REALLY Smart Wearables Customize Their Power Supply

Abstract: In this design solution, we review the operation of optical sensors used in wearables for health and fitness and show how they can waste power even in less demanding conditions. We’ll then present an alternative power management IC (PMIC) that allows optical sensors to save power with dynamic voltage scaling techniques, particularly when they are not being used to their full potential.


“Citius, Altius, Fortius,” or “Faster, Higher Stronger” in its more recognizable form, is the motto of the Olympic games (Figure 1).

Olympic flag. Figure 1. Olympic flag.

Of course, not everyone who wears a health and fitness monitor to measure and record their vital signs is an aspiring Olympic athlete. For more sedentary individuals, like your humble scribe, an occasional workout may be the only time a wearable is pushed anywhere close to the level of performance for which it was designed. In addition, everyone (even Olympic athletes) needs to rest and sleep. Ultimately, this means that for a large part of their lifespan, wearables are often required to function well below their potential. Is it reasonable to assume that during this time they are consuming as little power as possible to conserve battery life? This may not always be the case.

In this design solution, we will review the operation of optical sensors used in health and fitness monitors and show how, even in less demanding conditions, they can waste power. We will then present a PMIC that does more than simply reduce current to allow them save even more power under these circumstances.

Optical Sensing

Optical sensors are commonly used to measure health indicators like heart rate and blood oxygenation (SpO2). These measurements are based on a technique called photoplethysmography (PPG). A PPG signal is obtained by illuminating skin using a light-emitting diode and detecting changes in the intensity of the reflected light (Figure 2) by using a photodiode which generates a current proportional to the amount of received light.

PPG using LED and photodiode. Figure 2. PPG using LED and photodiode.

Measurement Challenges

Making reliable measurements to provide accurate heart rate (or SpO2) information is challenging and can be affected by several different factors. The first of these is motion. For example, when a device wearer is asleep (or inactive), LED light pulses of lower frequency and amplitude are required compared to when the wearer is engaged in high-intensity exercise. Ambient lighting conditions have a similar impact. In a dark environment, fewer and lower intensity LED light pulses are required than in bright or sunny conditions. Put simply, user conditions that require brighter and more frequent LED pulses drain more power from the battery. In the course of a single day, a user may encounter varying lighting conditions and engage in different levels of activity. Therefore, ideally, an optical sensor should only use as much power as is required for a given use case. Since a typical user will spend between 7 and 9 hours a day sleeping, reducing power consumption during this time is important. To date, power-saving efforts have primarily focused on reducing current consumption, However, in the following section we consider the limitations of this approach.

LED Circuit

A simplified example of a circuit used to drive the LED in a wearable device is shown in Figure 3. Since some class of lithium battery (which can range from 3.2V to 4.35V) is used as a power source, a boost circuit is used to increase the circuit supply voltage to 5V.

High LED current circuit. Figure 3. High LED current circuit.

In this example scenario (when the user may be engaged in strenuous exercise), a high LED current is required (e.g. 100mA) with a voltage drop across the LED of 4V, allowing a drain voltage (VDRV) of 1V for the transistor to maintain operation in the linear region. However, under more favorable measurement conditions (for example, when the user may be asleep), less LED current is required (5mA) and the circuit condition in Figure 4 exists. Due to the lower current, the voltage drop across the LED is reduced to 3V, leaving the voltage across the transistor to be 2V. However, the lower current means that the transistor does not need this voltage level to maintain linear operation instead it requires a VDRV of only 0.2V.

Lower LED current circuit. Figure 4. Lower LED current circuit.

This means that there is an excess voltage of 1.8V and therefore an effective power wastage of 9mW (i.e. 1.8V x 5mA) in this circuit condition.

Dynamic Voltage Scaling

A preferable scenario would be one where the supply voltage provided by the boost circuit changes in response to LED current such that the transistor always has sufficient (but not excess) voltage to maintain linear operation. This technique, referred to as Dynamic Voltage Scaling, can be implemented by simply using a “lookup” table (Table 1) to determine the output voltage required by the boost converter as shown in Figure 5. This can provide significant power savings in less demanding measurement conditions.

Table 1. DVS Lookup Table

LED Current (mA) VDRV (mV)
128 950
96 480
64 320
32 160

Dynamic voltage scaling. Figure 5. Dynamic voltage scaling.

For this circuit:


Many wearable systems manage LED current to reduce power consumption. DVS now provides the option to also manage voltage for even greater power savings. The PMIC in Figure 6 features a buck-boost converter that uses DVS on its output voltage.

MAX20345 PMIC for wearables. Figure 6. MAX20345 PMIC for wearables.

This PMIC also features three buck regulators, three low-dropout (LDO) linear regulators, and a buck-boost regulator, providing up to seven regulated voltages, each with an ultra-low quiescent current. This allows it to power multiple sensors (including the MAX86140 optical AFE) and a microcontroller in a typical design for a wearable device (Figure 7). It is available in a 56-bump, 0.4mm pitch, 3.37mm x 3.05mm wafer-level package (WLP).

MAX20345 typical application circuit. Figure 7. MAX20345 typical application circuit.


The power consumed by an optical sensor in a health and fitness wearable varies widely in response to changing user conditions. Until now, current consumption has been the focus of power-saving techniques used in these devices. Although effective, it neglects the other part of the power equation, namely voltage. As we have shown, power is wasted in an optical sensor that uses a fixed supply voltage in all circumstances. In this design solution, we presented a PMIC that uses a Dynamic Voltage Scaling technique to provide the appropriate (but not excessive) voltage required to supply the current needed by an optical sensor’s LED for a given use case. The ability to dynamically adapt voltage in response to current demand offers a new degree of freedom for power-saving approaches in wearable devices.