LED Current Sense Resistor Tolerance: How It Moves Regulated Current

This guide builds a working budget for LED current sense resistor tolerance. It is deliberately not about the series resistor that limits current from a fixed supply, which is a different circuit with a different calculation; the distinction is the first thing worth fixing before the numbers mean anything.

Quick Answer: A Small Resistor Sets a Wide Window

A driver regulates by holding a fixed voltage across the sense resistor, so the current it delivers is that reference divided by the resistance. Every error in the resistance becomes an error in current, in the opposite direction and with the same magnitude, before any other term is added.

Two percent of resistance error is two percent of current error, and that is only the beginning. Reference tolerance, comparator offset, and the temperature coefficient of the resistor itself add to the same budget. A design that treats the resistor as exact has, in practice, already spent its accuracy allowance before the first real component is soldered.

What the Sense Resistor Is Doing That a Series Resistor Is Not

A series resistor limits current by dropping the difference between the supply and the LED forward voltage. Its value is chosen once, from a nominal supply, and the resulting current varies with everything that moves the forward voltage. The circuit is open loop: nothing measures what actually flows.

An LED sense resistor sits inside a closed loop. The driver compares the voltage across it with an internal reference and adjusts the pass element until the two match. That structure is what makes constant current accuracy possible at all, and it is also why the sense resistor's own error cannot be corrected by the loop: the loop is faithfully regulating to a value that is slightly wrong.

The classic limiting-resistor calculation is covered separately in LED current limiting resistor calculation, which is the right reference for open-loop designs. Everything below assumes a regulated loop.

The Sense Equation and Where Error Enters

In its simplest form the regulated current is the reference voltage divided by the sense resistance. Three error sources act directly on that ratio: the initial tolerance of the resistor, the initial tolerance of the reference, and the offset of the amplifier that compares them.

Note the last row. Self-heating is not independent of the others: it converts electrical error into thermal error, and the thermal error feeds back through the temperature coefficient. A resistor that is comfortably inside its power rating can still be the largest single term in the budget if its package has a high thermal resistance to the board.

The offset term deserves a warning of its own. Fixed offset voltage becomes a larger fraction of the measurement as the sense voltage falls, so designs that lower the sense voltage to save power pay for it in accuracy. There is a floor below which the amplifier, not the resistor, dominates.

Building the Tolerance Stack

A tolerance stack is not a sum of percentages unless every term is a worst case in the same direction. The practical approach is to compute a worst-case window and a statistical estimate, then design to the worst case and check that the statistical figure is not so much tighter that the design is being over-constrained.

Worked example with round numbers. A 0.1 ohm sense resistor with one percent initial tolerance, a 100 millivolt reference with one percent tolerance, and a current-sense amplifier with one millivolt of offset gives a current near one amp. The resistor contributes plus or minus one percent, the reference contributes plus or minus one percent, and the offset contributes plus or minus one percent of the sense voltage, because one millivolt against one hundred millivolts is one percent.

Worst case in one direction, those three terms add to about three percent, so the loop may deliver roughly 0.97 to 1.03 amps before temperature is considered. Statistical combination would put the typical spread closer to 1.7 percent, which is the number a production distribution will resemble, but the worst case is the number that has to be safe. Seeing how resistor tolerance affects LED driver current in a worked example is more useful than arguing about percentages in the abstract.

Add a temperature coefficient of 100 parts per million per degree Celsius and a 40 degree rise, and the resistor term grows by another 0.4 percent, asymmetrically, because resistance rises with temperature and current falls. If the sense resistor is a low-value metal-element type with a much lower coefficient, that term shrinks and the reference and offset terms become the ones worth improving.

For the LED side of the budget, it helps to remember what the current error actually does to light output. Forward current and optical output are related but not proportional, and the relationship near the rated point is described in LED forward current vs rated current. A three percent current window usually produces a smaller optical window, which is why driver accuracy requirements can sometimes be relaxed and sometimes cannot.

Temperature Coefficient and Self-Heating

The temperature coefficient of resistance is quoted per degree Celsius, and its sign depends on the material. Metal film and metal element parts are low and positive; some thick-film and carbon types are higher and can be positive or negative. For a current-sense application the absolute magnitude matters more than the sign, because either direction consumes margin.

Self-heating is the part most often omitted. The resistor dissipates the square of the current times its resistance, and that heat has to leave through the package, the solder joints and the copper underneath. A small package with narrow thermal paths can run far above board temperature at modest dissipation, and the resulting drift is a systematic offset rather than random noise, so it does not average out across units.

• Calculate dissipation at the maximum regulated current, not the nominal one.
• Estimate the resistor's temperature rise from its package thermal resistance, then apply the coefficient to that rise.
• Prefer a larger package or a lower resistance with more gain when dissipation is the dominant term.
• Keep the sense resistor away from other heat sources, and do not place it in the airflow shadow of a driver IC.
• Check the coefficient at the actual operating temperature, not at room temperature, because it is not constant across the range.

Heat is not only a resistor problem. Once the loop is accurate, the remaining temperature question is whether the LED junction itself stays inside its limits, which is a separate measurement described in LED thermal resistance RthJA vs RthJC.

Converting the Stack Into a Worst-Case Current Range

The output of the exercise should be a range written on the specification, not a single target. That range becomes the acceptance criterion for incoming inspection and the input to the LED's own drive conditions.

Three checks make the range usable. First, the upper end must stay inside the LED's rated continuous current at the maximum ambient the product allows. Second, the lower end must still deliver the optical output the application needs at end of life, since output degrades over time. Third, the range must be consistent with whatever the display's brightness specification promises, because a brightness tolerance narrower than the current tolerance is a promise the driver cannot keep.

Where the range comes out too wide, the levers are ordered by cost. Tightening the LED driver resistor tolerance is usually the cheapest effective lever. Specifying a lower temperature coefficient is cheap in low volumes and increasingly worthwhile in high ones. Reducing amplifier offset means a different device, which may also change the sense voltage and the power loss. Changing the reference is the most expensive because it touches the driver architecture.

One architecture-level caution: the loop can only regulate while it has headroom. If the supply falls below what the LED string needs plus the sense voltage plus the dropout of the pass element, the driver leaves regulation and current falls with the supply. That failure mode looks like tolerance drift but is not, and it is covered in LED driver compliance voltage.

Bench Verification

Verification confirms the budget rather than replacing it. The measurement that matters is current at the LED, at two temperatures, at two supply voltages, across at least a handful of units so that the spread is visible.

1. Measure the actual sense resistance on a sample of boards, including the trace resistance in series with it if the layout uses a long return path.
2. Measure regulated current with a meter in series, at nominal supply and at the minimum supply the product allows.
3. Repeat after thermal soak at the maximum operating ambient, allowing the board to stabilise before recording.
4. Compare the measured window against the calculated one; a wider measured window means a term was omitted from the budget.
5. Keep the sense voltage waveform on the scope as well, because ripple and switching noise can shift the average the loop regulates to.

If the measured spread is much narrower than the worst-case calculation, the design is conservative and can be relaxed. If it is wider, the missing term is usually layout resistance, thermal coupling or amplifier offset at low sense voltage, in that order of likelihood.

Frequently Asked Questions

How does resistor tolerance affect LED driver current?

Directly and inversely: a resistor one percent low produces a current about one percent high, because the loop regulates a fixed voltage across it. That error then adds to reference, offset and thermal terms rather than replacing them.

Is a one percent sense resistor accurate enough for constant current driver tolerance calculation?

It depends on the other terms. If reference and offset are also about one percent, the combined window is roughly three percent before temperature, which is often acceptable but rarely tight.

Why does current drift as the product warms up?

Because the sense resistor's temperature coefficient changes its value, and self-heating raises its temperature above board temperature. Both effects are systematic and appear on every unit in the same direction.

Can I use a lower sense voltage to reduce power loss?

Yes, but the amplifier offset becomes a larger fraction of the signal. Below a certain sense voltage the driver's offset, not the resistor, sets the accuracy.

Does LED current sense resistor accuracy change with drive current?

The percentage terms from tolerance and reference stay roughly constant, while offset and self-heating scale differently. This is why a design verified at one current setting is not automatically verified at another.

Should the sense resistor be a tight-tolerance part or a trimmed one?

Tight tolerance is usually enough, because the loop corrects the absolute value only through the reference. Trimming helps when the reference is fixed and the total window must be very narrow.

Conclusion: Budget the Window, Then Publish It

The sense resistor is small, inexpensive and decisive. Its tolerance, coefficient and self-heating set the width of the current window that the driver can actually hold, and that window propagates into brightness tolerance, thermal margin and the credibility of the datasheet.

Build the stack, convert it into a range, verify the range on real boards at real temperatures, and write the result into the specification. If you are choosing a driver and display combination and want the current window checked against the LED's rated conditions, send the drive requirements to the HOUKEM team; the reason a regulated loop is used at all is explained in why LEDs require constant current control.