Ceramic Package LED: Technical Overview

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 ceramic package parts 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.

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.

The ceramic package is not a higher-grade plastic package. It is a different thermal architecture, and every electrical, optical and reliability claim follows from that one structural decision. This overview explains what changes when the die is mounted on an engineered ceramic substrate rather than moulded into a filled epoxy body, and it maps the eighteen ceramic series in the QUEENDOM range to the applications each one actually fits.

Eighteen series, thirty table sets and one absolute-maximum section is a lot of ground. The sections below are ordered the way an engineer actually evaluates a ceramic LED: first the package families and how they differ, then how to read the numbers, then how to choose.

1. What a ceramic package changes

A conventional SMD LED is moulded: the die is attached inside an epoxy body that carries heat away largely by conduction through the mould compound and out through the leads. That path is long, and it is not symmetric across the die area. The centre of a large die runs hotter than its edge, and the resulting temperature gradient across the junction is one of the things that limits both flux and lifetime.

QUEENDOM ceramic packages replace the moulded body with an engineered ceramic substrate — aluminium oxide or aluminium nitride — carrying a metallised heat spreader. Two consequences follow directly. The thermal path is shorter and far more symmetric, so the junction temperature is both lower and more uniform. And because heat leaves through the substrate instead of through the mould, the package accepts substantially higher drive current: the white ceramic series in this handbook are specified up to 1500 mA, against 350-700 mA for typical plastic packages.

That higher drive current is the commercially decisive point. It is not simply “more lumens”. It is lumens per package, which means fewer packages for the same delivered lumens, which means a smaller board, a smaller heatsink and a lower system cost. For a manufacturer building a high-output luminaire, that is the difference between a one-board and a three-board design.

The diagram above is schematic, and deliberately so. It shows the topology of the heat path rather than a specific part: dielectric substrate on top, die attach and die beneath, heat spreader and solder joint at the base. The qualitative point is the arrow length and the symmetry of the stack, not any absolute dimension. Dimensional data for each series is in the individual series sections of the handbook.

2. The eighteen ceramic series, mapped

The range is organised by package size first, then by die count, then by function. Size sets the thermal and mechanical envelope; die count sets current handling and colour capability; function sets the photometric output. The map below places every series on those two axes.

Read the map vertically and the die-count progression is clear: 2020 through 5050 single-die packages, then the 3535 dual- and triple-die parts, then the quad-die multi-colour families at 3535 and 5050, and finally the large-body multi-die white packages at 7070 and 9090. The 7070 and 9090 packages are not simply bigger 5050s: they are where multi-die white construction is taken to its limit for single-emitter high-output fixtures.

The table below lists all eighteen with their die count, part-number prefix and rated drive current.

Two details in that table carry real design consequences. First, the Typ. IF column is not uniform: 350 mA for single-die parts, 700 mA for dual- and triple-die 3535 families, 1400-1500 mA for the 5050 and larger families. A driver designed for the 3535 single-die parts will not drive the 5050 family correctly. Second, the multi-die white 5050 packages appear twice — §14 ball-top and §15 flat — with the same die logic but a different optical profile. Ball-top raises the emission for wide-angle flood applications; flat is the low-profile option for linear and panel luminaires.

3. Test conditions, binning and how to read a datasheet

Every table in the handbook is measured under stated conditions, and those conditions change the numbers. The following table collects them.

Two of these deserve emphasis because they are the usual source of confusion when comparing suppliers.

The first is that luminous flux is a binning range, not a single value. A part specified 80-100 lm does not output 90 lm; it outputs somewhere in that range, and the manufacturer bins parts across it. Any efficiency figure derived from a flux bin is therefore a range too. Treating the bin mid-point as a measured value is the most common arithmetic error in LED datasheet comparison, and it is the reason two datasheets can appear to disagree when they are in fact consistent.

The second is that the lm/W column in the handbook is derived, not measured. It is calculated from the flux bin mid-point and the forward-voltage range at the rated current. That is a useful engineering convenience, but it inherits the width of both inputs: the same part has a real efficacy range roughly three times wider than the single number suggests. Section 4 shows the arithmetic.

Binning is the mechanism behind both points. Manufacturers sort production dies into bins by luminous intensity or flux, dominant wavelength and forward voltage, so a bin code such as “H1” or a three-part intensity/wavelength/VF code defines the exact electrical and colorimetric window you receive. Two practical consequences follow. Where a specification falls between bins, custom binning is the normal route, subject to volume. And for multi-LED assemblies — channel-letter signs, backlit panels, display pixels — a single intensity and wavelength bin should always be specified, or brightness and colour mismatch become visible on the finished article.

What follows if a supplier’s datasheet is read against these conditions. The same colour temperature exists at several current ratings with different flux, so a specification saying only “6500 K, 140 lm” is not a specification: the flux figure is meaningless without the current attached. Comparing efficiency without the test current systematically favours whichever part was measured at lower drive, because forward voltage rises with current and efficiency falls with it. A second common error is comparing packages at equal die count rather than equal package size: a 3535 dual-die and a 5050 single-die may deliver similar flux, but they differ in thermal resistance, drive current and mounting method, and only one of those is a photometric comparison.

Die count drives colour mixing; package size drives thermals and mechanics. Both belong in the comparison, and a datasheet that quotes one without the other is withholding half the information needed to choose between parts.

4. Reading the efficacy numbers correctly

Efficacy in lumens per watt is the figure that decides whether a design is viable. Because it is derived from a flux range and a voltage range, the correct calculation produces a range, not a point. The formula is:

Efficacy minimum = Flux(min) / (VF(max) × IF)
Efficacy maximum = Flux(max) / (VF(min) × IF)

The extremes are the minimum flux bin driven at the maximum forward voltage, and the maximum flux bin at the minimum voltage. Applying this to the handbook data:

Each stated figure sits inside its derived range, which confirms the handbook is internally consistent. The width of the ranges is the real message: the 5050 family is quoted at 250 lm/W but the arithmetic allows anything from 250 to 429 lm/W depending on which part of the bin and voltage spread the sample represents. Design margin should be built against the lower bound.

The chart shows how the derived ranges widen and rise with package size, which is another way of stating that larger ceramic packages are more efficient but less predictable part-to-part. Predictability is often worth more than headline efficiency in a production line, and that trade-off is worth making explicitly rather than by accident.

5. Choosing between ceramic and a plastic package

The decision is usually made on thermal grounds, but the downstream effects run further than that. The comparison below sets out the five points where the two architectures genuinely diverge, and what each one means for the design.

The practical test is straightforward. If the design can be built at 350 mA per package on an SMD bracket with adequate board area, ceramic is unlikely to pay for itself. If it needs high output from a constrained footprint, operates outdoors or at elevated ambient, or has to survive automotive vibration and thermal cycling, the higher drive current and lower junction temperature usually make ceramic the only viable route.

6. Applications and what each one demands

Applications are not interchangeable requirements. Each places a different constraint on the package, and the series that fits one application well is frequently wrong for another.

Two of these warrant specific comment. Medical and automotive applications both demand colour and lumen stability over time rather than a headline figure at 25 °C, which pushes the decision toward parts with documented reliability programmes rather than toward the brightest bin available. Horticulture and new energy applications instead sit at the opposite extreme: sustained high drive, long hours, and ambient temperature that varies through the cycle — exactly the conditions where the ceramic thermal advantage does the most work.

7. Colour options and CRI

The range covers white from 1800 K amber through 3000 K, 4000 K and 6500 K, full monochromatic in red, green and blue, and multi-colour combinations in RGB, RGBW and RGBY. The table below gives the colour options with the CRI figures stated in the handbook.

Note the CRI column. The handbook states Ra 70 for cool white, 80 for warm white and 60 for amber, and does not state CRI for the natural white or the monochromatic parts. If your specification depends on a minimum Ra, confirm the figure for the exact part number rather than assuming it from the colour temperature — Ra does not track CCT monotonically across phosphor types. The CRI figures are handbook values, not a guarantee of a delivered bin; if the application is safety- or specification-critical, ask for the measured distribution rather than the nominal one.

8. Flux bins and colour point by series

The table below collects the binning ranges, colour temperature and stated CRI for the main white and amber families, so the operating points from Section 2 can be read directly as purchasable output.

Two points are worth noting before a design is locked to a value. The same colour temperature appears at more than one flux level — 3030 Cool White and 3535 Cool White are both 6500 K, but at different drive currents and different flux ranges; they are not alternatives at the same output but different operating points. And warm white carries a higher stated CRI than cool white across every family in the handbook, which is the opposite of the usual assumption.

9. How to read a ceramic part number

The part number prefixes in the series table are not arbitrary codes. Each one encodes the package, and where it applies, the die count and the colour or emission type, which is why so many of them share a common root. Reading the code saves a catalogue lookup and, more usefully, exposes the cases where two visually similar parts are electrically different.

Four conventions cover the range in the handbook. The first is the C-series white root, where C followed by the package size and a letter identifies a white part: C2020W, C3030W and C3535W are single-die white packages of increasing size. The second is the letter-suffix monochromatic root, where R, G, B, A and IR appended to the package root give red, green, blue, amber and infrared — C3535R and C3535G are the same 3535 platform as C3535W with a different emitter. The third is the RGB triple-die root, where C3535RGB and its permutations C3535RBG, C3535RYG and C3535RGW place three emitters on one 3535 body rather than three separate packages. The fourth is the W-suffix multi-colour root, where C3535RGBW and C3535RGBY, C5050RGBW and C5050RGBY add a dedicated white die to the RGB set, which is why the white in an RGBW part is specified separately from the coloured dies.

Two consequences follow directly from those conventions. First, package size and die count are independent axes: C5050RGBW is a four-die part and C5050W is a single-die part that happens to share the same package, so the two are not drop-in alternatives despite the identical footprint. Second, a multi-colour part number describes the emitter set but not the bin, and neither does a white part number describe its correlated colour temperature — 3030W covers 3000 K, 4000 K and 6500 K parts alike. The full ordering code has to include the bin and colour designations discussed in Section 3; a package root on its own is a family reference, not a purchasable part.

10. Drive current, junction temperature and derating

Rated current is a ceiling, not a design point. Two effects intervene before a part reaches it in the field: ambient temperature raises the junction temperature, and a mismatch between the constant-current driver set-point and the actual forward voltage spreads current across the bin.

The derating column is the part most often omitted from a specification, and it decides whether a design survives. A part operated continuously at its rating with no ambient derating will exceed its own test conditions the first hot afternoon, and the resulting excess junction temperature accelerates the phosphor and die-attach degradation described in our reliability guidance. If your application cannot tolerate that, specify a lower current or a larger die count rather than relying on the rating.

Current drive also interacts with forward-voltage spread. Because the packages are binned on flux, a batch will contain a range of forward voltages. A resistive driver — a single resistor feeding several packages in series — converts that spread directly into current spread, and the lowest-forward-voltage part carries the highest current. This is the mechanism behind early failure in multi-package arrays, and it is why a constant-current driver is preferred for any array where the parts are not individually binned on voltage.

11. Frequently asked questions

Related products and applications

• High-Power LED product category
• LED Package and Reliability Glossary
• Thermal Management Design Guide for LED Applications
• High-Power LED Failure Analysis
• LM-80 Testing and LED Lifetime Estimation
• Horticulture LED Solutions
• Request ceramic series data, test reports or engineering review

The ceramic package series in detail

This page is the overview. The four pages below take the same series apart in turn: what the ratings are, how to read the electrical and optical characteristics, how to assemble and protect the parts, and how to choose between them. Read in that order if the application is new, or go straight to the one that answers the question in front of you.

• Absolute maximum ratings and current derating — forward current, peak current, power dissipation, reverse voltage, junction temperature, thermal resistance, operating and storage ranges, reflow and ESD. Start here if you need to know what a part may be asked to survive.
• Electrical and optical characteristics — forward voltage by die configuration and colour band, flux ranges at a stated test current, wavelength versus CCT versus CRI, viewing angle, radiant flux, and how to verify a derived efficacy figure against its own limits.
• Reliability and soldering — the reflow rating read as a peak limit with a dwell time, thermal cycling and solder fatigue, the ESD rating as a qualification figure rather than an exemption, and what to request before committing to a part.
• Selection guide — the four questions that narrow a choice, the footprint families and their power classes, the colour and voltage bands, what to do when the part is not in the catalogue, and four worked examples.