New smoke alarm regulations often require a 10-year battery life if the battery is the only power source. Battery selection in these alarms is highly dependent on the system power consumption, and lower power alarms have smaller, lower cost batteries. Achieving low power consumption requires low standby currents and careful management of the smoke alarm regulators, drivers, and amplifiers. The TPS8802 smoke alarm analog front end (AFE) integrates configurable regulators, drivers, and amplifiers with low active and standby currents for achieving 10-year battery life with a single lithium primary battery. The TPS8802 supports multi-wavelength photoelectric smoke sensors, carbon monoxide (CO) sensors, and a variety of battery configurations, all of which are explored in the context of power consumption.

The external component configuration has a large influence on the system power consumption. In particular, the VCC, MCU, and LED supplies must be selected to minimize power consumption. Different options and their advantages are discussed in this section.

The TPS8802 operates with a battery voltage between 2.0 V and 15.6 V and a VCC voltage between 2.6 V and 15.6 V. When the battery voltage is less than 2.6 V, the boost converter increases the voltage to adequately supply VCC. This allows the TPS8802 to be powered from a variety of batteries, including 3-V lithium, series-connected 1.5-V alkaline, and 9-V alkaline or lithium. Alkaline batteries typically have a shorter shelf life than lithium batteries, making lithium batteries more viable for 10-year smoke alarms. The VCC voltage must be at least 2.6 V for the internal amplifiers to function properly, making the system most efficient when VCC is close to 2.6 V. Therefore, 3-V lithium manganese dioxide (LiMnO₂) batteries and two series-connected 1.5-V lithium iron disulfide (LiFeS₂) batteries provide the best capacity per cost in this application. Other lithium battery chemistries such as lithium ion and lithium thionyl chloride are not considered due to their higher cost. With the battery voltage selected, the system architecture can be designed and current consumption can be calculated. A specific battery size is selected after the capacity requirement is calculated.

There are several ways to supply power to the TPS8802 VCC input with a 3-V battery. Two options are explored in detail: connecting VBST to VCC and connecting VBAT to VCC through a switch. These configurations are demonstrated in the example smoke and CO system and smoke-only system.

When using the TPS8802 with a 3-V battery, the boost converter can be used to power the horn, use the battery test load, and power a blue LED. The TPS8802 automatically enables the boost converter on power-up when over 2.0 V is applied to VCC. Connecting VCC to VBST allows the boost converter to provide the minimum 2.6 V required on VCC with a battery voltage as low as 2.0 V. Connecting VCC to VBST increases the usable voltage range of the battery, improving the battery life. During operation, the TPS8802 VCCLOW_BST feature can be used to automatically enable the boost converter when the VCC voltage drops below 2.4 V and disable the boost converter when the VCC voltage is above 2.5 V. The VCCLOW_BST feature sustains the VCC voltage above 2.4 V with low power consumption. The functional performance is unaffected when the VCC voltage is between 2.4 V and 2.6 V, although parametric performance may be affected.

In systems where the CO amplifier, MCU LDO, 300-mV reference, VCCLOW monitor, interconnect, and sleep timer do not need to be continuously powered, the TPS8802 can be unpowered in between smoke measurements to reduce the system standby current. To achieve this, the MCU controls a load switch between VBAT and VCC to disconnect the TPS8802 in between smoke measurements. This allows the smoke alarm to achieve an idle current less than the TPS8802 typical 3.8-μA standby current. This configuration provides the best power savings in smoke-only alarms where the CO amplifier is not used. The battery voltage must be above 2.6 V for parametric performance and above 2.4 V for functional performance. While this configuration has a tighter battery voltage range and cannot use the full capacity of the battery, the lower power consumption of this configuration gives it a longer battery life. VCC and PLDO are shorted together to remove the voltage drop caused by the internal PLDO block.

The MCU can be powered with the TPS8802 MCU LDO, the battery directly, or an external LDO. VMCU sets the voltage of the TPS8802 digital inputs and outputs. VMCU must be the same voltage as the MCU supply to ensure reliable digital communication between the MCU and TPS8802.

It is most efficient to connect the MCU directly to the battery if the MCU can withstand the full battery voltage range. Connect MCUSEL to GND to set the MCU LDO to 1.8 V during the power-up sequence. The battery voltage overrides VMCU without drawing significant current. When the system is powered up, disable the MCU LDO to save power. Ensure the battery voltage does not exceed 3.5 V when connected to VMCU.

If the MCU cannot be directly connected to the battery, the MCU LDO can be enabled to output 1.5 V, 1.8 V, 2.5 V, or 3.3 V to the MCU while consuming 2.0 μA of current. Using a 3.3 V microcontroller requires the boost converter to be periodically enabled, greatly increasing the system current consumption. For this reason it is recommended to use a 1.5 V, 1.8 V, or 2.5 V microcontroller with the MCU LDO. Because the MCU LDO is register programmable, the LDO can be set to different voltages during the smoke alarm operation; for example, 3.3 V during measurements and 1.8 V between measurements.

If VCC is connected to VBAT through a switch, the MCU LDO cannot be used to power the MCU because the MCU LDO is not powered when the switch disconnects power to VCC. There are three options to power the MCU:

• Connect the MCU to the battery and VMCU to VCC.
• Use an external LDO to power the MCU and the MCU LDO to power VMCU. The MCU LDO is programmed nearest to the value of the external LDO: 1.5 V, 1.8 V, 2.5 V, or 3.3 V. The 2.0 μA MCU LDO current is consumed only while the TPS880x is powered.
• Use an external LDO to power the MCU and a switch to connect the external LDO to VMCU while the VCC switch is enabled.

The photoelectric smoke sensor LED can be powered by the battery, PLDO, or LEDLDO. For maximum power savings, the LED should be supplied directly from the battery if the voltage supports it. If direct battery voltage is not supported, use the boost converter with VBST set to the minimum required voltage. Before selecting how to power the LED, the minimum LED supply voltage V_LED(min) must be calculated using Equation 1. V_F is the LED forward voltage, V_DINA(drop) is the LED driver dropout voltage (300 mV at 150 mA and 500 mV at 500 mA), V_CSA is the voltage at the CSA current sense pin, and ΔV_LED is the voltage drop caused by the capacitive voltage supply on the LED. Using a higher capacitance on the LED supply decreases the supply voltage drop when pulsing the LED as shown in Equation 2.

A low current infrared LED can be powered directly from the battery through a current-limiting resistor and large capacitor. This is the most efficient way to power the LED. Ensure the minimum battery voltage is greater than V_LED(MIN).

An infrared LED or low current blue LED can be powered through PLDO with a low-leakage Schottky diode. PLDO is supplied by VCC and can supply a voltage higher than VBAT if VCC is connected to VBST. PLDO has two operating modes: a pass-through mode when VCC is less than 5 V (typical), and a regulation mode when VCC is greater than 5 V (typical). The pass-through mode shorts PLDO to VCC. The regulation mode has a minimum 1 V drop between VCC and PLDO. Therefore, PLDO is most efficient when it is operating in the pass-through mode. The Schottky diode must be low-leakage, approximately 0.1 µA, to prevent the LED capacitor from discharging into PLDO when the boost converter is disabled. A silicon diode can be used instead of a Schottky diode if the higher voltage drop is tolerable.

Any LED can be powered through the LED LDO. The LED LDO is current limited and has a series diode to block reverse current. The LED LDO is supplied by VBST and regulates to a programmable voltage between 7.5 V and 10 V, but does not require regulation in order to charge the LED supply capacitor. When the LED LDO is enabled and VBST is less than the programmed voltage, the LED LDO outputs the VBST voltage with a typical 1 V diode drop and 3 mA current limit. If VCC is connected to VBST, the LED LDO output is lower than PLDO in the pass-through mode and higher than PLDO in the regulation mode. For this reason, the LED LDO is best used when V_LED(MIN) is above 4.5 V.

Schematics for the smoke-only system and smoke and CO system are shown. The power consumption of these systems are calculated and measured.

The smoke and CO system connects VBST to VCC to allow battery voltages down to 2.0 V. Here the MCU LDO is set to 1.8 V, a low voltage that can be powered without the boost converter being enabled. The boost converter can be enabled to 2.7 V, 3.8 V, or 4.7 V to supply the IR LED through PLDO with a low-leakage (0.1 µA) Schottky diode. The blue LED is supplied by the LED LDO to support the higher blue LED forward voltage. The CO amplifier is continuously enabled to track the slowly-varying CO concentration signal.

The smoke-only system uses a microcontroller that connects directly to the battery. Because the TPS8802 MCU LDO is not required, the TPS8802 can be unpowered between measurements. A TPS22919 load switch is used to connect the battery to VCC during measurements and discharge VINT between measurements. VCC and PLDO can be shorted because of the low battery voltage. VMCU is connected to VCC to set the TPS8802 digital logic level equal to the microcontroller’s supply voltage. With the battery voltage used to supply VCC, the minimum battery voltage is limited by the 2.6 V minimum VCC voltage. The IR LED is driven at a low enough current such that 2.6 V is sufficient. The blue LED has a higher voltage that requires the boost converter, thus the LED LDO is used.

Various blocks and measurement timing have an impact on the system power consumption. These blocks will be discussed in the context of the system power consumption.

In this report, standby current is the current consumption if no measurements are taken. The two primary contributors to standby current are from the TPS8802 and microcontroller.

The TPS8802 typically consumes 3.8 μA current when 3 V is applied to VCC. This current is always drawn when VCC is supplied with 3 V. The TPS8802 standby current supplies the power LDO (PLDO), 2.3 V INT LDO analog supply, and digital core. I²C communication and the sleep timer are active and do not require additional current to use. The primary blocks in the TPS8802 that add to system standby current consumption are the MCU LDO, CO amplifier, and VCCLOW monitor. The MCU LDO uses 2.0 μA current, CO amplifier uses 0.6 μA current, and the VCCLOW monitor uses 0.9 μA current. Other blocks such as the photo amplifier and boost converter consume several hundred microamps of current and are strategically enabled for a short amount of time during measurements to make their overall contribution to total power negligible.

The microcontroller standby and active currents contribute to the total system power consumption. While there is no specific requirement for the microcontroller, a microcontroller in the range of 1 μA standby and 1 mA active current is generally sufficient to not dominate the total power consumption. It is critical that the microcontroller enters its standby mode in between smoke measurements. Some microcontrollers are able to enter an ultra-low power mode, less than 100 nA, if the microcontroller clocks are disabled.

The TPS8802 sleep timer allows the ultra-low power mode to be entered in between measurements. This can be done by writing the sleep time duration to the TPS8802 registers SLPTMR1 and SLPTMR2, then writing 1 to SLP_EN to start the sleep timer. The MCU can enter the ultra-low power mode while the TPS8802 keeps track of time. When the sleep timer finishes, the TPS8802 wakes up the MCU by sending a pin interrupt to the microcontroller.

The microcontroller supply voltage range has a large influence on the microcontroller current. If the boost converter is required to power the microcontroller, all of the microcontroller current is scaled by the boost converter efficiency and input-to-output voltage ratio. If an LDO is required to power the microcontroller, the LDO standby current adds to the total system current. For this reason it is most efficient to use a microcontroller that can be directly powered by the battery. The options for powering the microcontroller are discussed in Section 2.3.

In this report, measurement current is the current consumption from taking measurements. These measurements include smoke, CO, battery voltage, and the horn functionality.

Taking a smoke measurement generally follows the procedure:

1. Enable the photo amplifier and analog multiplexer (AMUX) buffer.
2. Take an ADC measurement after the photo amplifier and AMUX settles. This measures the ambient light entering the photo chamber.
3. Enable the LED driver.
4. Take an ADC measurement after the photo amplifier and AMUX settles. This measures the ambient light and smoke concentration in the photo chamber.
5. Disable the LED driver, photo amplifier, and AMUX buffer.
6. Process the ADC measurements to determine the smoke concentration.

The power consumed from taking a smoke measurement is dominated by the LED driver current. The average LED current is shown in Equation 3, where I_LED is the LED pulse current, T_LED is the LED pulse duration, and f_MEAS is the measurement frequency.

If the LED is supplied by the boost converter, the average LED current is scaled by the boost input-output voltage ratio and efficiency. The average LED current draw from the battery is calculated in Equation 4. Connecting the LED directly to the battery is the most efficient method of powering the LED.

Secondary sources of power dissipation during the smoke measurement are the MCU active current and boost charging current. The MCU active current is drawn during the length of a measurement while the MCU is not in standby. Optimizing measurements for speed is essential to reducing the MCU active current. Boost charging current is caused by charging and discharging the VBST capacitor. Boost charging is further discussed in Section 3.3.1.

The CO amplifier is continuously powered to amplify the electrochemical CO sensor current. Current is consumed from powering the CO amplifier and taking measurements of the CO amplifier output. Measuring the CO concentration requires enabling the AMUX buffer to output the COO voltage and taking an ADC measurement. The majority of power consumption from taking the CO measurement comes from the MCU active current. To take the CO measurement as fast as possible, take the CO measurement directly before the smoke measurement.

The CO sensor connectivity test is not a significant contributor to total power consumption due to the infrequency of the test.

The high current draw required for the battery test can make it a significant contributor to total power consumption. The average test load current is calculated in Equation 5, where I_BATTEST is the battery test load current, T_BATTEST is the battery test duration, and f_BATTEST is the battery test frequency.

When using the TPS8802 battery test load, the load current is scaled by the boost input-output voltage ratio and efficiency. The average battery current draw caused by the battery test load is calculated in Equation 6.

The last standard test that contributes to the total power consumption is the weekly or monthly alarm testing. In this test, the horn driver and boost converter are enabled by the end user for several seconds. The average current draw from this test can be estimated based on the horn driver current using the following equation:

Any time the boost converter is enabled, charge is transferred from the battery to the VBST capacitor. When the boost converter is disabled, the VBST capacitor charge supplies power to any circuitry connected to VBST. This energy transfer dissipates power because the boost capacitor cannot be charged or discharged with 100% efficiency. The TPS8802 boost converter typically operates between 65% and 85% efficiency when the load current is much higher than the boost converter active current. Using a lower inductor peak current limit BST_CLIM improves the efficiency with the tradeoff of decreased maximum output current. The majority of circuitry connected to VBST consumes constant current independent of the VBST voltage. Therefore, an increase in the VBST voltage directly increases the power consumption of the connected circuitry.

The extra current from boost charging is calculated using Equation 8, where:

• f_CHARGE is the boost charge frequency
• V_BAT is the battery voltage
• C_BST is the capacitance
• V_LOW is the voltage before charging and after discharging
• V_BST is the boost converter output voltage
• η_BST is the boost converter efficiency

Equation 8 is derived by taking the difference of the energy used to charge the VBST capacitor and the energy used from discharging the VBST capacitor. The boost charging current is in addition to all other currents.

Equation 8 is particularly useful in calculating power consumption with the TPS8802 VCCLOW_BST feature enabled. VCCLOW_BST automatically enables the boost converter when the VCCLOW monitor is enabled and a low VCC voltage is detected. In this scenario, the boost charge frequency depends on the capacitance, standby current, boost Schottky diode leakage current, boost converter voltage, and VCCLOW detection voltage. The standby current here is the current continuously drawn from the VBST capacitor. The frequency and current are calculated in Equation 9 and Equation 10. Equation 10 highlights the importance of using a low-leakage Schottky diode. The Schottky diode leakage effectively adds to the standby current. Using a Schottky diode with 1 μA leakage current reduces the power drawn when VCCLOW_BST is active. Low leakage Schottky diodes generally have a higher forward voltage and reduce the boost converter efficiency. Using a Schottky diode with less than 1 µA leakage current can reduce the total system battery life because of the higher forward voltage.

In systems where a load switch periodically applies power to VCC, power is dissipated during the TPS8802 initialization. The initialization current has several components: VCC charging, VINT charging, MCUSEL sensing, and VBST charging. The VCC, VINT, and VBST waveforms and battery current draw during the TPS8802 initialization are shown in Figure 3-1.

The VCC and VINT capacitors are charged when the load switch is enabled. The amount of power required to charge these capacitors depends on the capacitance, voltage, and frequency of initialization. The MCUSEL pin is sensed during initialization to determine the default MCU LDO voltage. Sensing the MCUSEL pin typically draws 2.2 mA current for 1.2 ms. The initialization current is calculated in Equation 11.

The initialization current caused by boost charging depends on the charging frequency and VBST voltage change. During initialization, the boost converter is automatically enabled to 3.8 V with 500 mA peak inductor current. If VBST is below 3.8 V, the boost converter switches until VBST is above 3.8 V. If the VBST load is low between measurements, the boost converter does not switch during every initialization.

The VBST voltage change depends on the VBST load between measurements. The minimum VBST voltage change is the boost converter ripple, and the maximum VBST voltage change is VBST minus VBAT. In Figure 2-8, the only load on VBST between measurements is the Schottky diode leakage. Therefore the Schottky diode leakage has a large impact on the boost charge current. The boost converter ripple is calculated in Equation 12, the charging frequency is calculated in Equation 13, and the boost charging initialization current is calculated in Equation 14.

All of the identified sources of current consumption are tabulated in a spreadsheet to calculate the system power consumption. The spreadsheet uses a set of system parameters and measurement subroutines to break down the power consumption block-by-block for each measurement subroutine. The decomposition allows users to identify the biggest sources of power consumption when testing their system and further reduce the system power.

The power calculation is based on the schematics in Section 2.5. These schematics and associated power calculation can be modified based on the system requirements. One of the key calculation parameters is that the IR LED is the primary method of smoke detection. When the smoke concentration is low, the IR LED is used to measure the smoke level. When the smoke concentration rises, the IR and blue LED are used to measure the smoke level. Because only the IR LED is pulsed most of the time, only the IR LED power consumption is considered.