AMS1117-3.3 PCB Design Guide: Footprint, Pinout, and Alternatives
1 A LDO with about 1.1 V dropout; ubiquitous 5 V to 3.3 V regulator
The AMS1117-3.3 is the three-legged regulator on nearly every cheap dev board that turns 5 V into 3.3 V. It is a fixed-output linear regulator in SOT-223 rated for 1 A, it costs a few cents, and in its home application — a solid 5 V input feeding a 3.3 V rail — it just works. The plain AMS1117-3.3 ordering code is the SOT-223 part; TO-252 and SOIC variants carry different suffixes.
The word LDO in the datasheet title is the trap. Dropout is 1.1 V typical and 1.3 V max at 0.8 A — and AMS specifies dropout only up to 0.8 A, noting it climbs further above that. By modern standards this is not a low-dropout part, and designs that treat it like one (most famously, regulating a 3.7 V Li-ion cell down to 3.3 V) fall out of regulation.
The recurring AMS1117 failures are all predictable from the datasheet: too little input headroom, the wrong output capacitor, quiescent current that quietly drains batteries, back-fed current at power-down, and thermal math nobody ran. Each is covered below.
What breaks boards
Dropout is over a volt — Li-ion to 3.3 V does not work
Dropout is 1.1 V typ / 1.3 V max at 0.8 A, and the datasheet notes dropout rises above 0.8 A (it is not specified at 1 A at all). 5 V in, 3.3 V out leaves comfortable margin. A 3.7 V Li-ion cell does not: the moment the cell sags below roughly 4.4–4.6 V the output falls out of regulation, which is most of the discharge curve. For battery rails use a true low-dropout part such as an HT7333, RT9013, or AP2112K instead.
The output capacitor is part of the compensation — use 22 µF tantalum-class
The Stability section of the datasheet is explicit: the output capacitor is required as part of the device frequency compensation, and 22 µF solid tantalum "will ensure stability for all operating conditions." Ceramics are never qualified and no minimum-ESR figure is published, so a low-ESR ceramic-only output is outside the characterized envelope — the widely reported ceramic-output oscillation is consistent with that, even though the datasheet never states it verbatim. Fit a tantalum or a ceramic with a deliberate series resistance.
5 mA quiescent current kills battery budgets
Quiescent current is 5 mA typical and 11 mA max. That is irrelevant on a mains-fed board and fatal in anything that sleeps: 5 mA is roughly 43 Ah per year, more than most small Li-ion or coin-cell budgets outright. If your design has a low-power sleep state, the AMS1117 is the wrong regulator regardless of dropout.
No reverse-current protection — add a diode with large output capacitance
If the input rail collapses while the output capacitance is still charged (big bulk caps downstream, or a backup supply on the 3.3 V rail), current flows backwards through the regulator. The standard fix is a diode from output to input so the output cap discharges through the diode instead of the die. With only tens of µF on the output it rarely matters; with millifarads or a second supply it does.
Run the thermal math: SOT-223 is a 1–2 W package, not a 4 W one
Dissipation is PD = (VIN − VOUT) × IOUT. At 5 V in, 3.3 V out, 0.8 A that is 1.36 W — workable on a decent copper pour (θJA spans roughly 46 °C/W with generous copper to over 90 °C/W with minimal copper). But 12 V in at 3.3 V / 0.5 A is 4.35 W, which no SOT-223 layout survives: the part hits thermal shutdown or cooks. For high step-down ratios put a buck converter (e.g. MP1584EN) in front and let the AMS1117 do the last volt or two.
Key specifications
| Parameter | Value | Source |
|---|---|---|
| Dropout at 0.8 A | 1.1 V typ / 1.3 V max at IOUT = 0.8 A (not specified at 1 A; Note 4: dropout is higher above 0.8 A) | ds1117.pdf Electrical Characteristics, 'Dropout Voltage (VIN - VOUT)' row + Note 4 |
| Iq | 5 mA typ / 11 mA max at (VIN − VOUT) = 1.5 V | ds1117.pdf Electrical Characteristics, 'Quiescent Current' row |
| Max input | 15 V (absolute maximum input voltage) | ds1117.pdf Absolute Maximum Ratings |
| Output accuracy | ±1.5% at 25 °C (3.251–3.349 V, VIN = 4.8 V); ±3% over full operating range (3.201–3.399 V) | ds1117.pdf Electrical Characteristics, 'Output Voltage AMS1117-3.3' rows |
| Required output cap / ESR | 22 µF solid tantalum (output capacitor is part of the frequency compensation; 'will ensure stability for all operating conditions'); no numeric ESR limit published | ds1117.pdf Application Hints, 'Stability' section; ripple-rejection test conditions also use COUT = 22 µF tantalum |
| θJA (SOT-223) | 90 °C/W (46 °C/W to >90 °C/W depending on copper area/mounting); θJC 15 °C/W | ds1117.pdf Absolute Maximum Ratings, 'Thermal Resistance' + footnote |
Verified against the manufacturer datasheet on 2026-07-09. Confirm the current revision before production use.
Alternatives
- AP2112K-3.3 — a true LDO: ceramic-stable, 600 mA, enable pin, low quiescent current — the right part for Li-ion rails, but input tops out at 6 V.
- MCP1700-3302E — 1.6 µA quiescent current for battery designs that sleep; only 250 mA and 6 V max input.
- LD1117S33 — ST's spec-controlled major-vendor equivalent in the same 1117 lineage (which originated at Linear Technology as the LT1117), same dropout class, with a traceable supply chain.
Common questions
- What is the AMS1117 dropout voltage?
- 1.1 V typical, 1.3 V max at 0.8 A output. AMS does not specify dropout at the full 1 A rating and notes it is higher above 0.8 A, so plan on roughly 4.6 V minimum input for a solid 3.3 V rail at load.
- Can I power an AMS1117-3.3 from a 3.7 V Li-ion battery?
- No. With over a volt of dropout the input must stay above roughly 4.4–4.6 V under load, and a Li-ion cell spends most of its discharge below that. Use a genuinely low-dropout regulator such as an HT7333, RT9013, or AP2112K-3.3.
- What output capacitor does the AMS1117 need?
- The datasheet calls for 22 µF solid tantalum and states the output capacitor is part of the frequency compensation. Ceramic-only outputs are not qualified and no ESR limit is published, so stick with tantalum-class capacitance or add deliberate series resistance to a ceramic.
- AMS1117 vs LD1117 — what is the difference?
- They are the same 1117 architecture (a lineage that started as the Linear Technology LT1117). The LD1117 is ST's spec-controlled equivalent with tighter production control and a traceable supply chain; electrically they are the same dropout and capacitor class, so layout and caveats carry over.