Dave Freeman, Texas Instruments
The Internet of Things (IoT) has been the topic conferences, articles, and blogs. Much of this has focused on the standards for communication and security of the information and devices. Just as important is powering the myriad of devices that comprise the IoT. It is worthwhile to first describe what makes up the IoT. The idea behind the IoT is that everything worth talking or listening to is connected for communications.
In many cases, the connected item can be an existing device that recently acquired communication capability, or new products created to enrich the information environment. Most often these devices are connected wirelessly. This connectivity sets the bar for the power developer. Wireless communication provides a high degree of flexibility that should not be restricted by any needs for any special power connectivity.
Figure 1 shows one view of the IoT. As the picture illustrates, everything can be connected. We also see that a subset of this concept is the wireless cloud. The wireless cloud provides connectivity for users and their devices. It also points to an important part of power consumption. A white paper published by the Center for Energy-Efficient Telecommunication, CEET, in April, 2013, predicted that the wireless cloud will consume 43 TWh this year (2015). Of that power, wireless networks will use 90%.
Click image to enlarge
Figure 1. View of the Internet of Things
In 2012, the wireless cloud only consumed 9.2 TWh, so this is a very large increase in energy. It is expected to grow as more devices and traffic become part of the IoT infrastructure. This becomes the challenge for power supply designers. At some point in the future, expansion of the IoT will be limited by the energy it consumes. So let’s get down to some of the tasks the power supply designer needs to address.
As Figure 1 shows there are small devices or wireless sensors that monitor the environment, appliances that need to communicate their status, wearable electronics, security systems, automobiles, and industrial equipment, as well as wireless networking equipment mentioned earlier. So there are many devices and associated infrastructure that are integrated to form the IoT. The description of the IoT leave many people thinking of the IoT as being made primarily of wireless sensors that communicate important information – so we will start there.
Wireless sensors are often placed in environments that are difficult or expensive to access. This drives the energy supply to be something that either lasts for a very long time, more than 10 years, or can be supplied from the environment in a dependable fashion. So power management must be very frugal with energy consumption it uses to manage the energy. Wireless sensors also have a very high peak-to-average power ratio, in some cases greater than 100.
Figure 2 shows an example of various power modes for a wireless sensor. In this example, the sensor sleeps most of the time but may wake up on communication to take a measurement, which also serves to let the system know the sensor is available. At a much longer interval the sensor may provide much more information to the system. This transmission may require much more energy, so there is a dependency on the available stored energy. The power management solution must be able to supply the needed peak power while consuming very little of the average energy. Any of these systems where the environmental energy is low, the power management solution must collect the energy until there is sufficient energy for the required utility.
Click image to enlarge
Figure 2. Wireless sensor power profile
Figure 3 shows an implementation example of such a system. In this example, maximum power point tracking (MPPT) is implemented based on the ratio of the MPPT voltage to the open circuit voltage of the photovoltaic source. This MPPT implementation minimizes the energy consumed while performing the MPPT function. This example also integrates the energy storage function. Since the life time of the energy storage element is very important, care must be taken not to over-discharge or charge this element. In this example, levels are set for the minimum and maximum storage voltages.
Click image to enlarge
Figure 3. Energy management for a wireless sensor
In order to inform the system about the stored energy level, the designer can configure externally the voltage level where this notification, VBAT_OK can be given. A buck regulator is also integrated into the solution to power the system load. This entire system has a typical quiescent current of only 500 nA resulting as high efficiency, even at low current. For example, at 500 mV input and 100 uA charging current, the boost converter efficiency is greater than 70 percent.
Another category of things that form the IoT is appliances. Many times we do not consider how such personal devices are part of the IoT, but it is through the IoT that we can exchange information with our clothes washers, refrigerators, and the like. Traditionally, these devices did not have much to say. You simply plugged them into the grid, set some information, and they did their job. In the example of the clothes washer, it might chime when the cycle is complete, but that was about it. However, today’s connected appliances can give you information that does not depend on you hearing a chime. How does this affect the power designer?
These appliances are going from only on when they have a task to perform to being always on, or at least some functions must be always on. These functions must be powered efficiently as they are always on and ready to exchange information. It is this new requirement that adds to the task of the power designer where traditionally they only needed to concern themselves with providing power to perform the task of the appliance.
Since these devices require higher power to complete their task, they are tied to the grid in most cases. Therefore, energy harvesting is not required. However, since they are always on, quiescent power and efficiency is important for the new connectivity function. Many times these connectivity functions are performed wirelessly and communicate with a local network. This sets the power level requirements to below 10W. This low-power level is typically satisfied by a AC/DC fly-back solution. There are many integrated fly-back solutions to choose from, but this particular application has its own requirements. Figure 4 shows an example of such a power solution to satisfy the need for connecting to the IoT.
Click image to enlarge
Figure 4. Low power AC/DC for appliance connectivity
There are several key features in the Figure 4 fly-back example. First is that it has very low standby power, less than 30 mW. This is important because connectivity must always we ready, even if the appliance is idle. Another aspect is low electromagnetic interference (EMI) as this device most often will be powering a wireless communication circuit. In this example, the controller uses valley switching and frequency dithering to help with EMI reduction.
Another aspect is size of the power solution. The size itself is typically not the issue, but it is how the size affects the final cost. Although the IoT is an exciting technology and having your washer tell you that your clothes are ready for the drier, or the refrigerator tell you that someone left the door open via a message sent to your cell phone are desirable, the consumer does not want to pay any more than necessary. So the solution needs to minimize the cost of the power solution. One way this can be accomplished is by reducing the size. Size reduction is accomplished by operating at a higher frequency, in this case 115 kHz.
Let’s move to what is becoming the big concern for the IoT. As mentioned at the start, it is the wireless network that is the primary energy hog. There is plenty of power design development work in play to help solve this problem. Everything from envelope tracking to digital RF power amplifiers are being researched and developed for the base station, which is certainly more than can be covered here. As many base stations are powered from the grid, there is an opportunity to make that front-end power factor control (PFC) supply more efficient. One such approach is shown in Figure 5. This is the power stage of bridgeless PFC. By removing the diode bridge, the system can be made more efficient. There are many different bridgeless PFC topologies, but we will focus on continuous conduction-mode (CCM) totem-pole version.
Clcik image to enlarge
Figure 5. Totem-pole bridgeless PFC
The benefits from this topology are reduced component count and removing bridge losses. We can further improve efficiency by taking advantage of Gallium Nitride, GaN, switching devices. These devices, Q3 and Q4, offer lower gate losses allowing higher frequency operation. The other parasitic losses like Coss are also lower. Additionally, there is no intrinsic body diode, so there is minimum reverse recovery loss. Q1 and Q2 are switched at the line frequency and they can be silicon MOSFETs while providing additional loss reduction over diodes alone. There have been several published papers that go into detail about this topology as it will benefit the overall efficiency of high-power grid connected system.
The IoT provides many new challenges for power designers with only a few mentioned here. Adoption and coverage of the IoT depends heavily on reducing the energy demands from harvesting environment energy, to minimizing home energy, and reducing the total network energy requirements. As we develop new technologies for energy harvesting, we must remember that reducing the energy needs will still be important to fuel growth. The lower the energy needs, the more likely it can be supplied from the environment.
Reducing the energy need from the grid is also important. Thinking about each individual grid-powered application may disguise the impact that a few tenths of a percent of efficiency has. It is the total that draws the attention of governments. It’s not the one washer or the one base station, but the millions that create the energy need. Luckily power designers have new technology to address these challenges. In some cases it will be process technologies that enable high-voltage components to be integrated with low-voltage control. In other cases it can be the WBG devices that improve high-voltage conversion by allowing low loss at higher switching speeds. Times are definitely getting more exciting for the power designer.
1. Center for Energy-Efficient Telecommunications (CEET), “The Power of The Wireless Cloud”, April, 2013
2. bq25570 datasheet, Texas Instruments