Application Note

Designing for Multiple Power Sources: Seamless Performance Optimization with Enhanced Linear Regulators

Application Note PSAN004

Download PDF

Introduction

Portable electronics demand ultra-low noise power delivery to sensitive analog circuits. These products increasingly operate from dual power sources—USB connectivity for stationary use and internal batteries for portability—creating a fundamental challenge for power management design. The regulator must deliver stable, low-noise output voltage while efficiently handling input sources that vary from 3.5 V (Li-Ion battery) to 5 V (USB power).

This application note presents an adaptive dual-mode regulation technique that maintains the noise performance and simplicity of linear regulation while achieving high efficiency across the full input voltage range. The approach uses a single Polaris Semiconductor enhanced linear regulator (ELR) to eliminate the conventional two-stage buck-plus-LDO architecture, reducing cost, board area, and design complexity while delivering superior performance in noise-critical applications.

The Challenge

Consider a portable precision instrument such as a spectrum analyzer, data acquisition unit, or signal conditioning system. In these products, supply noise directly limits measurement accuracy and system dynamic range. Such instruments typically require a stable 3.3 V supply with noise performance in the low microvolt range.

Linear regulators offer a compelling solution for providing a clean, stable 3.3 V rail from a single-cell Li-Ion source. Since their efficiency is largely dictated by the ratio of output to input voltage, these regulators can achieve upwards of 85% efficiency when stepping down from 3.7 V (nominal battery voltage) to 3.3 V, while simultaneously offering extremely low output noise and a remarkably compact footprint. The topology is straightforward: a series pass element and feedback network with minimal external components—typically just input and output capacitors. For battery-powered operation, this represents an effective solution combining high efficiency, zero switching noise, small size, and low cost.

When the power source switches from battery to USB, however, the higher 5 V input voltage creates a substantial efficiency penalty. The same linear regulator now operates with a 1.7 V drop across the pass element, reducing efficiency to the range of 60%–65%. For a typical portable instrument drawing 500 mA at 3.3 V output, the regulator must now dissipate 900 mW as heat when powered from a 5 V USB source—more than tripling the thermal load compared to battery operation (see Figure 1). This heat dissipation creates thermal management challenges in compact enclosures, forces reduced battery charging rates, and in some cases requires limiting output power to prevent thermal shutdown.

Conceptual illustration showing a linear regulator dissipating significant heat when supplying 3.3 V from multi-input sources, with much greater thermal load from a 5 V USB source than from a 3.7 V battery.
Figure 1. Heat generation is a critical drawback of LDOs for supplying 3.3 V from multi-input sources.

The straightforward response to this efficiency problem is to avoid using linear regulators when the input-output differential becomes large. Switching regulators can handle the 5 V to 3.3 V conversion with 85%–90% efficiency, making them the standard choice for USB-powered operation. However, their switching noise conflicts with the noise requirements of precision analog circuitry. This tension between efficiency and noise performance has driven many designs toward a compromise approach.

To address this dilemma, designers often opt for a two-stage power architecture: a buck switching regulator handles the primary conversion from 5 V down to an intermediate voltage (3.8–4.3 V), followed by an LDO that provides the final step to 3.3 V while filtering switching noise. This achieves a respectable overall efficiency of 70%–75% from 5 V input while maintaining adequately low output noise.

This two-stage approach, however, carries significant overhead. The buck converter requires an inductor, controller IC, compensation network, and EMI-aware layout with ground plane partitioning—all of which expand board area and design complexity. Even with proper filtering, switching harmonics can reach the output and create intermodulation products with sensitive signals. For noise-critical applications, the switching pre-regulator is a thermal necessity rather than an optimal solution.

The ELR Approach

Polaris Semiconductor’s enhanced linear regulators (ELRs) address this problem at its root. Rather than adding a switching pre-regulator to compensate for the LDO’s efficiency limitation, ELRs use integrated GaAs photovoltaic-output optocouplers to recycle energy that a conventional LDO would dissipate as heat. The result is a linear regulator that exceeds the conventional VOUTV_{\mathrm{OUT}}/VINV_{\mathrm{IN}} efficiency ceiling—without introducing any switching noise.

For dual- and variable-source applications, a single ELR can operate in two modes—an optocoupler-enhanced mode for higher input voltages and a conventional bypass mode for lower input voltages—automatically selecting the optimal mode based on input voltage. This adaptive approach delivers the noise performance of a standalone LDO, the efficiency advantages of a two-stage architecture, and the simplicity of a single-device solution. Table 1 summarizes the comparison.

Table 1. Qualitative comparison of power management approaches for 3.3 V output from dual USB/battery sources.
ParameterLDOBuck + LDOAdaptive ELR
3.7 V EfficiencyHighMed-HighHigh
5.0 V EfficiencyLowMed-HighMed-High
Output Noisea<1 µVRMSFilter-dependent<1 µVRMS
Switching ContentNonePresentNone
Component CountLowHighLow
Board AreaSmallLargeSmall
Design ComplexityLowHighLow
  1. a 10 Hz–100 kHz

The remainder of this application note explains how the two ELR operating modes work, describes a demonstration circuit that switches between them automatically, and presents measured efficiency, transient, and thermal performance across the full input voltage range.

How Enhanced Linear Regulation Works

While any of Polaris Semiconductor’s step-down enhanced linear regulators can be applied to multi-input performance optimization, this application note uses the BK301D33L for demonstration purposes. The device co-packages proprietary GaAs photovoltaic-output optocouplers with the Analog Devices LT3045, selected for its exceptional noise performance in precision analog applications. BK301D33L retains these best-in-class noise and power supply ripple rejection (PSRR) properties (see Figure 2), while significantly improving the electrical efficiency. The ELR technology is described in detail in white paper PSWP001 [1]; this section provides the conceptual overview needed to understand the adaptive dual-mode design.

Figure 2. PSRR (top) and noise spectral density (bottom) of BK301D33L operating in its enhanced mode, matching performance in bypass mode (not pictured).

Optocoupler Energy Recycling

In a conventional LDO, all input current flows through the series pass element and exits at the output. The voltage difference between input and output, multiplied by this current, is dissipated as heat. The efficiency ceiling is therefore VOUTV_{\mathrm{OUT}}/VINV_{\mathrm{IN}}\approx 66% for a 5 V to 3.3 V conversion.

The BK301D33L breaks this ceiling by providing a parallel current path through integrated optocouplers. Input current passes through an internal LED array, generating photons that photovoltaic (PV) cells capture and convert back to electrical current. This recycled current flows directly to the output, supplementing the current delivered through the pass element. Because the PV cells deliver current directly to the output, the regulator’s input current is less than the load current—the balance is supplied by the optocoupler path. The optocouplers achieve current transfer ratios exceeding 70%, enabling them to recycle a substantial fraction of the energy that would otherwise be lost as heat. However, the forward voltage of the LED array increases the minimum input headroom required for regulation—a trade-off that motivates the two operating modes described next.

Simplified block diagram of the BK301D33L showing the LDO pass transistor, the internal LED array between the BYP pin and the output, and four series photovoltaic cells from OUT,PV to ground, together with the external switch that selects between enhanced and bypass operating modes.
Figure 3. Simplified block diagram of BK301D33L with application facilitating switchover between enhanced and bypass modes. Power good, undervoltage lockout, and current limit functionalities are not used in this example.

Enhanced Mode vs. Bypass Mode

Figure 3 shows the device architecture and how the two operating modes are implemented. The BYP pin connects internally to both the collector of the pass transistor as well as the anodes of the LED array, while the LED cathodes connect to the output pin. On the photovoltaic side, four devices are configured in series with their positive terminal at OUT,PV and their negative terminal at ground. This internal topology enables the device to operate in two distinct configurations through external pin connections.

Enhanced Mode. OUT and OUT,PV are connected together while an external switch between BYP and OUT remains open. Current flows through the LED array, generating photons that the PV cells convert to useful current supplementing the output. The device can be operated in this mode permanently through a fixed pin configuration (by floating BYP and connecting OUT to OUT,PV), or switched into this mode dynamically—as demonstrated in this application note.

However, enhanced mode imposes a minimum input voltage requirement. The voltage dropped across the LED array, combined with the LDO’s dropout voltage and the output voltage, establishes a threshold below which enhanced mode cannot regulate. For the BK301D33L producing 3.3 V output, this threshold is in the range of 4.6–5 V depending on load current (see Figure 6). Enhanced mode is therefore well-suited for USB power at 5 V but cannot support Li-Ion battery operation (for VOUTV_{\mathrm{OUT}} = 3.3 V).

Bypass Mode. The external switch closes to connect BYP directly to OUT. This shorts the LED array and configures the pass transistor to drive the output directly, just as in a conventional LDO. The PV cells no longer contribute current, and the device operates with the low dropout characteristics of a traditional linear regulator. Efficiency in this configuration follows the familiar VOUTV_{\mathrm{OUT}}/VINV_{\mathrm{IN}} relationship. Bypass mode operates efficiently at input voltages as low as 3.5 V (for VOUTV_{\mathrm{OUT}} = 3.3 V), making it appropriate for Li-Ion battery operation throughout the discharge curve.

The adaptive approach combines both modes: bypass mode at low input voltages where it operates efficiently, and enhanced mode at high input voltages where it recaptures energy that would otherwise be wasted as heat. The following sections describe the circuit that achieves this automatic transition and present measured performance across the full input voltage range.

Flow diagram of the adaptive dual-mode implementation, with solid lines tracing the power path through the power multiplexer and ELR and dashed lines tracing the signal and logic path through the threshold detection and load-switching blocks.
Figure 4. Adaptive dual-mode implementation flow diagram. Solid lines represent power path. Dashed lines represent signal/logic path.
Photograph of the adaptive ELR demonstration board, showing the BK301D33L regulator, the TPS2117 power multiplexer, the TLV4041 comparator, the load-switch MOSFET, and the BATT IN and USB IN connectors.
Figure 5. Adaptive ELR demonstration board.

Adaptive Dual-Mode Implementation

The flow diagram (Figure 4), demonstration board (Figure 5), and schematic in Figure A1 implement automatic mode switching based on the input voltage, allowing the regulator to seamlessly optimize efficiency as the power source changes between battery and USB.

The control circuit consists of three main functional blocks (Figure 4): input source selection with threshold detection, operating mode threshold detection, and MOSFET load switching. Proper selection of switching thresholds requires understanding the relationship between input voltage, load current, and the regulator’s operating modes.

Operating Boundaries

Figure 6 illustrates the key voltage thresholds and operating boundaries for the BK301D33L in this application. Two distinct threshold detection circuits govern the system’s behavior: one internal to the power multiplexer (THMUX\mathrm{TH_{MUX}}) that determines input source selection, and one external (THCOMP\mathrm{TH_{COMP}}) that determines whether the regulator operates in bypass or enhanced mode.

The curve labeled ELR VONV_{\mathrm{ON}} in Figure 6 represents the minimum input voltage at which the enhanced linear regulator can maintain output regulation at a given load current (25 °C operation). This threshold increases with load current because higher currents require greater forward voltage across the internal LED array, in addition to increasing LDO dropout voltage. At 50 mA, ELR VONV_{\mathrm{ON}} is approximately 4.6 V; at 500 mA, it rises to nearly 5.0 V. If the input voltage falls below ELR VONV_{\mathrm{ON}} while in enhanced mode, the regulator enters dropout and the output voltage decreases—behavior analogous to a conventional LDO operating with insufficient input voltage.

The VONV_{\mathrm{ON}} threshold also varies with temperature. At 300 mA, BK301D33L measurements show VONV_{\mathrm{ON}} decreasing at approximately −0.36 mV/°C, corresponding to a shift of approximately ±9 mV over a 0 °C to 50 °C operating range. This variation is modest but should be accounted for when selecting the operating mode switching threshold. As the relationship varies with load current, consult our datasheets or speak with an application engineer to determine the appropriate margin for your operating conditions.

Plot of input voltage thresholds versus load current for the adaptive mode switching circuit. The ELR V_ON curve rises from about 4.6 V at 50 mA to nearly 5 V at 500 mA; the V_DO plus V_OUT curve stays near 3.5 to 3.7 V; the TH_MUX and TH_COMP threshold bands are marked between them.
Figure 6. Operating thresholds and voltage boundaries for the adaptive mode switching circuit. ELR V_ON is the minimum input voltage for regulation at 25 °C in enhanced mode; V_DO+V_OUT is the minimum for bypass mode. Two threshold detection circuits govern system operation: TH_MUX (internal to the power multiplexer) controls input source selection, while TH_COMP (external comparator circuit) controls operating mode selection. The demonstration circuit is designed for maximum load currents below 300 mA.

The curve labeled VDOV_{\mathrm{DO}}+VOUTV_{\mathrm{OUT}} represents the minimum input voltage for bypass mode operation: the sum of the 3.3 V output and the pass element dropout voltage. This threshold is less sensitive to load current, remaining in the range of 3.5–3.7 V across the full load range.

Input Source Selection (THMUX)

Input source management is handled by the TPS2117 (U1) from Texas Instruments [2], a power multiplexer that automatically selects between two input sources. The device features an internal comparator that can be connected to the USB input through a voltage divider. When the USB voltage exceeds the internally-set threshold (THMUX\mathrm{TH_{MUX}} in Figure 6, approximately 4.5 V), the multiplexer switches from battery to USB power. This threshold-based approach ensures the multiplexer selects USB power only when the USB source is capable of supplying the load. The TPS2117 transitions smoothly between sources following a sub-20 µs break-before-make switchover delay, making it appropriate for applications where both sources may be present simultaneously, such as when a portable device charges while in use. The selected input voltage then feeds the BK301D33L regulator and the operating mode selection circuit.

As shown in Figure 6, THMUX\mathrm{TH_{MUX}} is set below ELR VONV_{\mathrm{ON}} across the full load current range. This means the system may briefly operate from USB power in bypass mode before the input voltage rises sufficiently to enable enhanced mode operation.

Operating Mode Selection (THCOMP)

The second threshold detection circuit uses a precision comparator (U3, TLV4041R1 from Texas Instruments [3]) configured with an external resistive voltage divider to determine when the input voltage crosses the boundary between bypass and enhanced operating modes. The shaded band labeled THCOMP\mathrm{TH_{COMP}} in Figure 6 indicates this threshold, including its hysteresis.

The divider network formed by R9 (31.6 kΩ) and R10 (10.2 kΩ) scales the input voltage down by a factor of approximately 4.1:1. This scaled voltage is compared against the TLV4041’s internal 1.2 V reference for state switching. The part features 20 mV of built-in hysteresis: the low-to-high transition occurs at 1.2 V, while the high-to-low transition occurs at 1.18 V. With the specified divider ratio, this translates to nominal switching thresholds of 4.92 V (rising) and 4.84 V (falling). In practice, the transitions were found to occur at 4.90 V (rising) and 4.83 V (falling), well-within the tolerance range of U3 and the associated passives (see Figure 7).

For reliable operation, THCOMP\mathrm{TH_{COMP}} must be set above ELR VONV_{\mathrm{ON}} at the maximum expected load current. As shown in Figure 6, the demonstration circuit’s comparator threshold of approximately 4.9 V provides adequate margin for maximum load currents below approximately 300 mA. Applications requiring higher load currents would need to raise THCOMP\mathrm{TH_{COMP}} accordingly, though this reduces the input voltage range over which enhanced mode is active. Conversely, applications with lower maximum load currents could reduce THCOMP\mathrm{TH_{COMP}} to extend enhanced mode operation to lower input voltages.

Plot of adaptive ELR input current versus input voltage, with one trace for sweeping low-to-high (IT+) and another for high-to-low (IT−); the two traces form a hysteresis loop around the operating mode transition near 4.9 V.
Figure 7. Adaptive ELR input current versus input voltage behavior when sweeping from low-to-high (IT+) and high-to-low (IT−), illustrating the hysteresis in operating mode transition.

The comparator sensing network (R9/R10) draws approximately 120 µA from the input at 5 V. As discussed in the Component Selection section, this divider can be resized with higher-value resistors to reduce current draw at the expense of increased mode-switching delay.

For applications requiring different operating mode threshold voltages, the divider ratio can be adjusted by changing R9 and R10 values. The threshold voltage is given by:

Vthreshold=Vref×R9+R10R10V_{\mathrm{threshold}} = V_{\mathrm{ref}} \times \frac{R_9 + R_{10}}{R_{10}}

where VrefV_{\mathrm{ref}} is the comparator’s internal reference: 1.2 V for the positive-going input threshold and 1.18 V for the negative-going input threshold. To simplify the design, the comparator—as well as the power mux—is powered from the same rail that it monitors, eliminating the need for an additional supply.

Design Considerations for USB Voltage Tolerance

Standard USB bus voltage is specified as 4.75 V to 5.5 V [4]. At higher load currents, the ELR’s required headroom (VONV_{\mathrm{ON}}) may exceed the lower bound of this tolerance (4.75 V). The demonstration circuit thresholds are optimized to maximize Enhanced Mode efficiency for sources maintaining ≥ 4.9 V under load. For input voltages below this threshold, the device automatically transitions to Bypass Mode to maintain regulation. Designers wishing to extend Enhanced Mode operation to lower input voltages may adjust the resistive divider network to lower the switching threshold (THCOMP\mathrm{TH_{COMP}}), accepting that the maximum supportable load current in Enhanced Mode will be reduced.

Plot of output voltage and regulator efficiency versus input voltage at 200 mA load current, showing the adaptive ELR holding a regulated 3.3 V output while efficiency steps up as the device transitions from bypass to enhanced mode near 4.9 V.
Figure 8. Output voltage and regulator efficiency versus input voltage at 200 mA load current.
Table 2. Input current required to supply 200 mA load current for V_OUT = 3.3 V.
Input current (mA)
@ VIN=3.7 V@ VIN=5.0 V
Conventional LDO208208
Enhanced Mode ELRn/a180
Adaptive ELR208180

Load Switching

The comparator drives a P-channel MOSFET switch (Q1, DMG2305UX-7) that controls the connection between the BYP and OUT pins. The TLV4041 features a push-pull output that actively drives both high and low states. When operating from battery power at approximately 3.7 V, the scaled voltage at the comparator’s input remains below 1.2 V, causing the comparator output to drive low. This pulls the gate of Q1 negative relative to its source, turning the P-channel MOSFET on and connecting BYP to OUT to engage bypass mode. When USB power is present at 5 V, the scaled voltage exceeds 1.2 V, and the comparator output drives high, turning Q1 off and disconnecting BYP from OUT to enable enhanced mode operation.

Resistor R12 (330 kΩ) provides a weak pull-up to define the MOSFET gate state during the comparator’s power-up sequence, defaulting the switch to the open (Enhanced Mode) state. Designers prioritizing a default-closed (Bypass Mode) startup condition may substitute a pull-down resistor. The high resistance value ensures negligible impact on the comparator’s push-pull drive capability once normal operation begins. Gate resistor R11 (24.9 Ω) limits the transition slew rate, reducing inrush current and preventing potential oscillation during threshold crossings.

Component Selection and Design Tradeoffs

Component selection for the adaptive switching circuit prioritizes fast response, simplicity, and low power consumption. The TPS2117 and TLV4041 consume less than 4 µA of quiescent supply current combined. Total steady-state current drawn by the adaptive switching circuit—including resistive voltage dividers—is approximately 120 µA from battery and 400 µA when USB is connected, with the U1 priority-pin divider (R6/R7) and U3 mode-switch divider contributing the majority of the total current draw (combined approximately 90 µA from battery source and 350 µA when USB is connected).

The DMG2305UX P-channel MOSFET offers low gate charge for fast switching with the micropower comparator, a 20 V voltage rating to accommodate the full input range, and on-resistance sufficiently low to minimize voltage drop in bypass mode at the current levels of this application (approximately 20 mV drop at 250 mA load current).

Reducing current consumption. For applications that do not require as fast a mode-switch response, the resistor divider network impedance can be increased to significantly reduce current consumption. A variant of the demonstration board assembled with R9 = 3.57 MΩ and R10 = 1.15 MΩ maintained approximately the same operating mode threshold transitions while reducing total board-level steady-state overhead current from approximately 400 µA maximum to 280 µA. The additional threshold setpoint error was not significant despite the orders-of-magnitude increase in resistor values. However, care should be taken to minimize parasitic capacitance at the comparator input node when selecting such high-impedance resistors, as this increases the RC time constant and the comparator’s effective propagation delay. In practice, the propagation delay increased by approximately 6 µs with the high impedance resistor network, consistent with approximately 7 pF stray capacitance for a Thévenin equivalent source resistance of 0.87 MΩ.

The other key driver of total board-level overhead current is the resistor divider formed at the priority input of TPS2117, which accounts for approximately 230 µA of the remaining 280 µA. This resistor divider can be similarly adjusted to reduce current consumption if desired.

Measured Performance

Figure 8 shows the measured output voltage and efficiency of the adaptive ELR at 200 mA load current. The dotted trace shows a conventional LDO’s efficiency following the VOUTV_{\mathrm{OUT}}/VINV_{\mathrm{IN}} relationship, declining steadily as input voltage increases. The dash-dot trace shows the BK301D33L output voltage operating exclusively in enhanced mode. Below approximately 4.8 V, the output voltage collapses as the device enters dropout, illustrating that enhanced mode alone cannot support Li-Ion battery operation for 3.3 V output.

The solid traces show the adaptive ELR, which combines both operating modes. At input voltages below approximately 4.9 V, the device operates in bypass mode with performance matching the conventional LDO. When input voltage exceeds the switching threshold, the device transitions to enhanced mode, producing an immediate efficiency increase—from 63% to 73% at 5 V and 200 mA. Table 2 confirms the mechanism: input current drops from 208 mA in bypass mode to 180 mA in enhanced mode, reflecting the contribution of optocoupler-recycled current to the output. In relative terms, the BK301D33L’s enhanced mode represents a 15–18% efficiency improvement over a conventional LDO above the enhanced-mode turn-on voltage. In higher voltage systems, other BK30-series ELR parts can be leveraged for significantly greater performance enhancement (e.g. 12 V→5 V with BK300D50E is 70% higher relative to the LDO).

Figure 9 provides a more comprehensive view of BK301D33L’s relationship between adaptive ELR efficiency, input voltage, and load current. The thick white line delineates the boundary between bypass and enhanced operating modes1.

Together, Figures 89 demonstrate that the adaptive ELR maintains regulated output across the full range of desired input voltages while automatically optimizing for high efficiency. Both modes maintain output noise below 1 µVRMS (see Figure 2), and neither introduces switching frequency content—the output spectrum is indistinguishable from that of a conventional linear regulator in either operating mode.

Contour map of adaptive ELR demonstration board electrical efficiency as a function of output current and input voltage, with a thick white line marking the boundary between bypass and optocoupler-enhanced modes of operation.
Figure 9. Contour map of adaptive ELR demonstration board electrical efficiency as a function of output current and input voltage. A thick white line delineates the operating mode switch between bypass and optocoupler-enhanced modes of operation.
Oscilloscope capture of the voltage regulator input voltage, output voltage, and input current while a constant 3.7 V battery is held at BATT IN and a 5 V USB is rapidly connected and disconnected at USB IN; a light green hatch marks intervals of ELR mode operation.
Figure 10. Oscilloscope channels measuring the voltage regulator's (VR) input and output voltage as well as its input current. Measurement was taken while supplying a constant 3.7 V at the BATT IN terminal while rapidly connecting/disconnecting a 5 V USB at the USB IN terminal. Operating mode is designated by a light green hatch for ELR mode; no hatch/shading for conventional LDO mode.

Transient Response

To assess the Adaptive ELR’s behavior during power source transitions, oscilloscope recordings were captured during USB connect and disconnect events (see Figure 10). During these measurements, a 3.7 V DC source simulating a Li-Ion battery was continuously applied at the BATT IN terminals. Oscilloscope probes monitored the voltage at the ELR chip’s input (the power MUX output node, labeled VRIN\mathrm{VR_{IN}}), the ELR output voltage (VROUT\mathrm{VR_{OUT}}), and the ELR input current. The USB IN voltage served as the trigger event and is indicated by the vertical dashed lines marking USB connect and disconnect. Input current was monitored indirectly through the BK301D33L’s ILIM pin, which provides a scaled representation of regulator input current [7]. While this measurement does not capture total board-level input current, the steady-state difference is well below 1 mA. A series resistor at the USB input was implemented to monitor board-level input current. The adaptive ELR supplied a resistive load of nominally 16.5 Ω—200 mA at 3.3 V—for this experiment.

USB Connection. Prior to connecting USB power, input current registered 208 mA, consistent with bypass mode operation. Table 3 summarizes the timing of downstream events following USB connection. The TPS2117 power MUX asserted its ST (status) flag approximately 6 µs after the USB was connected—an open-drain output indicating that the priority power path (USB IN) is active. This sub-10 µs response time is consistent with the TPS2117 datasheet [2]. After an additional 18 µs, the voltage at the ELR input reached the comparator’s switching threshold. The operating mode transition completed approximately 26 µs after the initial USB connection event, at which point input current settled to 180 mA, confirming enhanced-mode operation. Given the TLV4041’s specified propagation delay of a few hundred nanoseconds [3], the measured delay of 2 µs is likely an overestimate, given the uncertainty in the exact threshold value as well as measurement fidelity.

Table 3. Measured time between USB connect/disconnect and various downstream events.
Time Delay After USB Event
ConnectDisconnect
MUX ST Flag Change+6 µs+4 µs
VRIN Threshold Met+24 µs+12 µs
Load Switch+26 µs+14 µs

Throughout this sequence, the output voltage remained stable. The ELR’s 10 µF input capacitor (MLCC, X7R dielectric) maintained sufficient headroom for the LDO to regulate during the TPS2117’s break-before-make switchover. A board-level current inrush of up to 1.6 A was observed during the USB connection event, measured separately using a 20 mΩ series resistor (R13) at the board’s USB input; this inrush exceeded 1 A for approximately 6 µs. It is not visible in Figure 10 because the ILIM pin monitors only the ELR’s input current downstream of the power MUX. The inrush is associated with charging input capacitors (C1, C4) during the voltage transient. Applications targeting USB 2.0 compatibility may benefit from less input capacitance (discussion to follow), an NTC thermistor, or selecting a power MUX with adjustable output slew and/or current limiting (e.g. TPS2121 [8]). TPS2117 features conditional slew rate control, which is active only when the MUX’s output voltage is below 1 V; it was disabled in this measurement since the switchover condition was from 3.7 V to 5 V.

USB Disconnection. Compared to connection, the disconnect sequence exhibited faster timing (Table 3), with operating mode transition completed approximately 14 µs after the USB disconnect event. There were two minor output voltage perturbations during disconnect.

The first—±4 mV over less than 100 ns—occurred as the operating mode transitioned from enhanced to bypass. This perturbation arises because a finite delay exists between VRIN\mathrm{VR_{IN}} crossing THCOMP\mathrm{TH_{COMP}} and the bypass MOSFET actually engaging, during which the enhanced-mode ELR approaches dropout as its input voltage falls. As a result, regulator input current fell to as low as 134 mA (as the output capacitor discharged to support the load) to a peak of 368 mA (as the capacitor recharged once bypass mode operation was restored). The inrush exceeded 300 mA for approximately 250 ns.

The second perturbation occurred as the power MUX executed its break-before-make transition from USB back to battery power. During this interval, VRIN\mathrm{VR_{IN}} fell to as low as 3.39 V causing the regulator—now in bypass mode—to enter dropout. Output voltage dropped 42 mV before recovering, with the total excursion lasting approximately 7 µs. During the transition, input current swung from as low as 143 mA to a peak of 218 mA.

Side-by-side oscilloscope captures of the voltage regulator input voltage, output voltage, and input current during USB disconnection events; the left panels use a 10 µF ELR input capacitor and the right panels use a 4.7 µF input capacitor, with larger transient excursions for the smaller capacitor.
Figure 11. Oscilloscope channels measuring the voltage regulator's (VR) input and output voltage as well as its input current during USB disconnection events. Left panels are for the adaptive board with ELR C_IN = 10 µF; right panels use ELR C_IN = 4.7 µF.
Three FLIR thermal images of the adaptive ELR demonstration board at 3.3 V output and 500 mA, sharing a fixed color scale: the left panel at 3.7 V input, the top right panel at 5 V input forced into conventional LDO mode (hottest), and the bottom right panel at 5 V input in enhanced mode (noticeably cooler).
Figure 12. FLIR thermal images of adaptive ELR demonstration board operating at output voltage and current of 3.3 V and 500 mA, respectively. Input voltage is either 3.7 V (left panel) or 5 V (right panel). At V_IN=5 V, images are captured both when forcing the regulator into conventional LDO mode (top right panel) and when enabling enhanced mode (bottom right panel). Color scale is fixed across all three images to emphasize enhanced mode's impact on thermal dissipation.

A note on input capacitor selection. For applications that do not require as tight output voltage regulation during source transitions, the ELR’s input capacitor value can be reduced to the minimum recommended value of 4.7 µF. As shown in the USB disconnection events of Figure 11, the output voltage excursion was larger when reducing the value of C1, but it was still a relatively minor event—with the largest deviation from nominal lasting 14 µs with output voltage falling as low as 3.16 V before recovering. The transient swings in ELR input current were also more significant when the ELR’s CINC_{\mathrm{IN}} = 4.7 µF as a result of more substantial discharging and charging of the output capacitor. During the USB connection event, the output voltage remained stable regardless of the ELR’s CINC_{\mathrm{IN}} value (not pictured). The lower CINC_{\mathrm{IN}} of 4.7 µF has an additional benefit of reducing the board-level inrush current during USB connection.

A note on output capacitor selection. The transient behavior during break-before-make transitions depends on the output capacitor’s ESR and capacitance. The 10 µF MLCC (X7R) used in this design has very low ESR, which is favorable for minimizing voltage excursions. If a designer substitutes a higher-ESR capacitor (e.g., tantalum or electrolytic), the voltage perturbations during source transitions will increase. Low-ESR (<20 mΩ) MLCCs are recommended for use with the BK30 series of ELRs.

A note on mode boundary stability. The 70 mV hysteresis band (4.90 V rising / 4.83 V falling) prevents mode toggling under normal operating conditions. However, if the input voltage is near THCOMP\mathrm{TH_{COMP}} and the source impedance is high, load transients can momentarily depress the input voltage. The R9/R10 divider impedance directly affects how susceptible the comparator input is to such transient-induced excursions. For applications operating near the mode boundary, verify that the source impedance and load transient magnitude do not produce input voltage swings exceeding the hysteresis band.

Thermal Performance

Table 4. Thermal and electrical performance of the adaptive ELR demonstration board at V_OUT = 3.3 V, I_LOAD ≈ 0.5 A.
VIN=3.7 VVIN=5.0 V
BypassBypassaEnhanced
TMAX35.0 °C57.0 °C45.5 °C
IINb524 mA524 mA456 mA
η86%63%73%
PDISS276 mW957 mW617 mW
  1. a The adaptive ELR was forced into bypass mode at VIN=5.0 V by grounding the comparator’s input.
  2. b Measured input current values suggest the nominally 6.6 Ω load resistor was slightly lower in practice. This has been accounted for in efficiency and dissipated power calculations.

Thermal imaging provides a direct visualization of enhanced mode’s benefit at higher load currents. In the experiment shown in Figure 12, the ELR supplied 3.3 V into a nominally 6.6 Ω chassis resistor. Although the demonstration circuit’s mode-switching thresholds are optimized for load currents below 300 mA, this thermal evaluation was conducted at the ELR’s 500 mA maximum rating to illustrate the upper bound of thermal dissipation savings. To ensure valid enhanced-mode operation at this elevated current level, a stiff 5.0 V source was maintained at the USB input, ensuring the voltage remained strictly above the VONV_{\mathrm{ON}} requirement. Three operating conditions were captured with a fixed color scale across all images to facilitate direct comparison.

At VINV_{\mathrm{IN}} = 3.7 V, the ELR operated in bypass (LDO) mode with high efficiency (η\eta\approx 86%), registering a maximum board temperature (TMAXT_{\mathrm{MAX}}) of 35.0 °C and estimated power dissipation of 276 mW (see Table 4). For VINV_{\mathrm{IN}} = 5 V with the adaptive function disabled—achieved by shorting the comparator input to ground through the high-impedance divider, with negligible effect on current draw—the regulator operated in conventional LDO mode and registered TMAXT_{\mathrm{MAX}} = 57.0 °C with 957 mW dissipated. Upon re-enabling the adaptive function, the regulator entered enhanced mode, reducing TMAXT_{\mathrm{MAX}} to 45.5 °C and dissipation to 617 mW. This represents a 36% reduction in both power dissipation as well as peak temperature rise above ambient.

Conclusions

This application note demonstrated an adaptive dual-mode regulation technique using the Polaris Semiconductor BK301D33L enhanced linear regulator to maintain high efficiency across input voltages from 3.5 V to 5.5 V—the full range spanned by single-cell Li-Ion and USB power sources. By automatically switching between optocoupler-enhanced and conventional LDO operating modes based on input voltage, a single regulator replaces the two-stage buck-plus-LDO architectures traditionally required for efficient, low-noise multi-source power management.

At 200 mA output current, the adaptive ELR achieved 86% efficiency from a 3.7 V battery source in bypass mode and 73% from a 5 V USB source in enhanced mode, versus 63% for a conventional LDO at 5 V. At the device’s maximum rated current of 500 mA, enhanced mode delivered 73% efficiency from 5 V—reducing power dissipation from that of conventional LDO operation—957 mW—down to 617 mW, with a corresponding 36% reduction in peak board temperature rise above ambient. Both operating modes maintained output noise below 1 µVRMS (10 Hz to 100 kHz) with no switching-frequency content, preserving the spectral purity that makes linear regulation essential in precision analog systems. Power source transitions completed within 30 µs with output voltage excursions recoverable within microseconds and manageable through straightforward capacitor sizing.

The adaptive switching circuit adds minimal overhead: a power multiplexer, a micropower comparator, a single MOSFET, and a handful of passive components. The steady-state current overhead is approximately 120 µA when operating from battery alone, rising to approximately 400 µA when USB is connected—dominated by the resistive voltage dividers on the power MUX’s priority input and external comparator’s input, which can both be resized for lower current draw. This compares favorably to the inductor, controller IC, compensation network, and EMI-aware layout demanded by a buck pre-regulator stage—resulting in a meaningful reduction in board area, component count, and design complexity.

While the dual USB/battery scenario presented here is a natural application for this technique, the underlying principle—dynamically selecting between enhanced and bypass operating modes based on input voltage—applies to any system where the input voltage range spans the ELR’s enhanced-mode operating threshold. Systems powered by variable-voltage adapters, energy-harvesting sources, or wide-tolerance industrial rails can benefit from the same approach when paired with the appropriate Polaris Semiconductor ELR for the target output voltage.

Appendix

Multi-Input Performance Optimizer Schematic

Full schematic for the Multi-Input Performance Optimizer, showing the BK301D33L ELR, the TPS2117 power multiplexer with its priority divider, the TLV4041 comparator with its R9/R10 sensing divider, the DMG2305UX load-switch MOSFET, the BATT IN and USB IN inputs, and the supporting passives and test points.
Figure A1. Schematic for Multi-Input Performance Optimizer.

Multi-Input Performance Optimizer Bill of Materials

Table A1. Bill of Materials for the Demonstration Board.
IDDescriptionPartManufacturer
C1, C310 µF ceramic X7RGRM31CR71E106KA12KMurata Electronics
C24.7 µF ceramic X7RGRM31CR71H475KA12LMurata Electronics
C4, C61 µF ceramic X7RCL21B105KBFNNNESamsung Electro-Mechanics
C5100 nF ceramic X7RCL10B104KB8NNWCSamsung Electro-Mechanics
J1–J6Terminal Turret Connector2501-2-00-80-00-00-07-0Mill-Max Manufacturing Corp.
J7CONN RCP USB2.0 TYP CUSB4105-GF-AGCT
Q1MOSFET P-CH 20 V 4.2 ADMG2305UX-7Diodes Incorporated
R1200 kΩ 5% 1/8 WCRCW0603200KJNEAVishay Dale
R233 kΩ 0.1% 1/8 WRG2012P-333-B-T5Susumu
R3442 kΩ 1% 1/8 WCRCW0603442KFKEAVishay Dale
R449.9 kΩ 1% 1/8 WCRCW060349K9FKEAVishay Dale
R5249 Ω 1% 1/8 WCRCW0603249RFKEAVishay Dale
R616.9 kΩRR1220P-1692-D-MSusumu
R74.87 kΩRR0816P-4871-D-67HSusumu
R8, R1010.2 kΩRR0816P-1022-D-02CSusumu
R931.6 kΩRR0816P-3162-D-49CSusumu
R1124.9 ΩRR0816Q-24R9-D-39RSusumu
R12330 kΩ 1% 1/8 WCRCW0603330KFKEAVishay Dale
R1320 mΩ 1% 1/4 WRCWE080520L0FMEAVishay Dale
R14, R155.1 kΩRC0603FR-075K1LYageo
TP1, TP3, TP4, TP6, TP12Multipurpose Test Point5010Keystone
TP2, TP13Multipurpose Test Point5011Keystone
TP5, TP7–TP11Multipurpose Test Point5012Keystone
U1Power Switch/DriverTPS2117DRLRTexas Instruments
U22–4 V ELRBK301D33LPolaris Semiconductor
U3IC COMPARATORTLV4041R1YKARTexas Instruments
Component placement drawing for the demonstration board, showing reference designators and footprints for the ELR, the power multiplexer, the comparator, the load-switch MOSFET, and the passives.
(a) Component locations
Front copper, solder paste and silkscreen composite layer for the demonstration board.
(b) Front copper, paste, and silkscreen
Rear copper layer for the demonstration board, showing the ground pour and routed traces.
(c) Rear copper
Figure A2. Component locations, front copper and silkscreen, and rear copper and silkscreen layout for the demonstration board.

References

  1. M. P. Lumb, K. J. Schmieder, J. DeLombard, J. Carlin, L. Kaliszewski, A. Price, D. Hollingshead, and T. Grassman, “Breaking the rules of linear regulators: High efficiency buck and boost conversion without switching.” Polaris Semiconductor; White Paper PSWP001, 2025. Available: https://www.polarissemiconductor.com/techdocs/#WP
  2. TPS2117 1.6-V to 5.5-V, 4-A low IQ power mux with manual and priority switchover. Texas Instruments; Datasheet, 2023. Available: https://www.ti.com/lit/ds/symlink/tps2117.pdf
  3. TLV40x1 small-size, low-power comparator with precision reference. Texas Instruments; Datasheet, 2021. Available: https://www.ti.com/lit/ds/symlink/tlv4041.pdf
  4. Universal serial bus 3.2 specification. USB Implementers Forum; Revision 1.1, 2022. Available: https://www.usb.org/document-library/usb-32-revision-11-june-2022
  5. Universal serial bus power delivery specification. USB Implementers Forum; Revision 3.2, Version 1.1, 2024. Available: https://www.usb.org/document-library/usb-power-delivery
  6. Universal serial bus Type-C cable and connector specification. USB Implementers Forum; Release 2.4, 2024. Available: https://www.usb.org/document-library/usb-type-cr-cable-and-connector-specification-release-24
  7. LT3045 20V, 500mA, ultralow noise, ultrahigh PSRR linear regulator. Analog Devices; Datasheet, 2022. Available: https://www.analog.com/media/en/technical-documentation/data-sheets/lt3045.pdf
  8. TPS212x 2.8-v to 22-v priority power MUX with seamless switchover. Texas Instruments; Datasheet, 2020. Available: https://www.ti.com/lit/ds/symlink/tps2121.pdf

Footnotes

  1. The slight shifts in this boundary as a function of load current are measurement artifacts due to the limited precision and granularity of voltage steps. In practice, the operating mode changeover is dictated solely by input voltage and is invariant with respect to load current.

Get started

Ready to design it in?

Request evaluation samples, detailed datasheets, and SPICE models — an engineer answers, not a sales funnel.