LDO vs Buck Converter: Which Should You Use?
Use an LDO when the voltage drop and current keep heat small and simplicity or low noise matters. Use a buck when the step-down ratio, current or battery target makes linear loss unacceptable. Use both when an efficient first conversion must feed a sensitive local rail.
The answer in one calculation
First calculate (VIN,max − VOUT) × IOUT,max. If the resulting LDO heat cannot be removed while keeping junction temperature within the selected device’s limits, stop: a buck or another switching topology is the practical starting point. If the heat is small, compare noise spectrum, quiescent current, transients, footprint, cost and validation effort.
LDO: best when input sits close to output, current is modest and a simple low-noise rail has real value.
Buck: best when a large voltage drop, higher current or battery runtime makes efficiency decisive.
Buck + LDO: useful when the large drop needs switching efficiency but the final load needs cleaner local power.
Neither is universal: dropout, minimum on-time, mode changes, stability and thermal conditions can overturn the headline choice.
Scope: This guide covers non-isolated step-down rails. It is not a safety design for mains, medical, traction, high-energy or other regulated power systems.
LDO vs buck converter at a glance
The category labels describe the usual trade-off, not a guarantee for every modern device. Compare exact parts at the actual voltage, load, temperature and operating mode.
Swipe the table horizontally to compare all columns.
| Decision factor | LDO | Buck converter | What must be verified |
|---|---|---|---|
| Energy conversion | Excess voltage is dissipated in the pass device. | Switches and an inductor transfer energy to the load. | Worst-case input power and loss. |
| Efficiency ceiling | Approximately VOUT/VIN when IQ is small. | Not limited by that voltage ratio; depends on real loss mechanisms. | Datasheet curve at the exact operating point. |
| Noise and EMI | No switching node; can be very low noise, but PSRR is finite. | Ripple, harmonics and switch-node coupling require control. | Spectrum, probing method, layout and load sensitivity. |
| External parts | Usually input/output capacitors and a few optional parts. | Inductor plus capacitors and possibly compensation, filter or snubber. | Complete BOM, area, height and derating. |
| Light-load behavior | IQ and dropout can dominate. | PFM, pulse skipping or forced PWM changes efficiency and ripple. | Mission profile, shutdown current and mode transitions. |
| Design risk | Heat, dropout, capacitor stability and reverse current. | Minimum on-time, inductor saturation, EMI and loop/layout behavior. | Prototype across voltage, load, temperature and tolerance. |
Why the two regulators lose power differently
An LDO controls a pass transistor
The feedback loop moves a transistor between input and output until the sensed output matches a reference. The voltage difference appears across that transistor. With load current flowing through it, the unused electrical power becomes heat.
Dropout is the minimum headroom the pass stage needs to maintain regulation at stated load and conditions. Once headroom disappears, the output follows the falling input instead of remaining regulated.
A buck alternates energy paths
The high-side switch charges the inductor; the low-side switch or diode provides a current path during the other phase. The output capacitor reduces ripple while feedback adjusts duty cycle.
The ideal continuous-conduction duty ratio is approximately D = VOUT/VIN. Real loss comes from MOSFET resistance and switching transitions, gate drive, inductor DCR and core loss, capacitor ESR, controller current and PCB resistance. Analog Devices’ AN-140 explains both operating mechanisms and the 12V-to-3.3V duty-cycle example.
Screen direct-LDO loss before selecting a regulator
Enter the worst-case input, regulated output and continuous load. The tool compares an idealized LDO with a buck efficiency assumption and estimates LDO junction rise from a user-supplied thermal impedance.
LDO vs buck first-pass calculator
This is a screening aid, not a component guarantee. It omits LDO IQ, temperature-dependent dropout, transient thermal impedance and all converter-specific curves.
- LDO dissipation
- 0.870 W
- Idealized LDO efficiency
- 27.5%
- Estimated LDO rise
- 52.2°C
- Buck loss at assumed efficiency
- 0.037 W
- Output power
- 0.330 W
- Screening result
- Start with buck
The LDO result omits internal-current loss. Confirm actual package, PCB, ambient, efficiency curve and dynamic behavior.
Two voltage ratios lead to different answers
12V to 3.3V at 100mA
The load receives 0.33W. A direct LDO dissipates (12 − 3.3)×0.1 = 0.87W, before internal current, and has an idealized 27.5% efficiency ceiling. If the real board produced an effective 60°C/W junction-to-ambient path, the first estimate is a 52.2°C junction rise.
At an illustrative 90% buck efficiency, input power is 0.367W and converter loss is about 0.037W. The 90% and 60°C/W figures are assumptions, not claims for a part. Even so, the gap explains why a buck is the natural first architecture for this ratio.
3.6V to 3.3V at 20mA
A direct LDO dissipates approximately (3.6 − 3.3)×0.02 = 6mW, before internal current, with a 91.7% idealized efficiency ceiling. That small loss may make simplicity and low noise more valuable than the incremental conversion gain from a buck.
But at a 20µA sleep load, regulator current matters. TI lists the TPS7A20 with 6.5µA typical IQ, 7µV RMS output noise and a 300mA rating. Using only that typical IQ, a simplified 3.6V, 20µA calculation falls to roughly 69% efficiency. Maximum current, shutdown behavior and the complete duty cycle must drive a runtime commitment.
An LDO is not a universal ripple eraser
An LDO has no switching node of its own and can reach very low output noise. It still has internal reference and control noise, transient disturbances and finite power-supply rejection ratio (PSRR). PSRR measures how much input ripple is attenuated at a stated frequency and operating condition.
At 40dB, the voltage ratio is 100:1, so 100mV of input ripple would ideally become 1mV at the output—only if that PSRR applies at the real frequency, headroom, current, capacitor and temperature. A 1kHz headline cannot prove rejection at a 2MHz switching frequency or its ringing harmonics.
A buck adds switching ripple, harmonics, control-mode behavior and switch-node ringing. Layout and probing can change what appears on an oscilloscope. Define the sensitive band first: an ADC reference, RF synthesizer, image sensor and digital core do not judge the same spectrum in the same way. TI’s LDO Basics covers the dependence of dropout, capacitors, thermals, IQ, PSRR and noise on operating conditions.
Compare energy over the load profile, not one efficiency point
A product that draws 500mA for one second and 5µA for an hour cannot be judged from its active current alone. For every mode, estimate input power and multiply it by time. Then add the modes.
A buck can excel during a radio burst yet lose ground in long sleep because of controller current, feedback-divider current or power-save behavior. An ultralow-IQ LDO can win near the battery voltage in sleep, yet waste substantial energy during a high-current burst with a large voltage drop.
Modern bucks use PFM, pulse skipping or sleep modes for light-load efficiency. Those modes can add low-frequency ripple or tones. Forced PWM gives more predictable spectral content but may consume more current. TI lists the active TPS62160 as a 3V-to-17V, 1A synchronous buck with 17µA typical IQ, 100% maximum duty and light-load efficiency features. That is a documented example, not a universal comparison with any LDO.
Thermal area and power-stage area belong in the comparison
Do not treat θJA as a package constant
The common estimate TJ ≈ TA + PDθJA is a starting point. Published thermal resistance depends on the test board. Copper, vias, airflow, enclosure temperature, nearby components, orientation and power duration change the real path.
A buck lowers total loss but can concentrate it in the IC, inductor or diode. Check RMS and peak current, saturation, DCR, core loss, capacitor ripple current and hot-condition efficiency.
Compare the complete solution
An LDO may need only capacitors, but a hot design can require wide copper, thermal vias or a larger package. A buck adds an inductor and more switching parts, yet its lower heat can reduce the effective board and enclosure burden.
For sourcing, compare the regulator, inductor, capacitors, optional EMI parts, land pattern, maximum component height, approved alternates and lifecycle together. The smallest IC rarely defines the real footprint.
Buck plus LDO works only with deliberate headroom
A hybrid chain uses a buck for the large voltage drop and an LDO for final local regulation and filtering. A common planning example is 12V to 3.6V by buck, then 3.6V to a quiet 3.3V rail by LDO.
The LDO input must stay above VOUT + VDROPOUT,max through buck tolerance, ripple, wiring loss, load steps and temperature. Extra headroom can improve rejection for some LDOs but raises heat; less headroom improves efficiency but risks dropout.
If the buck is 90% efficient and the LDO stage is ideally 3.3/3.6 = 91.7%, the chain is roughly 82.5% efficient before LDO internal-current effects. That is far better than the 27.5% idealized ceiling of a direct 12V-to-3.3V LDO, but it is not automatically quieter unless the frequency-domain conditions are satisfied.
Start from the load, then verify the exception
Swipe the table horizontally to compare all columns.
| Application | Likely starting point | Evidence that decides it |
|---|---|---|
| 5V to 3.3V MCU at 20mA | LDO | Worst-case heat, sleep current, input minimum, dropout and load steps. |
| 12V to 3.3V at 100mA | Buck | Efficiency curve, minimum on-time, ripple, EMI, inductor and light-load mode. |
| Precision ADC or RF rail from 12V | Buck + low-noise LDO | Noise spectrum, PSRR at buck frequencies, headroom, transient and total heat. |
| Battery close to rail voltage | Low-IQ LDO or 100%-duty buck | Discharge curve, dropout, pass-through behavior, standby current and burst current. |
| High-current processor core | Synchronous buck | Transient response, sequencing, current limit, remote sensing and thermal design. |
| Always-on microamp sensor | Ultralow-IQ LDO or low-IQ buck | Duty cycle, shutdown current, PFM ripple, voltage ratio and battery range. |
A part substitution can become an architecture change
The cheaper regulator was not the cheaper approved solution
A contract manufacturer receives a BOM for a 12V-to-3.3V, 100mA sensor controller. The original design uses a buck. A buyer finds an available low-cost LDO with the same output and enough headline current, then proposes it as a substitute to avoid an inductor shortage.
The first loss calculation shows 0.87W in the LDO before internal current. That changes junction temperature, copper area and enclosure heat. Removing the switching stage may simplify EMI work, but the proposed part also changes startup, current-limit behavior, package, capacitors and fault response. It is not a drop-in substitute because the architecture and layout are different.
The useful procurement outcome is a controlled choice: source the exact buck or an engineering-approved buck alternate with the same validated power-stage constraints; redesign with an LDO only as a documented board revision. The case is composite, but the decision consequences come directly from the published loss mechanisms.
For exact manufacturer coverage and sourcing review, see YURUNOX resources for Texas Instruments components, Analog Devices components and quality-assurance evidence.
Use this order before approving the rail
- Define the real input envelopeInclude tolerance, ripple, startup, battery discharge, surge, reverse conditions and wiring drop.
- Define the load in time and frequencyRecord continuous, peak, sleep and burst currents; load-step edge rate; allowable droop; sequencing and sensitive noise bands.
- Eliminate impossible LDO casesCalculate loss at maximum input and load, then estimate junction temperature with the actual package and board assumptions.
- Compare energy, not only efficiencyUse every operating mode and its duration. Include IQ, shutdown current and converter mode transitions.
- Design the complete solutionVerify capacitors, inductor, saturation, feedback, compensation, filters, thermal copper, test points and component tolerances.
- Check sourceable exact partsLock package, grade, lifecycle, approved alternates and documentation. A regulator category is not a purchasable identity.
- Prototype at the cornersMeasure temperature, efficiency, startup, ripple, EMI and transient behavior across input, load, temperature and tolerance.
Input minimum/nominal/maximum · output voltage and tolerance · continuous/peak/sleep current · load profile · noise and transient limits · ambient and board thermal conditions · package/height · lifecycle · approved manufacturers · whether architecture or only part-number substitution is allowed.
Source the rail as a controlled power stage
Send the exact regulator model or the full electrical envelope, quantity, package, inductor and capacitor constraints, lifecycle, traceability and alternate-approval rules. YURUNOX can review the sourcing record without treating a different topology as a drop-in replacement.
Availability and suitability must be confirmed for the exact orderable parts and the released schematic, layout and BOM.
Questions engineers and buyers ask before choosing
Is an LDO always quieter than a buck converter?
No. An LDO has no switching node and can be extremely low noise, but it still has internal noise, finite PSRR and transient disturbances. A well-designed low-noise buck may meet the load directly. Compare the required spectrum and the measured power-distribution network.
How do I calculate LDO power loss?
Use approximately (VIN − VOUT)×IOUT + VIN×IQ. Apply worst-case voltage, load, internal current, temperature, package and board conditions. Include startup and fault states where relevant.
Why is LDO efficiency approximately VOUT divided by VIN?
When quiescent current is small, input and output current are nearly equal. Output power divided by input power then reduces to about VOUT/VIN; the remaining voltage is dissipated as heat.
Can a buck converter replace an LDO directly?
Sometimes at the functional level, but it adds an inductor, switching ripple, EMI, mode behavior and different startup and protection. It also changes the PCB. Verify the complete schematic, layout and load requirements before approval.
What is LDO dropout voltage?
It is the input-to-output headroom required to maintain regulation under specified load and conditions. Use the maximum dropout across required current and temperature, then add margin for ripple, sag and transients.
Which regulator is better for battery-powered products?
It depends on battery range and load profile. A buck often saves energy when input is far above output or active current is high. A low-IQ LDO can win when voltages are close and the product sleeps at tiny load for long periods.
Should I place an LDO after a buck converter?
Use the combination when the buck efficiently handles the large voltage drop and the LDO materially improves the sensitive rail. Verify headroom, PSRR at switching frequencies, total efficiency, heat, startup and stability.
Does a higher buck switching frequency make the design better?
Not automatically. Higher frequency can reduce inductor and capacitor size, but generally raises switching loss and can tighten minimum on-time and EMI constraints. Choose frequency from the full size, loss, noise, transient and thermal trade-off.
Sources and limitations
- Analog Devices AN-140: Basic Concepts of Linear Regulator and Switching Mode Power SuppliesLinear loss, buck operating modes, duty ratio and architecture trade-offs.
- Texas Instruments: LDO BasicsDropout, capacitors, thermal management, quiescent current, PSRR and noise.
- Texas Instruments: Basic Calculation of a Buck Converter’s Power Stage, Rev. BDuty cycle, inductor current, output capacitor and loss-aware power-stage calculations.
- Texas Instruments TPS7A20 product page and Rev. H datasheetCurrent exact-device LDO example: noise, IQ, dropout, PSRR and capacitor requirements.
- Texas Instruments TPS62160 product page and Rev. E datasheetCurrent exact-device synchronous-buck example: voltage range, current, IQ, light-load behavior and 100% duty operation.
Recheck the latest datasheet revision, lifecycle, efficiency and PSRR curves, package thermal conditions, external parts and layout guidance before design release or substitution. All unreferenced numerical comparisons above are transparent calculations or explicitly labeled assumptions, not measured YURUNOX results.