Achieving 'low power' used to be easy. Whenever possible, put your MCU to sleep. Now, MCUs come with a wide range of sleep modes, each of which brings its own unique challenges. Microchip explains how these different modes work and when to use them.
As more electronic devices become battery powered, conserving that power has become paramount. Recently, manufacturers of microcontrollers (MCUs) have added novel ways to control power consumption through various implementations of electronic "switches." By removing power from parts of the chip, dramatic power savings are possible. Additionally, advances have been made in voltage supervisory circuits such that they may be used continuously, consuming only tiny amounts of battery power. This article will explore these emerging technologies and provide some insight into when and how they should be used.
Simply sleeping is not enough
Traditionally, reduced power consumption has been accomplished in microcontroller-based applications by putting the MCU in sleep mode. This method has worked fine for many years; however, there are several trends that are now converging that make this solution inadequate.
More complexity has been added to MCUs in recent years along with the use of more advanced and therefore leaky, process nodes. Combine this with the market shift toward 10 to 20 year battery-powered applications such as utility meters, smoke detectors, and products associated with the green movement, and it becomes apparent that the traditional sleep mode is often no longer adequate. Extreme low-power is now required.
Counteracting the trends
Recently, manufacturers have added novel modes to their microcontrollers that offset the ill effects of increasing complexity and smaller process geometries. These modes might go by the name of LPM5, Standby, STOP2, or Deep Sleep. For discussion purposes, let's refer to them as "Deep Sleep."
Figure 1: The PIC24F16KA XLP family is one example of MCUs employing Deep Sleep.
At the top level, the variety of Deep Sleep implementations operate in the same way. The premise is that power is removed from significant areas of the chip using embedded software controlled switches. By powering off the transistors in areas of the chip, the transistor leakage is removed, and the battery life can be extended significantly. Figure 2 illustrates which circuits are typically enabled while in Deep Sleep—all other circuits in gray are removed from power.
Figure 2: Circuits enabled in Deep Sleep (green sections are on).
While the degree of improvement derived from Deep Sleep varies by manufacturer, an 80 percent reduction in sleep current is common. In fact, some MCUs can now achieve as low as 20 nA while in Deep Sleep mode. By combining low currents with batteries that have low self-discharge rates, Deep Sleep can add years to the life of an application.
Tradeoffs of Deep Sleep verses Sleep
With the benefit of dramatically lower current consumption comes a classic tradeoff. A longer startup time is associated with Deep Sleep modes. While wake from standard sleep was 1 to 10 µs, waking from Deep Sleep modes can take 300 µs to 3 ms depending on the manufacturer. The reason for the increase in wake time is to allow for power-up sequences to terminate and on-chip regulators to stabilize. Since power was removed from parts of the chip, exiting a Deep Sleep mode behaves much like a Power-on Reset condition.
In addition to power-up time, most implementations of Deep Sleep remove power from RAM, peripheral registers, and I/O. This is quite different from standard sleep mode where execution begins precisely where it was stopped. With Deep Sleep modes, the program context must be restored from a non-volatile memory source such as flash or EEPROM. Some manufacturers even offer small "backup" RAM areas that are not powered down in Deep Sleep. Since code execution is required for this restoration step, there is a power penalty paid for using Deep Sleep.
Thus, the tradeoffs for using Deep Sleep are longer wake-up times and the current used to restore the state before the execution was stopped.
The good news is, however, that many applications benefit from Deep Sleep in spite of the tradeoffs. A simple calculation makes it easy to understand when to use and when not to use Deep Sleep. (See Equation 1).
Equation 1:
Tbe
= Breakeven Time where charge in Sleep equals charge in Deep Sleep
Tinit
= Initialization time to resume full power operation
Idd
= Current consumed during run mode
Tpor
= Time Required for Power-on Reset
Ipor
= Power-on Reset Current (includes regulator stabilization capacitor current, if present)
Ipdslp
= Static Current in Sleep Mode
Ipdds
= Static Current in Deep Sleep Mode
This equation models the charge in Sleep and Deep Sleep. The break-even time, Tbe, is the point at which the charge in each mode is equal. Beyond Tbe, Deep Sleep provides the maximum benefit. Example 1, below, illustrates this technique.
Example 1:
Suppose we have an MCU with Deep Sleep mode that has the following characteristics:
Initialization execution time = Tinit = 200 µs
Current during execution = Idd = 400 µA
Power-on Reset Time = Tpor = 600 µs
Current in POR = Ipor = 30 mA*
Current in Sleep mode = Ipdslp = 3.5 µA
Current in Deep Sleep mode = Ipdds = 28 nA
*30 mA includes current for regulator stabilization capacitor
So with Tbe equal to 5.2 seconds, an application that remains in Deep Sleep longer than 5.2 seconds will benefit.
Waking up from Deep Sleep
We have seen how Deep Sleep use increases battery life by driving consumption to extreme levels as low as 20 nA. Now, the question is, "How do I wake up if I have removed power from most of the chip?"
With traditional sleep mode, we had a variety of ways to wake-up. They included interrupts, timers, communication reception, end of ADC conversion, supply-voltage changes, and so on. These wake-up sources are very useful. Fortunately, MCU vendors have included many of these same features in Deep Sleep mode.
Sources of wake-up available in Deep Sleep mode can include interrupts, reset pins, power-on reset, real-time clock alarms, watchdog timers, and brownout detection. You might notice that wake from communication reception and end of ADC conversion are missing from this list. Since these portions of the device do not have power, these wake-up features cannot be supported in Deep Sleep mode. There is a wide gamut of wakeup source implementations among the different manufacturers; therefore, it is very important to review which capabilities have been included for a specific device family.
For example, some vendors only exit Deep Sleep by the assertion of the RESET pin. This is great for applications that have an "on" button and consume no additional current. By pressing the button, the application wakes from Deep Sleep, restores state, and is ready to run. This is ideal for applications such as thermometers and handheld devices. Another use for Deep Sleep lengthens the shelf life of battery-powered products when they are shipped in the Deep Sleep mode.
For a more complete system implementation, some vendors provide increased flexibility through the addition of real-time clock and calendar functions. These additions allow an application to be autonomous, adding as little as 500 nA to the Deep Sleep current. Rather than waiting for a button, the clock's alarm wakes the device. This is very important for an application such as a smoke detector where it must wake 2 to 3 times per minute to sample the air quality.
A battery-powered sensor that wakes only a few times per day to transmit data is another example.
By matching the needs of the application with Deep Sleep wake-up features, dramatically longer battery lifetimes can result.
Application safety
The goal of using Deep Sleep is to increase the application lifetime while on battery power. By using new MCUs with Deep Sleep modes, this goal is achieved. However, as the battery is consumed and we approach the end of the useful life of the battery, the risk of improper operation increases.
This issue is traditionally solved using supervisory circuits, such as brownout reset (BOR) circuits and watchdog timers (WDTs). Brownout circuits can detect if the battery output is too low for safe operation and force the application to a safe state. Watchdog timers offer similar protection against errant code execution should the MCU attempt to execute in an unsafe voltage/frequency region. The main drawback to these circuits is that they have traditionally consumed as much as 5 to 50 µA. Therefore, these traditional solutions are no longer compatible with the new Deep Sleep modes.
Recently, MCUs have emerged with a variety of new, lower-current BOR and WDT circuits specifically designed for Deep Sleep mode.
Sometimes called Deep Sleep BOR (DSBOR) or Zero-Power BOR, these brownout circuits offer diminished accuracy in return for outstanding current consumption—as low as 45 nA. This capability is important not only for protection at the end of battery life but also in the case of momentary power loss due to battery holder flex, a common issue with battery-powered systems. The implementation of low-current BORs varies from vendor to vendor, as some may be turned off and some are permanently on. Not all manufacturers of MCUs with Deep Sleep offer a BOR, and thus, again, it is important to review MCUs carefully for compatibility with your application.
Like BOR, the watchdog timer currents have been reduced in MCUs with Deep Sleep. While only a few vendors have included WDT with Deep Sleep, the current performance of the circuit is as low as 400 nA.
With these advancements, both supervisor circuits can now remain powered while in Deep Sleep, as the combined current consumption can be as low as 445 nA — 99 percent lower than the prior generation of MCUs. Now, using Equation 1, the break-even time (Tbe) with both supervisor circuits is only 5.9 seconds. Therefore, the current consumption of these new supervisory circuits facilitates safer applications for a whole host of applications that remain asleep for more than six seconds.
Deep Sleep mode in combination with new low-current supervisory circuits has emerged as a viable way to dramatically increase battery life in MCU-based applications. Through careful design of wake-up circuits and powering down wasteful parts of MCUs, vendors have thwarted the high current leakage from shrinking process geometries to deliver sleep currents that are 80 percent lower than before. Deep Sleep can be successfully applied in a variety of applications requiring extreme low power by simply understanding the power breakeven point. This is great news for the myriad of designers who are creating the next portable gadgets, as it promises to bring less waste, robust safety, and more capabilities to our battery-powered products.
