The derating curve is the most operationally important plot on an LED datasheet, and the most casually read. It draws the boundary between drive conditions that reach useful lifetime and conditions that cook the junction. Parts operated inside the curve meet the rated numbers; parts outside it meet nothing except heat.
What the Plot Shows
The horizontal axis is usually case temperature or pad temperature; the vertical axis is the maximum allowed forward current. At low temperatures the line runs flat at the rated current. At some threshold, commonly 55C to 85C for power parts, the line slopes downward: every additional degree of case temperature removes a fixed number of milliamps from the allowed current. Below the line is the safe operating area.
The slope is the story. It is set by the thermal resistance from junction to the reference point and by the maximum junction temperature the maker allows. A steep slope means a hot part with nowhere to dump its heat; a flat slope means the thermal path is generous.
Reading It Correctly
Three details decide whether a design sits inside or outside the curve. First, the reference temperature: a curve referenced to the solder pad looks different from one referenced to the case bottom, and the difference is exactly the thermal resistance of that segment. Second, the condition of the reference: a measurement on an ideal infinite heatsink flatters the part; a real PCB with limited copper does not. Third, the pulse conditions, if the plot was made with short pulses rather than DC, because pulsed ratings always exceed continuous ones.
| Curve detail | What to check | Reading trap |
|---|---|---|
| Reference point | Pad, case or ambient | Mixing references between suppliers |
| Slope start | Where derating begins | Assuming the rated current survives at any temperature |
| Slope value | mA per degree | Comparing parts with different package sizes |
| Measurement basis | DC or pulsed | Using pulse numbers for a continuous design |
From Curve to Drive Budget
The honest workflow runs backwards from the application. Estimate the board-side thermal resistance from the pad to ambient for your actual PCB stackup. Add the package resistance from the datasheet to get junction-to-ambient. With the target junction temperature, often 85C to 105C for long lifetime, solve for the case temperature the part will actually sit at during worst-case operation. Then read the derating curve at that case temperature. The current you find there is your ceiling, not the headline rating.
Designers who skip the case-temperature step quote the flat-region rating and discover at qualification that the part runs derated, losing twenty to forty percent of the planned flux.
Margin and Parallel Strings
Leave margin for two things. Thermal stacking, where neighboring parts on the same board heat each other, moves every part’s local ambient up. Production spread moves the actual thermal resistance above the nominal value. A practical rule keeps the design current ten to twenty percent below the derated limit. For parallel strings, remember that current sharing worsens as parts heat unevenly, so the derating margin also buys electrical stability, not only thermal safety.
FAQ
Is the derating curve the same as the lifetime curve? No. The derating curve protects the junction from immediate overstress. Lifetime curves, LM-80 based, describe slow degradation within the safe area. Operating at the derating boundary is legal and still shortens life.
Why do two suppliers with identical packages derate differently? Their allowed junction temperature and their measurement conditions differ. Compare curves at the same reference point and the same basis before ranking parts.
Does pulsing raise the limit permanently? Only within the pulse conditions stated. Duty cycle and pulse width define how much average heat the part dissipates; outside those conditions the DC curve governs.
