Every LED design begins with the same two-sign resistor question, and a wrong answer shows up as a dimmed array, a warm resistor or a field failure eighteen months later. This page gives the working formulas, a live calculator for the common supply voltages, and the branch tables our applications engineers use when they check a customer layout. The arithmetic is simple; the discipline is in the details the formulas hide – forward voltage spread, resistor tolerance, and the heat the resistor itself has to shed.

1. The current limiting problem

An LED is a diode, not a lamp. Its forward voltage changes only a few tens of millivolts for a large change in current, which means that connecting it directly across a supply produces a current set by the supply impedance and the diode’s own slope resistance – almost never the value the datasheet specifies. A series resistor converts the supply into a current source by absorbing the difference between the supply voltage and the LED forward voltage.

The governing relation is Ohm’s law applied to the voltage left over after the LED has taken its share:

R = (Vsupply - n × Vf) / If

where n is the number of LEDs in the series string, Vf is the forward voltage of one LED at the intended current, and If is that current in amperes. Three consequences follow directly and each one is a common source of error.

First, the resistor sets current only as well as Vf is known. A 3.2 V nominal LED with a ±0.2 V bin window moves the current by roughly ±20 percent at a 5 V supply, more at lower supply headroom. Where current accuracy matters, either bin the LEDs tightly or drive them from a constant-current source instead.

Second, headroom is everything for efficiency. A 3.2 V LED on a 5 V supply wastes 36 percent of the energy in the resistor. The same LED in a three-LED string on a 12 V supply wastes only 20 percent, because the string forward voltage is closer to the rail.

Third, the resistor must survive its own dissipation. Power in the resistor is P = (Vsupply - n × Vf) × If, and it must stay inside the package rating at the ambient temperature inside the fixture, which is frequently 60 to 80 degrees Celsius rather than the 25 degrees of a test bench.

2. Calculator

Enter the supply voltage, the LED forward voltage at the intended current, the forward current, and the number of LEDs in series. The calculator returns the exact resistor value, the nearest E24 standard value, the dissipation, and the recommended package.

The calculator above solves the two equations that govern a resistor-fed LED branch. The first is Ohm’s law applied to the resistor: the resistance equals the supply voltage minus the total LED forward voltage, divided by the intended current. The second is the power equation applied to the same resistor: dissipation equals the voltage across it multiplied by the current through it. Both are exact for a regulated supply and one resistor per branch, which is the case the calculator covers.

The calculator assumes a regulated DC supply and one resistor per branch. For mains-derived or unregulated supplies, allow for the supply tolerance and the ripple minimum rather than the nominal voltage.

3. Reading the resistor value curves

The first chart plots the required series resistance against forward current for three branch configurations on a 12 V supply. The shape is a rectangular hyperbola, which is why raising the current into the upper half of the chart rapidly reduces the resistor value and increases the sensitivity of the current to Vf spread.

The three curves above are drawn for a fixed 12 V supply and three branch configurations, and the shape of each is worth reading carefully. All three start at the same left-hand point because at very low current the resistor becomes very large, and the required value is dominated by the supply voltage rather than by the LED count. As the current rises, the curves separate because a branch with three LEDs in series leaves less voltage across the resistor than a branch with one, so it needs a smaller resistance for the same current.

The practical consequence is that a long string is more efficient than a short one. A branch of three 3.2 V LEDs on a 12 V supply leaves 2.4 V across the resistor at the operating point, so roughly one fifth of the total power is wasted. A single LED on the same supply leaves 8.8 V across the resistor, and most of the power becomes heat. Where the supply is fixed and the LED count is flexible, adding LEDs in series until the remaining headroom is a reasonable fraction of the total is the single most effective efficiency measure available without changing to a switching driver.

4. Resistor power rating

The second chart shows how quickly the resistor dissipation grows as the supply voltage rises above the LED forward voltage. At the point where the curve crosses the 1/4 W rule the design has two honest options: move to a larger resistor package, or reduce the headroom by putting more LEDs in series.

Resistor power rating is the second half of the design and the half that is most often skipped. The current calculation gives a resistance value; the power calculation gives a package size, and the two must both be satisfied before the design is complete. A 1/4 W resistor that dissipates 0.3 W will drift upwards in value as it heats, which reduces the current, which reduces the light output, which is a slow and confusing failure mode to diagnose.

The two dashed rules in the chart above mark the continuous ratings of a 1/4 W and a 1/2 W resistor at a 70 degrees C ambient. Note the words continuous and ambient: a resistor rated 0.25 W at 70 degrees C is rated less at 100 degrees C, and a resistor carrying a pulsed current may tolerate a higher peak as long as the average dissipation stays inside the rating. Where a design lands close to a rule, the correct responses in order of preference are to raise the resistance and accept a lower current, to split the current across two parallel branches so that each carries half, or to move to the next package size. Running a small resistor hot is the option that looks cheapest on the bill of materials and costs the most in field returns.

A useful margin is a factor of two: choose a package whose rating is at least double the calculated dissipation. The margin covers the ambient being warmer than assumed, the supply being at its high tolerance limit, and the resistor body not being in free air. Where the resistor is potted, enclosed or mounted against a board, the derating is more aggressive and the margin should be larger.

5. Standard resistor values and tolerance

A 5 percent resistor already contributes up to 5 percent of current error before any allowance for Vf spread or temperature. Where an array must look uniform, specify E96 at 1 percent and bin the LEDs to a 0.1 V Vf window; the two measures together hold channel-to-channel current within a few percent.

Standard values matter because the calculated resistance almost never lands on a purchasable part. The E24 series divides each decade into 24 values spaced roughly 10 percent apart, and the E96 series divides it into 96 values spaced roughly 2.4 percent apart. A design that needs 187 ohms will therefore be built with 180 ohms or 200 ohms from the E24 series, and the question is which direction to round.

Round upwards. A larger resistance produces a lower current, which is safe; a smaller resistance produces a higher current, which pushes the LED past its intended operating point and, on a design that was already near the derating limit, past it entirely. The cost of rounding up is a small reduction in light output that is barely visible, and the benefit is a design that stays inside its current specification across the tolerance of every component in the branch.

Tolerance compounds. A five percent resistor and a forward voltage that varies by ten percent across the bin combine to give a current that may be fifteen percent away from the design value in the worst case. Where the current is critical, use a one percent resistor, which removes the resistor contribution almost entirely and leaves only the LED bin spread to manage. The extra cost of a one percent part is small compared with the cost of sorting LEDs or adding an active current regulator.

6. Branch configuration table

The table below is the reference our applications engineers use for a 12 V rail. The Vf column is the total string forward voltage at the stated current.

Note the last row. A four-LED string on a 12 V rail has no headroom at all; the string forward voltage already exceeds the supply, so no resistor value can set the current. This is the single most common mistake in 12 V strip and module designs, and it is why our 12 V modules are specified with three LEDs per three-resistor segment rather than four.

The table above compares the branch configurations that cover most small LED designs. The comparison is not about which one is best in the abstract but about which one fits the supply that is already available, because the supply voltage usually comes first in a project and the LED configuration has to live with it.

Three considerations decide between the rows. The first is headroom: a configuration that leaves too little voltage across the resistor gives poor current regulation because the resistor can no longer absorb the forward voltage spread, and a configuration that leaves too much wastes power as heat. The second is the number of LEDs available: a 24 V supply suits a long string, while a 5 V supply suits a single LED or none at all, because a single white LED at 3.2 V leaves only 1.8 V of headroom and a second LED in series would exceed the supply. The third is fault behaviour: a series string goes dark if any LED fails open, while a set of parallel branches loses only the affected branch, which may matter in a signalling or safety application.

7. Parallel branches and current sharing

Never parallel two LED branches across a single resistor. Because Vf falls with temperature, the branch that starts warmer draws more current, warms further, and takes an increasing share until it is carrying most of the total. The correct topology is one resistor per branch, which is why addressable and multi-segment modules are laid out with a resistor at every cut point.

Placing LEDs in parallel looks like a way to increase the total current without increasing the supply voltage, and it is, but it introduces a problem that a series string does not have. Two LEDs of the same type do not have the same forward voltage, and the one with the lower forward voltage carries more current, runs hotter, and drops further in forward voltage, so it takes even more of the current. The imbalance grows until the distribution stabilises, and the design ends up relying on the hottest LED not to fail.

The standard remedies are to give each parallel branch its own series resistor, which costs one component per branch and limits the imbalance, or to use a constant-current driver with per-string sensing, which removes the imbalance entirely. A single resistor shared by several parallel LEDs is a design that works on the bench with matched samples and fails in production with parts drawn from the full bin range. Where the parallel LEDs are on the same die or in the same package, as in a multi-chip emitter, the manufacturer has already matched them and the problem is controlled.

Thermal runaway is the failure mode to keep in mind. As an LED warms, its forward voltage falls, which at a fixed supply voltage increases the current, which increases the dissipation, which raises the temperature further. With a series resistor the negative feedback is stable because the resistor absorbs the voltage change. With parallel LEDs sharing one resistor the feedback is unstable, and one device progressively takes more than its share.

8. Dimming, PWM and the resistor’s role

Where the intended control is dimming, the series resistor is usually replaced by a constant-current driver with a PWM or analogue input. Two cautions apply when a resistor is retained in a PWM application.

The first is that the average current, not the peak, is what determines brightness, so a resistor sized for continuous 20 mA can be driven at a higher peak in PWM provided the average and the LED thermal budget are respected. The second is that PWM at rates below about 1 kHz becomes visible on moving objects and on camera, which matters for signage and for machine vision illumination; above 2 kHz no visible banding occurs in typical applications. Our addressable J-16 devices accept PWM directly at the protocol rate.

A series resistor is a poor dimmer. Reducing the resistance raises the current, but the relationship between current and light output is not linear and the colour shifts slightly as the current changes, so resistive dimming produces a dimmer that changes hue as it is turned down. It is acceptable for trimming an indicator to a fixed brightness and unacceptable for a lighting product.

Pulse-width modulation is the standard alternative, and it changes the role of the resistor rather than removing it. Under PWM the LED is driven at full current for part of each cycle and off for the rest, so the peak current is the design current and the average light output is set by the duty cycle. The resistor still sets the peak current and still has to be sized for it, but the average dissipation in the resistor is lower because the duty cycle reduces the average. A resistor sized for continuous operation is therefore conservative under PWM, which is a comfortable place to be.

Two rules keep a PWM design out of trouble. First, keep the modulation frequency above the audible range, generally above 1 kHz, and above 3 kHz where a camera or a high-speed sensor is present, because a low frequency produces visible banding on video. Second, do not exceed the peak current rating of the LED just because the average is low; the die sees the peak current and heats accordingly. Where the driver integrates the PWM function, the resistor becomes a current-setting element for a constant-current source rather than a ballast, and the dissipation moves into the driver where it can be managed with a proper thermal path.

9. Worked example with QUEENDOM parts

A designer wants a five-element status indicator running from a 12 V rail using J-17 3528 multi-colour parts at 15 mA per element. Each element is a single LED, so one resistor per element is required.

The 0.125 W figure sits inside a 1/4 W part with good margin at ambient, but a 1/2 W part is the safer specification inside a sealed enclosure where the local temperature may reach 70 degrees Celsius, because the continuous rating of a metal film resistor derates above 70 degrees.

The example above uses three of the multi-colour SMD parts in the J-17 and J-18 groups, which have published forward voltage bins and a nominal 20 mA per channel condition. Take a 12 V regulated supply, three LEDs in series at 3.2 V each, and an intended 20 mA. The headroom is 12 minus 9.6, or 2.4 V. Ohm’s law gives 2.4 divided by 0.020, or 120 ohms, which happens to be an E24 value, so no rounding is needed. The dissipation is 2.4 V multiplied by 20 mA, or 48 mW, which sits comfortably inside a 1/4 W resistor with a factor of five margin.

Now change one assumption. Suppose the same design uses a 24 V supply, which is common in industrial equipment. The headroom becomes 24 minus 9.6, or 14.4 V, and the required resistance becomes 720 ohms. The dissipation becomes 14.4 V multiplied by 20 mA, or 288 mW, which exceeds a 1/4 W resistor and requires a 1/2 W package. Worse, 288 mW of the 480 mW total is being turned into heat in the resistor rather than into light, so the efficiency of the branch is below 40 percent. The correct response is to add more LEDs in series, up to six at 3.2 V, which brings the headroom down to 4.8 V and the dissipation to 96 mW, and restores the efficiency to about 80 percent.

Where the forward voltage bin matters, size for the high end of the bin. A branch designed against the typical forward voltage will over-current on the low-voltage parts once the supply is at its high tolerance limit. Designing against the highest forward voltage in the bin and the highest supply voltage gives a current that is never exceeded, at the cost of a slightly lower current on the average part.

10. Where a resistor is the wrong answer

• High current arrays. Above roughly 100 mA total the resistor waste becomes a thermal problem in the fixture; move to a constant-current driver.
• Wide supply tolerance. An unregulated supply that varies from 10 to 14 V gives a 40 percent current swing with a fixed resistor; only a driver holds the current.
• Critical colour stability. Because peak wavelength shifts with current, an RGB element driven through a resistor from a varying rail will drift in colour. Use a driver per channel.
• Battery products. The resistor burns the difference between a full and empty cell; a switching driver delivers the same light for a longer runtime.

Four situations call for something other than a series resistor, and recognising them early saves a redesign later.

The first is high current. Above roughly 100 mA the dissipation in the resistor becomes large enough that the resistor is the dominant thermal load in the fixture, and a switching constant-current driver both saves that waste and regulates the current properly.

The second is a widely varying supply. A battery whose voltage falls from 12.6 V to 10 V over its discharge, or an unregulated mains-derived rail, changes the headroom across the resistor by more than the LED can tolerate. A constant-current driver holds the current regardless of the input within its compliance range.

The third is a requirement for accurate current matching. Where several LEDs must match to within a few percent, as in an RGB colour-mixing application, the forward voltage spread across bins makes resistor-based matching unreliable, and an active per-channel regulator is the practical answer.

The fourth is a requirement for dimming below the point where resistive control is stable. Where a product must dim to one percent, and hold its colour while doing so, pulse-width modulation or an analogue constant-current control is the only approach that meets the specification.

11. Contact and design review

QUEENDOM supplies the SMD and high-power families referenced above with published forward voltage bins and recommended drive conditions, and our applications engineers will check a resistor layout, a branch count or a thermal budget against the fixture ambient on request. Send the rail voltage, the branch topology, the intended current and the enclosure ambient, and we will return a checked schematic and a bill of materials.

Related products and applications

• Multi-colour SMD LED (J-17, J-18, J-19)
• SMD LED packages (J-01)
• Addressable RGB LED (J-16)
• Application overview: LED components application solutions
• More design resources: LED design resources

Reference Data for Resistor Design

The table and curve below extend the calculator with the two lookups most often needed at the bench: the forward voltage a given emitter color presents at 20 mA, and the current error a Vf bin introduces. Values are typical of volume-production SMD parts; always confirm against the specific datasheet bin.

Forward voltage by emitter color:

Series counts by rail assume a minimum 1.0 V of headroom so the resistor can regulate, which is the practical floor with binned parts. Strings above these limits sit below the knee and current falls out of specification with temperature.

Parallel branches multiply array current, but every branch must carry its own resistor. A shared resistor lets the branch with the lowest Vf steal current as temperatures diverge, which is the classic mechanism behind strip modules that darken from one end.