% over the temperature range. This flexibility allows designers to precisely adjust the voltage supplied to loads, optimizing energy efficiency and system performance.
As in many areas of modern electronics, AI techniques are also being studied and introduced in the power management sector. This area of study is referred to as cognitive power management. PMICs, for example, can use machine-learning techniques to predict load evolution over time, adjusting the output voltage value in real time.
Tools such as PMIC.AI, developed by AnDAPT, use AI to optimize PMIC architecture and component selection, while Alif Semiconductor’s autonomous intelligent power management (aiPM) tool dynamically manages power based on AI workloads. These solutions enable voltage scaling, increasing system efficiency and extending battery life.
The post Designer’s guide: PMICs for industrial applications appeared first on EDN.
