LDO vs DC-DC: efficiency, noise and external-part count
A practical decision guide across three axes — dropout loss, output noise and PSRR, and external parts / board area — including the pre-regulate-then-filter pattern and common layout mistakes.
The short answer: this is not a "which is better" question. Three axes decide it — how large the voltage difference is, how much current the load draws, and how sensitive the downstream circuit is to noise. Cross any one of them and the choice is essentially made for you.
Axis 1: dropout loss and efficiency
An LDO regulates linearly, so the voltage difference multiplied by load current becomes heat. Efficiency is roughly Vout / Vin: dropping 12 V to 3.3 V gives about 28%, with 72% turned into heat. At 8.7 V of dropout and 0.5 A, the LDO dissipates 4.35 W — far beyond a SOT-23 package, so a larger package or heatsinking is mandatory.
A DC-DC converter switches, typically reaching 85%–95% efficiency, and wins clearly when the voltage difference or the current is large. The price is an inductor, input/output capacitors and more careful layout, which raises board area and BOM cost.
Axis 2: output noise and PSRR
- LDO output noise can be in the microvolt range with high low-frequency PSRR — the right choice for analog front ends, RF, audio and ADC/DAC references.
- DC-DC output ripple is commonly in the tens of millivolts and shaped by switching frequency and loop design; the switch node is the main EMI source and will couple into sense lines if layout is poor.
- When you need DC-DC efficiency but also low noise, the standard pattern is pre-regulate then filter: let the DC-DC drop the voltage to slightly above target and use an LDO as the final stage.
Axis 3: external parts and board area
- LDO: usually one or two ceramic capacitors, minimal routing, high tolerance for layout mistakes and fast to debug.
- DC-DC: power inductor plus input/output capacitors plus feedback divider; the inductor must be chosen for saturation current and DCR, and the feedback trace must stay away from the switch node.
- For low volume, tight mechanical space or fast prototyping, the engineering cost of an LDO often outweighs the power it wastes.
Match the scenario
| Scenario | Choice | Why |
|---|---|---|
| Battery powered, large dropout, >300 mA | DC-DC | Linear dissipation is unacceptable; efficiency directly becomes runtime |
| 5 V to 3.3 V, <300 mA | LDO | Small dropout, manageable loss, simplest circuit |
| Downstream is ADC / RF / audio | LDO, or DC-DC + LDO | Noise and PSRR outrank efficiency |
| Wide or fluctuating input range | DC-DC (wide-Vin) | A linear part cannot survive the dropout |
| Many rails, each with small current | LDO array | Saves inductors and keeps layout clean |
Common mistakes
- Not calculating LDO heat: estimate (Vin − Vout) × Iout and compare against the package thermal resistance instead of only checking current rating.
- Routing the DC-DC feedback trace next to the inductor: take feedback directly from the output capacitor and keep it away from the switch node.
- Splitting the ground plane: join the switching loop ground and signal ground at a single point, or the return path breaks and noise appears out of nowhere.
- Floating enable pins: tie unused EN pins to a defined level per the datasheet, otherwise power-up sequencing becomes unpredictable.
- Ignoring inrush: large capacitive loads draw a surge at startup that can trip current limit and cause failed or repeated startup.
For a specific power solution, browse the Power & Charging category or send your input range, load current and noise requirement to our FAE for a selection review.
Send the part numbers and operating conditions; our FAE replies within 1 business day.