How it works
An LDO (low-dropout) regulator passes current from input to output through a PMOS or PNP pass element, burning off the difference as heat. The output voltage is regulated by a feedback loop that compares the output (via a resistor divider) to an internal reference.
Pdiss = (Vin − Vout) × Iout + Vin × Iq
Tj = Ta + Pdiss × θJATwo things kill LDO designs: insufficient dropout headroom and underestimated thermal dissipation.
Dropout voltage
The dropout voltage Vdo is the minimum Vin − Vout at which the LDO still regulates. Below this, the pass element saturates and the output follows the input minus a fixed offset — regulation is lost.
Vdo scales with load current. Most datasheets give it at the rated maximum. At lighter loads it drops significantly. For a 3.3 V design with AMS1117 (Vdo = 1.2 V), your input must be ≥ 4.5 V. That rules out a 3.7 V LiPo cell in most cases.
For battery-powered designs where Vin can drop to 3.0–3.6 V, choose a true-LDO with Vdo < 200 mV — MCP1700, XC6206, or TLV700xx family.
Thermal calculation
The LDO dissipates Pdiss as heat. The package thermal resistance θJA converts that to a junction temperature rise above ambient:
Tj = Ta + Pdiss × θJAKeep Tj < Tj_max (typically 125 °C). In practice, derate to 100 °C to avoid premature aging.
A SOT-23 package has θJA ≈ 250–335 °C/W, which limits dissipation to around 200–360 mW before Tj becomes problematic at 25 °C ambient. High-dissipation designs need SOT-223 (θJA ≈ 100–160 °C/W) or TO-252, ideally with a copper pour under the tab.
Example: 5 V in, 3.3 V out, 500 mA load, SOT-23, Ta = 25 °C:
Pdiss = (5 − 3.3) × 0.5 = 0.85 W
Tj = 25 + 0.85 × 300 = 280 °C ← thermal shutdownThat SOT-23 will protect itself but won’t regulate continuously. Use a buck converter instead, or a SOT-223 with copper pour.
Feedback resistors for adjustable LDOs
Adjustable LDOs (AMS1117-ADJ, NCP5662, TLV1117-ADJ) set their output with an external resistor divider from output to feedback pin to ground:
Vout = Vref × (1 + R1 / R2)
R1 = R2 × (Vout / Vref − 1)Vref is the internal reference (1.25 V for AMS1117, 0.8 V for NCP5662). A typical design uses R2 = 10 kΩ–100 kΩ; higher values reduce quiescent current drain but increase susceptibility to noise. With R2 = 10 kΩ and Vref = 1.25 V, the divider draws only 125 µA — negligible for most designs.
After calculating R1, round to the nearest E24 standard value. The calculator shows the resulting Vout error so you can decide whether to trim or accept it. For ±1% accuracy, use 1% resistors; E96 values aren’t usually necessary.
Common mistakes
Forgetting the minimum Vin over the full battery discharge curve. A fresh 2S LiPo delivers 8.4 V; a discharged one gives 6.0 V. If your LDO plus load needs 6.2 V minimum, the circuit will drop out before the battery is dead. Design for the end-of-discharge voltage, not the nominal.
Using AMS1117 on battery. The AMS1117’s 5 mA quiescent current drains a 250 mAh coin cell in 50 hours with zero load. Use XC6206 (1 µA Iq) or MCP1700 (1.6 µA Iq) for anything running from a battery.
No output capacitor, or wrong type. Many LDOs require a minimum output capacitance for stability — typically 1–10 µF. Some older designs (AMS1117) specifically need capacitors with an ESR in the 0.1–1 Ω range; an MLCC alone can cause oscillation. Check the datasheet’s stability section. Newer LDOs (MCP1700, TLV700xx) work fine with ceramic caps.
Ignoring the Iq path. Quiescent current flows from Vin to ground (or output) through the LDO’s bias circuitry, regardless of load. For high-Iq parts like AMS1117 (5 mA), this dominates the power budget at light loads. The calculator includes Iq in the dissipation calculation; verify the total Vin × Iq loss is acceptable.
Thermal runaway in high-temperature environments. Ta = 85 °C is common in automotive and outdoor designs. A dissipation that’s safe at 25 °C may exceed Tj_max at 85 °C. Always recalculate for the worst-case ambient.
Thermal resistance converter
The calculator’s second tab converts between thermal conductivity and thermal resistance, and computes series/parallel thermal networks.
Conductivity to resistance
Thermal resistance quantifies how much temperature rise a material produces per watt of heat flowing through it. For a flat slab:
Rth = L / (k × A)Where L is thickness (m), k is thermal conductivity (W/m·K), and A is cross-section area (m²). Result is in °C/W — identical to K/W, since the degree size is the same for both scales.
Common values matter for PCB stackups: FR4 sits at 0.3 W/m·K, making it a poor thermal conductor. A 1.6 mm FR4 board under a 10×10 mm pad gives Rth = 1.6 × 1000 / (0.3 × 100) = 53 °C/W. A 2×2 mm copper pour (401 W/m·K, same thickness) over the same area drops that path to under 1.1 °C/W — copper is roughly 1300x better than FR4 for the same geometry. Thermal paste between a package tab and a heatsink typically has k = 3–8 W/m·K; a 0.1 mm layer over 25 mm² gives Rth ≈ 0.5–0.8 °C/W depending on grade.
The thermal conductivity to thermal resistance converter accepts thickness in mm and area in mm² directly — no manual unit conversion needed.
Series and parallel networks
The full thermal path from die junction to ambient is a series chain:
Rth_total = Rθ_JC + Rθ_CS + Rθ_SA
junction-to-case + case-to-board + board-to-ambientA TO-252 LDO on a small heatsink might be: 10 °C/W (θJC, junction-to-case) + 0.8 °C/W (TIM interface, 0.1 mm over 25 mm²) + 8 °C/W (heatsink, natural convection) = 18.8 °C/W total. At 1 W dissipation that’s only an 18.8 °C rise. Compare to a SOT-223 on minimal copper (θJA ≈ 160 °C/W): at 1 W the junction rises 160 °C — already over the limit at 25 °C ambient.
Parallel paths combine as 1/Rth_total = 1/R1 + 1/R2. Heat spreading through a copper pour alongside the package body runs in parallel with the package’s own Rth_CS — reducing total resistance and improving cooling without adding a heatsink.
The thermal resistance unit converter tab handles both calculations — enter the individual Rth values in °C/W, choose Series or Parallel, and read the combined result along with ΔT projections at common dissipation levels.