The Importance of Sleep Mode Power Consumption in ZigBee/802.15.4 applications

The ZigBee standard actually mandates a 2-year battery life for battery-powered nodes. Here are some techniques for achieving and exceeding that goal.


September 11, 2006
URL:http://www.drdobbs.com/embedded-systems/the-importance-of-sleep-mode-power-consu/192701135

Two new standards, ZigBee and IEEE 802.15.4, now offer the ability to implement wireless industrial control applications. IEEE 802.15.4 is a low data rate, wireless standard that allows industrial control, embedded control and other systems to communicate without wires.

Sensors, switches and servo controllers can be located anywhere in the industrial environment and communicate reliably with the control application. IEEE 802.15.4 covers the basic RF part of the network: the physical layer and the media access control layer.

ZigBee covers the network layer, application framework and application profiles that that guarantee interoperability between equipment from different vendors. In addition, the ZigBee network layer specification defines networking topologies, specifically "mesh" networks.

The ZigBee standard is a superset of the 802.15.4 standard and specifies the network security layer, application framework and application profiles so that ZigBee-certified equipment from different vendors will interoperate seamlessly. A ZigBee-certified application must conform to both the ZigBee standard and the 802.15.4 standard.

It is also possible to have a functioning 802.15.4 wireless network that is not a ZigBee network. In fact, if the application does not require interoperability and will not be employing a complex mesh-type network, it may be preferable to implement a non-ZigBee 802.15.4 network because it will be much easier to deploy.

Network Topologies There are basically three types of network configurations that can be implemented under the 802.15.4 and ZigBee umbrella: point-to-multipoint (star) networks, tree networks and mesh networks. Point-to-multipoint (star) networks are typically used for low cost gaming or entertainment center control.

They are the simplest to implement and require the least amount of code for setup and control. They are typically limited in the quantity of nodes and coverage. Tree networks are more appropriate for applications such as access or industrial control sensing.

Since they allow more nodes, they can cover a larger area than point-to-multipoint networks. However, tree networks may suffer from latency effects that can cause unacceptable data delays for critical applications.

Tree networks also may be subject to critical node failure and cause system failure. Tree networks also usually need larger amounts of code to implement than multipoint systems.


Figure1: A ZigBee tree network.
Mesh networks represent the highest level of 802.15.4/ZigBee configuration and require the most network level code. Mesh networks route data dynamically creating the most efficient path among a multiple of network nodes.

The ability to route data among multiple paths provides mesh networks with a "self healing" capability. If a node in the path fails for any reason, the network identifies a new path using other nodes. This capability makes mesh networks ideal for large building control systems or wide area sensing. Mesh networks are by far the most difficult 802.15.4/ZigBee networks to design and implement.

Building a mesh network from scratch is a complex process. Anyone wanting to deploy a mesh network should conform to the ZigBee specification.

Regardless of the network top-logy, ZigBee/802.15.4 networks have three basic types of nodes: controller nodes, full function (FFD) nodes, and reduced function (RFD) nodes, each of which minimally has a radio, a microcontroller, 802.15.4 MAC and a network layer. Different types of nodes have different processing, memory and power consumption requirements.

Controller and full function nodes, such as those in gateway servers or electrical equipment, are usually hard wired to a power source, so power consumption, per se , is not an issue. Reduced function nodes, on the other hand, are connected to sensors and switches and are usually battery powered.


Figure 2: Nodes can be battery operated.

Battery-powered nodes should have a very long battery life. If possible the battery should outlast the industrial control equipment itself because many of these battery powered end-nodes will be in hard to reach places that will make replacing their batteries very difficult and expensive.

The ZigBee standard actually mandates a 2-year battery life for battery-powered nodes. However, where battery life is concerned, longer is always better. Imagine how annoying (not to mention expensive) it would be to replace all the sensors and servo controllers embedded in a flow control system.

The importance of Sleep Mode Power Consumption Several factors affect power consumption, including the supply voltages of the radio and microcontroller, the active current drawn by the radio and microcontroller, the clock frequency at which the controller operates, the number of external components required in the system (particularly power amplifiers), and the code size, in as much as it affects the MCU clock frequency.

Radios and controllers are available with supply voltages that range from 1.8 to 3.6 volts. Lower is better, but be sure to verify the supply voltage specifications. Some vendors claim 1.8 volt operation, but actually require 2.2 volts to write to flash memory.

Atmel offers the only radios and controllers with "true" 1.8 volt operation. This ability to operate at 1.8 V actually allows the battery to last about 20% longer than it would with a 2 V supply voltage.


Figure A: Typical battery lifetimes.

Radios draw different amounts of current during transmit and receive. ZigBee radios are available with transmit current consumption of 17 mA to 38 mA, and receive current consumption of 15 mA to 36 mA. Again, lower is obviously better.

However, in the case of radios, transmit power and receive sensitivity have an indirect effect on current consumption. If the radio's transmit power or receive sensitivity is too weak, power-hungry power amplifiers must be added to the design to beef up the signal. These external components can add up to 50 mA to node power consumption. When designing battery-powered nodes, the rule of thumb should be to use a radio with the highest transmit power and the highest receive sensitivity to minimize or eliminate power-hungry power amplifiers. The controller should be supported with a MAC that executes in the minimum number of clock cycles. Both radio and controller should support multiple supply voltages with the low-end being no higher than 1.8-volts, with true 1.8-volt operation. However, the most important factor affecting end-node power consumption and battery life is the sleep mode power consumption.

Most battery operated 802.15.4/ZigBee nodes (think sensors or switches) will be in sleep mode 99.9% of the time, waking up periodically for a few milliseconds to check a sensor or poll the other radios. Total power consumption of the node will actually approach sleep mode power consumption.

This is important because engineers and vendors tend to emphasize active power consumption. In face, sleep mode power consumption is frequently buried in the back of the data sheet or may not be in the data sheet at all. Even if it takes a call to the vendor, it is worth finding out the sleep mode power drain of both the radio and controller because it will have a substantial impact on the battery life of the end nodes.

In a system that is off most of the time, active power drain can be less important than sleep mode power consumption. To illustrate this point take the hypothetical example of an end node that wakes up once a minute to perform a task that takes 12 milliseconds with equal amounts of time spent on transmission and reception. The rest of the time the node is asleep.


Figure 3: A power budget.

The total power consumed by just the controller and radio is 0.0062 mA, with sleep mode power representing over one-third of the total. This is why close attention should be paid to sleep mode power as well as active power.

Estimating actual system battery life is virtually impossible because one must factor in all the external components in the reference design as well as the sensors or indicators.

However, in a real-life temperature sensor node application, the microcontroller has active current of 8 mA and sleep current of 1.5 uA with watchdog timer on, while the radio has transmit and receive currents of 17 mA and 15 mA and sleep current of 0.7 uA.

The actual power consumed by this application for wakeup, sense, ADC conversion, transmit data, receive acknowledgement, and transition back to sleep mode is 0.0011 mAh, including external components and sensors for each transmission cycle.

At a rate of one transmission per minute, this node would consume 0.0706 mA per hour of operation. At this rate, two AA 2700mah lithium-ion batteries last about 5.2 years. Increasing the sleep mode current on both the microcontroller and the radio by 1uA each reduces the battery life to 4.8 years " about 10% less.

In addition to paying close attention to sleep mode power consumption of both the radio and the controller, engineers should also check the controller data sheet to verify true 1.8-volt operating range when designing battery-powered nodes.

Some microcontrollers that claim 1.8V operation in their marketing literature actually require more than 2.0V to operate properly. The 0.2-volt difference can cut practical battery life by up to 30%. This information may be buried in a footnote, so it is a good idea to contact the vendor directly to verify the supply voltage. There's a significant need for accurate counters and brownout detection. Brownout detectors (BODs) are frequently used to monitor the supply voltage and shut the system down before it falls below a certain threshold. If a controller operates below its minimum voltage it may execute code incorrectly, causing code runaway and the corruption of non-volatile memory -- an event that can damage the application so it never works again.

While BODs protect the controller from code runaway, they also make a substantial contribution to sleep mode power consumption. An accurate BOD may consume as much as 20 uA. A common solution to this problem is to use a "zero-power@ brownout detector that consumes as little as a few nanoamps. Although this approach keeps power consumption to a minimum, too little current may compromise the BOD's accuracy and speed so that the system fails before the brownout is noticed or a reset executed.

A BOD requires at least 20 uA to achieve sufficient accuracy and speed to protect the system. One way to achieve this accuracy without increasing sleep mode power drain is to turn the BOD off as soon as the controller goes to sleep and wake the BOD up before the controller is allowed to execute any code. This approach provides better brownout protection without compromising sleep mode power consumption.


Figure 4: Brownout detection accuracy.

ZigBee and 802.15.4 end nodes frequently need to keep track of the time, waking up periodically to poll a sensor or check in with a controller node. In these systems, an accurate real time clock must be running in both active and sleep modes while consuming a minimal amount of power.

Either a real-time crystal oscillator or a very low power oscillator (VLO) can be used to effect timed wake from deep sleep mode. If timing accuracy in not important, a VLO can be a good choice. However, if there is even a hint that the timing must be accurate, verify that the microcontroller has a very accurate 32 kHz oscillator.

ZigBee/802.15.4 end-nodes should have a battery life of at least several years. This can be achieved by paying close attention to the supply voltage and to sleep mode power consumption.

Using a radio with high receiver sensitivity and transmit power eliminates the need for power amplifiers and other external components, further extending battery life. By using a controller with the ability to put the brown-out detector in sleep mode, the system will be protected, without unnecessarily draining the battery.

Chris Baumann is Director of Atmel's BiCMOS Products business unit. Before joining Atmel in 1989, he held various positions at Texas Instruments and Honeywell. His B.S. degree in Electrical Engineering and M.S.E.E. degree are both from the University of Notre Dame.

Terms of Service | Privacy Statement | Copyright © 2024 UBM Tech, All rights reserved.