0. If you are seeing a specific symptom, start here
This page is written to be read end to end. If you are here because something is already wrong, start below instead. The table maps reflow, thermal cycling and ESD symptoms to the three causes that account for most of them in the field, and to the one measurement that separates them. If the measurement contradicts the table, the cause is outside the range this page covers.
| Symptom you are seeing | The three most likely causes | The one measurement that separates them |
|---|---|---|
| Solder voids appear only on some boards from the same reel | Profile variance between lines, or paste volume outside the qualified window | Compare a good board and a bad board on cross-section |
| Parts fail after ESD handling in the workshop | The handling discipline does not match the device class, regardless of the rating | Audit the bench grounding and the handling sequence |
| Reliability looks fine on the datasheet but not in your application | The qualification covers a different environment from the one you operate in | Compare the qualified environment variables against yours, one by one |
Two things to do before you continue production
- Lower the drive current or the duty cycle before anything else. Almost every symptom below is current- or heat-driven, and reducing both is reversible within minutes.
- Measure before you swap parts. A forward-voltage measurement across the assembly, taken with the board powered, separates a driver problem from a LED problem in under five minutes and costs nothing.
When to contact us
Contact us when the measurement contradicts the table, when the same batch shows the same symptom in more than one assembly, or when you need a failure analysis to close a customer case. Bring the bin label, the forward voltage at operating current, and the ambient temperature at the moment the symptom appeared. Send the details to our engineering team.
The chapters below explain the underlying physics and specification in full.
1. What the reliability data does and does not cover
The ceramic series publish a short list of ratings that a designer has to match during assembly: a manual soldering temperature with a dwell time, an ESD withstand figure, and the temperature ranges the device survives at rest and in service. That list is short, and it is short for a reason. It is a set of limits for handling and mounting, not a set of lifetime results.
There is no lifetime claim anywhere in the ceramic handbook. The handbook says so explicitly under its general notes: lifetime claims are not published, and both lumen maintenance data and rank selection are handled through the sales office. This is deliberate rather than an omission, and it is worth understanding why before reading the rest of this page.
The reason is that lumen maintenance is a property of the specific part number, the bin, the drive current and the junction temperature it actually runs at. A single figure quoted for a whole series would be true for no part in it. The consequence for a buyer is practical: if a design needs an LM-80 figure or a lifetime claim, that has to be requested for the specific part number, and it cannot be read off these pages. For the estimation method itself, see the LM-80 testing and lifetime estimation guide.
What the ratings do cover is the assembly process. Reflow, electrostatic discharge and the thermal environment during mounting are the three places where a ceramic high-power LED is most often damaged, and all three have published limits. The rest of this page works through those limits and what they mean at the line.
2. Reflow: 260 degrees is a peak limit with a 5 second clock on it
The manual soldering temperature is published as 260 ± 10 °C for 3 to 5 seconds. The same value appears in the absolute maximum table on every package page, and it is the number most often misapplied in both directions.
The rating is a peak temperature limit with a dwell condition attached. Three separate things follow from that, and all three are common errors in the other direction.
- The ceiling is 270 °C, not 260 °C. The band is 250 to 270 °C. A profile peaking at 265 °C is inside the rating; a profile peaking at 272 °C is not.
- The 3 to 5 seconds is the total time above 250 °C, not a dwell to be repeated per cycle. Time spent above the lower limit accumulates.
- 260 °C is not a target. Running a profile at exactly the nominal value leaves no margin, and the nominal is a mid-band value rather than a preferred set point. A profile that aims at the middle of the band and holds a tight tolerance is the intent.
The other side of the same point: the rating bounds the peak and the dwell, and says nothing about the rest of the profile. It does not specify a solder paste, a ramp rate, or a soak temperature. Those come from the board, the paste and the oven, and they have to be qualified on the actual assembly. The published value is the constraint the profile has to fit inside, not a profile to copy.
For the thermal side of assembly, the same caution applies to the mount. The junction to board thermal resistance is published at 8 °C/W for the 5050 package, and that figure describes the path above the board only. It is a package measurement on a defined footprint, not a property of a finished luminaire, and it is used here only to explain why reflow is a single-event limit while thermal cycling is a lifetime limit.
3. Thermal cycling: the ceramic path solves heat, and moves stress elsewhere
The ceramic substrate is what makes these packages efficient at heat removal. That same property has a consequence that is easy to miss: a stiff, low-expansion substrate bonded to a PCB with a strain-sensitive solder joint moves the strain out of the package body and into the joint. The thermal path improves; the solder fatigue exposure does not disappear.
The temperature ranges explain the driver. Storage spans −40 to +80 °C and operating spans −30 to +60 °C, both wider than the junction limit of 125 or 150 °C. A luminaire that is switched on and off in an outdoor enclosure cycles the assembly through part of that swing every cycle, and each cycle works the joint.
Two design measures reduce the exposure, and both are standard practice rather than specialist. The first is thermal interface material and copper area on the board, which lowers the swing actually seen at the joint by keeping the whole assembly closer to the ambient. The second is matching the board’s thermal expansion behaviour to the ceramic, which for these packages means a land pattern that gives the joint room to shear rather than pulling it in tension. Neither changes the published ratings; both reduce how much of the rated swing the joint sees in service.
The steady-state mechanism is separate and much slower: holding the junction at its limit degrades the die and the phosphor regardless of how well the joint survives. Thermal management design sets out how to close that budget; the point here is that a part can pass thermal cycling and still fail from sustained over-temperature, and the two mechanisms are diagnosed differently.
4. ESD: a qualification number and a handling discipline are different things
The ESD withstand rating is 2,000 V Human Body Model, and it appears identically in the absolute maximum table on every package page. The same table lists ESD sensitivity as a parameter, which sounds contradictory until the model is understood.
The two figures describe the same part at different stages. A finished LED in a finished package has been qualified at 2,000 V under the Human Body Model, which simulates a person touching a charged assembly. That number does not transfer to the bare die inside the package, and it does not transfer to charged delivery, field handling or a machine that generates a discharge faster than a human body can.
The handbook is explicit that the device is ESD-sensitive and names the three precautions directly: a grounded wrist strap, a dissipative work surface and ionised air during assembly. Those are stated in the handbook’s own note on the rating rather than inferred from it, and they are the minimum, not the target. A 2,000 V qualification means a discharge of that severity will not destroy the part. It does not mean the assembly line can dispense with precautions, because the failure threshold is not the same as the handling threshold and production environments generate events the Human Body Model does not reproduce.
- Grounded wrist strap is the primary control, because the dominant human-borne charge path is direct contact.
- Dissipative work surface and ionised air address charge that arrives without contact, including handling of parts before they reach the grounded point.
- Humidity control is part of the same regime: dry conditions raise generated voltage, which is why the practice is seasonal in unconditioned shops.
Two models of discharge come up repeatedly and they are not interchangeable. The Human Body Model, which is what the 2,000 V rating is qualified against, simulates a person touching a charged assembly through a resistive body path. Charged Device Model simulates a machine or tool discharging directly onto a lead frame, and it is a faster, more local event. The published figure covers the first. Where a process introduces automated handling, pick-and-place or a board-level cleaning step, the second has to be considered separately.
The two discharge models and what each one applies to
| Model | Simulates | What the rating covers | Control |
|---|---|---|---|
| HBM, 2,000 V | A person touching a charged assembly through a body resistance | The published device rating | Wrist strap and dissipative surface |
| CDM | A charged tool or machine discharging onto the lead frame | Not covered by the published figure | Ionised air, controlled handling paths |
5. Failure modes and their signatures
The three mechanisms above produce different symptoms, and distinguishing them saves a misdiagnosis. Because the ratings bound the process rather than the lifetime, the diagnostic evidence has to come from the failure pattern and the assembly record rather than from a rating table.
The four assembly-driven failure modes and what distinguishes them
| Mode | Trigger | Typical signature | Where to look first |
|---|---|---|---|
| Solder fatigue | Repeated temperature cycling | Cracked joint at the pad edge, intermittent open | Joint cross-section, board thermal design |
| Reflow damage | Peak over 270 °C or dwell over 5 s | Voiding, weak bond, die attach separation | Reflow profile logs, cross-section |
| ESD damage | Unprotected handling | Sharp junction or gate rupture, low VF or open circuit | Handling records, ESD audit |
| Steady-state over-temperature | Junction held above its limit | Gradual flux degradation, colour shift | Thermal budget calculation |
The fourth row is on the list because it is the failure most often mistaken for a reliability problem when it is a design problem. Nothing in the assembly changed; the junction simply sat above its limit for a long time. If a device fails without any assembly event preceding it, the thermal budget is the first thing to recheck, not the supplier.
6. Soldering process parameters that are not in the rating
Everything in the preceding sections is a published limit. The parameters below are not published for the ceramic series, and the handbook is explicit that the individual datasheet should be requested for design-critical parameters. They are listed here as the questions to ask, not as values to assume, because assuming them is the most common cause of a reflow problem that no rating table would have predicted.
Process parameters the ratings do not specify
| Parameter | What the pages state | How to resolve it |
|---|---|---|
| Solder paste and alloy | Not specified | Board and process qualified value; request the part datasheet |
| Soak temperature and time | Not specified; ramp only is unconstrained | Paste vendor window, then oven profile |
| Ramp-up and ramp-down rates | Not specified | Oven capability against paste window |
| Peak tolerance across the board | Band is 250 to 270 °C | Thermocouple survey on an actual board |
| Number of reflow passes | Dwell is 3 to 5 s in total | One pass assumed; double reflow needs written confirmation |
| Solder mask and land pattern | Not specified | Board design; thermal expansion matching drives the land geometry |
The fifth row is the one to watch. A double reflow, common in double-sided assembly, doubles the total time above 250 °C and therefore moves the assembly outside the published dwell unless the two passes are individually short. Where a double-sided process is fixed, the part has to be confirmed for it rather than assumed acceptable.
7. What to request before committing to a part
Three requests close most of the gap between what the pages publish and what a design needs. All three come through the sales office, and all three are worth making before the design is frozen rather than after it is built.
The three requests that close the documentation gap
| Request | Why it is needed | When to raise it |
|---|---|---|
| Individual product datasheet | Pages are a family-level reference; specifications are subject to change | Before schematic freeze |
| Lumen maintenance for the part number | The handbook explicitly does not publish lifetime claims | Before the design is committed |
| Rank and bin selection | Acceptable bin is a design decision, routed through sales | At first sample |
| Reflow profile acceptance | Not covered by the published rating alone | When the assembly process is fixed |
| Operating envelope above Topr | Ratings stop applying beyond +60 °C ambient | If the ambient can exceed it |
Two further items are worth confirming in writing when the application is unusual: whether a reflow profile is acceptable for the specific part number, and what the supported operating envelope is when the ambient exceeds the published Topr range. Neither is answered by a rating table, and both are asked often enough to be worth raising early.
8. Frequently asked questions
Can I run a peak reflow temperature of 270 degrees?
That is the upper limit of the published band, so it is inside the rating provided the total time above 250 °C stays within 3 to 5 seconds. A profile that touches 270 °C with a long soak above 250 °C is outside the rating even though the peak is legal. Tolerance on the peak is not the same as tolerance on the dwell.
Does the 2,000 V ESD rating mean my assembly line does not need precautions?
No. The rating qualifies the packaged device against a Human Body Model discharge. The handbook states in the same note that the device is ESD-sensitive and requires a grounded wrist strap, a dissipative work surface and ionised air. The qualification number and the handling discipline answer different questions, and both apply.
Where does thermal cycling damage actually show up?
In the solder joint and the die attach, not usually in the ceramic. The substrate holds the die rigidly and conducts heat well, so it does not flex with the board; the strain goes into the joint at the pad. A cross-section of a joint that has seen a lot of cycles shows the crack at the pad edge.
Why is no lifetime figure published if the packages are qualified?
Because lumen maintenance is a property of the specific part number, bin, drive current and junction temperature rather than of the series. A single series-level figure would not describe any individual part. The handbook states that lifetime claims are not published and that LM-80 data is handled through the sales office, which is why requesting it for a specific part number is a normal part of the design process rather than an exception.
Is the ceramic package more reliable than a plastic one?
It behaves differently rather than simply being better. The ceramic path keeps the junction lower for the same input, which slows steady-state degradation. The stiff substrate transfers strain into the solder joint, which makes the mounting process and the board design more important. Both effects are real and they are separate, which is why the process ratings on this page matter as much as the thermal ones.
What changes if the ambient exceeds +60 degrees?
The published operating temperature range ends at +60 °C, so the ratings no longer apply beyond it. That is not the same as the device failing immediately, but it is the point at which the ratings no longer bound the design and the operating envelope has to be agreed for the specific part rather than inferred from the family pages. Raise it with sales while the design is still open.