SizingKitElectronics · driving an LED
LED resistor calculator
Give it a supply rail, a forward voltage and the current you want and it returns the exact resistance, the nearest E24 value you can buy, the current each of those actually produces, and the smallest standard wattage that survives the heat. It also prints how far the current moves for every 0.1 V of forward-voltage spread and over a stated junction temperature rise, which is the pair of numbers that decides whether the design works with the next reel of LEDs. Free, no account, and no part number is asked for at any point.
- 100% free
- No signup
- E24 stock values
- Wattage with derating
- Series-string comparison
The string and what feeds it
LEDs in series share one current, so the count multiplies the forward voltage and eats the headroom. LEDs in parallel do not share anything reliably, which is what the second panel is about.
The regulated rail, not the transformer's rating.
Default 2 V — the LED datasheet's own forward-voltage window, quoted at its test current — no standard value exists.
What you want through the string, not the absolute maximum.
This supply will carry at most 1 at 3.2 V each, and fewer with usable headroom.
Default 2× — resistor power ratings are stated at the ambient named on the datasheet — commonly 70 °C in free air — and derate linearly to zero at the maximum body temperature. Inside a sealed enclosure, under a lens, or on a board dense enough to raise local ambient, the derating curve is what decides. Ask for more margin there, not less.
Default 2 mV/°C — the forward-voltage temperature coefficient stated on LED datasheets as ΔVf/ΔTj. Published figures span roughly −1 to −4 mV/°C and depend on the die chemistry and the drive current. Take the slope from the part's own Vf-versus-temperature curve.
How far above ambient the die settles once it has been on for a while.
RESISTOR, EXACT
90.0 Ω
NEAREST STOCK VALUE
91 Ω
| Headroom the resistor works with | 1.800 V of 5 V (36.0%) |
|---|---|
| Current at the nearest stock value | 19.8 mA (-1.1% off target) |
| Current at the next value up | 91 Ω → 19.8 mA |
| Burned in the resistor / in the LEDs | 36 mW / 64 mW |
| Fraction of the supply power reaching the LEDs | 64.0% |
| Smallest rating to buy at your margin | 0.125 W |
| Current change per 0.1 V of Vf spread | 5.6% |
| Current change over a 40 °C rise | +4.4% |
Stock values from IEC 60063, series E24 — the values a ±5% resistor is made in; wattage from the standard axial-lead film resistor power ratings, each stated at its own datasheet's ambient. At 3 W and above the part is usually wirewound and cased differently, with its own derating curve and a much hotter surface. Above 5 W this list stops; read the manufacturer's series data instead of extrapolating it. The nearest E24 value is the honest arithmetic answer, but the next value up is the safe one — it lowers the current, and every LED parameter on a datasheet is quoted at or below a maximum.
Sanity-check the forward voltage against the color
An LED’s emission wavelength and its forward voltage come from the same band gap, so the color of the light puts a scale on Vf. If the two are far apart, one of the two numbers is from the wrong datasheet.
465 nm is a typical blue die, 630 nm a red one, 940 nm an infrared remote emitter.
PHOTON ENERGY AT THAT WAVELENGTH
2.666 eV
YOUR FORWARD VOLTAGE
3.20 V
The junction is running 0.534 V above the photon energy. A few tenths above is ordinary — that difference is the series resistance of the die and its contacts, and it is larger in the GaN parts that make blue, green and white than in the AlInGaP parts that make red and amber. A whole volt apart in either direction means the Vf and the wavelength did not come from the same part.
Photon energy is 1239.841984 eV·nm divided by the wavelength — hc/e, exact from the SI definitions of the Planck constant, the speed of light and the elementary charge (2019 redefinition). This is a plausibility check, not a measurement: the emission peak spans tens of nanometers and a phosphor white LED emits far from its die’s own wavelength, so treat a close match as reassurance and a wild mismatch as a question.
Many LEDs: how to split them
Every row below is one resistor per string. There is no row for several LEDs in parallel behind a single resistor, because that arrangement has no defined current per LED — see the note under the table.
The whole count in the fixture or on the board.
| Per string | Strings | Resistor each | Wasted in resistors | Per 0.1 V of Vf |
|---|---|---|---|---|
| 1 | 6 | 90.0 Ω | 216 mW | 5.6% |
Longer strings waste less and hold the current less securely — the last column is the same headroom argument read as a risk. Nothing here divides current between parallel LEDs on one resistor, and that is deliberate: parallel LEDs all sit at one node voltage, and the only thing deciding how the resistor’s current splits between them is the difference between their individual Vf curves, which the datasheet gives as a window rather than a number. The lowest-Vf part in the group takes the largest share, runs hottest, drops further as it heats, and takes more still. One resistor per string, or a constant-current driver, are the two arrangements with a defined answer.
How to pick a resistor that keeps working when the LEDs change
The division is trivial. Everything that goes wrong afterwards is about how little voltage the resistor was left to work with.
Take the forward voltage from the part, not from the color
Vf is stated in the datasheet as a window at a named test current — a white LED specified 3.0 to 3.4 V at 20 mA really is any of those values, and the one you get is not the typical figure. Type the typical value to design with, then re-run with the top and bottom of the window to see the current you would get from the worst part in the reel.
Read the headroom before you read the resistance
The panel prints the supply minus the string as its own figure. Under about 20% of the supply the current stops being something you set and becomes something you hope for: at 100 mV of headroom, a part 50 mV off the typical Vf is 50% off the target current. If the headroom is thin, drop one LED from the string or move up a supply rail rather than trimming the resistor.
Choose the next stock value up, then size the wattage
Both E24 neighbors are shown with the current each produces. Upward means more resistance and less current, which is the direction every LED maximum rating is written in. Then take the printed wattage, which is the dissipation multiplied by your margin and rounded to a rating you can buy — not the dissipation itself.
Technical specifications
| Stock values offered | E24, both neighbors of the exact figure, with the resulting current for each — the nearest and the next one up are usually different parts |
|---|---|
| Wattage list | 0.125, 0.25, 0.5, 1, 2, 3 and 5 W axial ratings; above 5 W the list stops rather than extrapolating |
| Default margin on the rating | 2×, because a rating is a point on a derating curve at the ambient the datasheet names, commonly 70 °C in free air — editable |
| Forward voltage | Editable, prefilled at 2.0 V, with no claim of authority: LED datasheets state a window at a test current and this page says so rather than shipping a color table |
| Junction coefficient | Editable, prefilled at 2 mV/°C of fall; published LED figures span roughly 1 to 4 mV/°C |
| Sensitivity figure | Current change per 0.1 V of Vf spread = 0.1 × LEDs in series ÷ headroom. At 1.8 V of headroom that is 5.6%; at 100 mV it is 100% |
| Wavelength cross-check | Photon energy from hc/e = 1239.841984 eV·nm, exact under the 2019 SI redefinition — a plausibility check on the Vf you typed, not a measurement |
| Privacy | Datasheet figures you enter stay in the tab; the page makes no request after it loads |
Frequently asked questions
What resistor do I need for a 3 V LED on 5 V at 20 mA?
100 Ω exactly, and 100 Ω is an E24 value so there is nothing to round. The resistor drops the 2 V of headroom and dissipates 40 mW, so a 1/8 W part clears a 2× margin. Change the LED to a 3.4 V part from the same reel and the same resistor gives 16 mA — a 20% shift from a 0.4 V difference, which is the whole argument for keeping headroom generous.
Can I put several LEDs in parallel behind one resistor?
Not with a defined current in each. Parallel LEDs all sit at the same node voltage, so the split is decided entirely by the differences between their forward-voltage curves — and the datasheet gives that as a window, not a number. The lowest-Vf part takes the largest share, runs hottest, drops further as it warms and takes more still. One resistor per string, or a constant-current driver, are the two arrangements with an answer.
How many LEDs can I run in series from my supply?
Fewer than the supply divided by the forward voltage, and the gap between those two numbers is the design. Five 3.2 V LEDs on 24 V leaves 8 V of headroom and a comfortable circuit; seven leaves 1.6 V and works; eight is 25.6 V and does not light at all. The arrangement table on this page lists every workable split with the resistor loss and the sensitivity for each.
Does the resistor go before or after the LED?
Either — it is a series circuit and the same current flows through both regardless of the order. What does change with position is which end of the resistor sits near the supply rail, which matters for a switched low side, for a shared anode display, and for probing: a resistor on the cathode side puts one end at ground and makes the voltage across it easy to measure, which is the fastest way to check the actual current.
Why does my LED get brighter and hotter after a few minutes?
Because the forward voltage falls as the junction warms — around 2 mV per degree — and every millivolt it gives up lands on the resistor as extra current. Over a 40 °C rise, one LED with 1.8 V of headroom gains about 4.4%, which nobody notices. The same LED on a 3.3 V rail with 100 mV of headroom gains 80%, which is how a marginal design ends up dimming permanently after a week.
Is a resistor good enough or do I need a constant-current driver?
A resistor is fine when the headroom is a healthy fraction of the supply and the wasted power is acceptable — a 20 mA indicator on 5 V wastes 40 mW and nobody cares. It stops being fine when the string voltage approaches the supply, when the supply itself moves (a battery from 4.2 V down to 3.2 V), or when the wasted power matters, which is most lighting rather than most indication. The wasted-power column in the arrangement table is where that decision usually gets made.
The resistor I calculated is not a value I can buy. What now?
Take the next E24 value up and accept slightly less current — the tool prints exactly how much less. Series or parallel combinations to hit the exact figure are almost never worth it here: the LED's own Vf tolerance moves the current further than the resistor rounding does, so a 5% resistor error sits well inside the noise you already have.
About headroom, forward-voltage spread and what a series resistor really controls
A diode is a poor thing to put a voltage across. Its current rises exponentially with forward voltage, so a tenth of a volt is the difference between a dim LED and a dead one, and no supply is specified tightly enough to sit safely on that curve. The series resistor exists to break the exponential: it converts a voltage source into something close to a current source, and the strength of that conversion is set entirely by how much voltage the resistor gets to hold. That difference — the supply minus the string — is the number this page treats as the answer, because every failure mode of the circuit is a statement about it.
The arithmetic is one line and the consequences are not. Current sensitivity to forward-voltage spread is the headroom in the denominator: at 1.8 V of headroom a part 0.1 V off typical is 5.6% off the target current, and at 0.1 V of headroom it is 100% off. The same denominator governs thermal behavior, because the junction drops roughly 2 mV less per degree as it warms and the resistor absorbs the difference as current. It also governs efficiency in the opposite direction: a long string wastes little and holds its current loosely, a single LED on a high rail wastes most of the power and holds the current precisely. There is no arrangement that wins on both, which is why the tool lists them all with both columns rather than nominating one.
Two things that get quietly assumed elsewhere are refused here. There is no forward voltage table by color, because a forward voltage is a property of a part number and its drive current, not of a hue — what the page offers instead is a wavelength cross-check against the photon energy of the emitted light, which is exact and catches the case where the Vf came from a different datasheet. And there is no arrangement in which several LEDs share one resistor in parallel, because there is no honest number to print for it. If you are choosing the resistor from a drawer, the resistor color code calculator turns the value into bands and back; if you are feeding something that reads a voltage rather than something that lights up, that is the voltage divider calculator, which is a different circuit with a different failure. Where the LED is being driven from a battery, the battery runtime calculator takes the supply current this page prints and turns it into hours, and the battery capacity calculator works back from a wanted runtime to the bank.
Where these datasheet figures go
Every number on this page is worked out by JavaScript running in the tab you are reading it in. Nothing you type — loads, lengths, nameplate ratings, the rates your utility charges you — is uploaded, logged or kept, which is also why the calculators carry on working in a mechanical room with no signal.
Forward voltages and drive currents you copy out of a datasheet are held in the page only. Nothing here identifies a part number, and nothing is stored between visits.