KEMI
/

KEMI Docs · Circuit Practice

LDO vs Buck Converter

Last updated: 2026-08-30

The first fork in any power design: linear (LDO) or switching (buck)? Only two variables actually decide it — the heat produced by voltage difference times current, and how noise-sensitive the load is. Here is how to decide with numbers.

1

One line each

An LDO is an automatically adjusted resistor that burns the excess voltage as heat — simple and quiet. A buck switches at hundreds of kHz to MHz, ferrying energy through an inductor instead of burning it — 85-95% efficient, at the price of switching noise and part count.

2

The thermal math that eliminates LDOs

LDO loss is one formula: P = (Vin − Vout) × Iout. From 12V to 3.3V at 500mA: P = 8.7 × 0.5 = 4.35W. A SOT-223's junction-to-ambient thermal resistance of roughly 60°C/W implies a 260°C rise — physically impossible without serious heatsinking. From 5V instead, P = 0.85W: manageable with copper pour as the heatsink. Large voltage difference × large current = LDO eliminated.

CaseLDO lossVerdict
5V → 3.3V, 100mA0.17WLDO fine — even SOT-23 class
5V → 3.3V, 500mA0.85WLDO possible with SOT-223 + copper heatsinking
12V → 3.3V, 500mA4.35WLDO impossible — buck required
12V → 5V, 2A14Wnot a discussion — buck
3.6V (Li-ion) → 3.3V, 200mA0.06WLDO ideal — quiet and cheap

3

Noise — where bucks can't go

Buck outputs carry switching ripple and harmonics. Digital logic doesn't care; ADC references, RF supplies, precision analog front-ends and audio paths do. The standard answer is two stages: buck down to target +0.3-0.5V for efficiency, then a low-noise LDO for the final clean-up (12V → buck → 3.6V → LDO → 3.3V). With so little voltage across it, the LDO's heat is negligible.

4

Cost, area, design difficulty

KEMI's default decision order

1) Compute the loss — above about 1W, buck. 2) Analog-sensitive load — buck + LDO two-stage. 3) Otherwise the cheap quiet LDO. When unsure, calculate first: watts decide, not intuition.

FAQ

What is LDO dropout?

The minimum input-output difference that keeps regulation. A 0.3V-dropout LDO making 3.3V needs at least 3.6V in — critical when running to the end of a battery discharge curve.

Is buck efficiency always 90%?

No — it is a curve versus load. Many parts fall off hard at light load (a few mA); battery devices should check the light-load (PFM) behaviour on the datasheet curve.

When do boost and buck-boost apply?

Input below output: boost (3.7V→5V). Input straddling output (Li-ion 3.0-4.2V→3.3V): buck-boost. Same efficiency/noise trade-off framework.

See also

This article summarises the working rules KEMI uses in real design and fabrication work. For production, part datasheets and fab specifications take precedence.