Marshe

Why This Guide Matters

In January 2026, a street lighting contractor in Qatar replaced about 500 modules, each rated 120W. The lens design was sound — but the lenses yellowed and the modules burned out.

The cause was thermal, not electrical. The heat sink was 200x60x37mm, while the standard 50-60W module in the same family already uses 300x70x40mm. In Gulf summer ambient, 120W on that heat sink has no margin.

What the file did not contain: a rated ambient temperature, a measured Tc, a calculated Tj. Three numbers that would have stopped the order before it shipped. That is the point of this guide. A specification is only worth what the test conditions behind it are worth.

We build custom LED optics and PCB modules in Shenzhen and ship to lighting contractors in more than 60 countries. Below are the seven specifications we would want a buyer to challenge us on.

1. Efficacy: Three Levels, and They Always Step Down

"Efficiency" is not one number. It is three, and a serious datasheet keeps them apart.

• Package level. The bare LED at a stated current and case temperature, often 25°C in a short pulse test. This is the highest number on the sheet.
• Module level. LED plus PCB plus lens, measured hot, no driver. Optical losses of 8-12% and thermal droop of 3-8% pull the number down.
• Luminaire level. Module plus driver plus housing, measured per LM-79. Driver losses of another 8-12% pull it down again.

A realistic chain looks like this: about 160 lm/W at package level, about 140 at module level, about 125 at luminaire level. Each step is lower than the one before. If a sheet shows a module number above its package number, the two numbers were not measured on the same product, or one of them is not what the label says.

That is also how to read a 200 lm/W claim. It is not a lie by default. It is a package-level number at a defined current and temperature, usually a top bin. Ask which level, which current, which temperature, and ask for the LM-80 report behind it. Then compare only numbers from the same level.

On our own sheets the numbers come from different products and different LED generations, so they must not be chained: the 30W SKD kit uses R3 LEDs rated 122 lm/W at package level, while the 50W module is rated 130-140 lm/W at module level with PH 3030 chips. Different SKU, different era, different boundary. Ask any supplier for all three levels on the same SKU, and the comparison becomes simple.

2. Light Distribution: Beam Angle Is Only the Start

Beam angle tells you how wide the light spreads. It does not tell you whether the road passes the lighting class.

What actually gets checked in a project:

• IES file for the exact model. Not a similar model, not a catalog average.
• Light distribution type. Type II for narrow roads, Type III for wider roads, Type IV for wall-adjacent layouts. Our 12W array lens is a 90x126° Type IV.
• Road class results. Average illuminance or luminance, uniformity (Uo), and surround ratio against EN 13201 M2, M3 or M4.
• Glare and spill. BUG rating per IES TM-15, and the uplight ratio. For road lighting, glare is normally expressed as TI (threshold increment) under EN 13201. GR values belong to the CIE 112 sports and area lighting system, not to road lighting.

Two modules with the same wattage and the same lumens can produce very different Uo and glare. If a supplier cannot send the IES file for the exact model, you are buying a picture, not a photometric product.

3. Optical Material: An Honest PMMA vs PC Comparison

Both materials work. The mistake is treating one as premium and the other as cheap.

Practical rules we give customers:

• Use PMMA where the optics run cooler, the gasket is silicone, and maximum transmittance matters.
• Use UV-stabilized optical PC where the module runs hot, the site is coastal or high-impact, or the housing design puts stress on the lens.
• Never pair a PMMA lens with a soft PVC gasket. The plasticizer migrates and the lens cracks at the screw holes. This is a slow failure that shows up months later.

One number here deserves a second look. Transmittance moves with the measurement: a 2 mm molded sample reads differently from the finished lens geometry, and it shifts again with thickness and surface finish. PMMA at 92–93 % is normal. Optical PC at 93 % is not — so when a PC lens is quoted at 93 % or more, ask for the measured report on that exact grade and thickness. This is the first number an experienced buyer re-tests.

4. Thermal: Judge Thermal Resistance and Measured Tc

Two statements you will hear are too simple to be useful: "always use thermal pads, never grease" and "1.6mm aluminium is the minimum".

What actually matters is the thermal path and the measured result:

• Thermal resistance from junction to heat sink. Copper thickness, dielectric conductivity and via design decide this. A thin board with good dielectric can beat a thick board with a poor one.
• Interface performance. A pad gives consistent bond line thickness and is easy to assemble. Grease gives lower thermal resistance when the bond line is thin and the pressure is controlled, but it pumps out over years. Judge by measured interface resistance, not by the material type.
• Heat sink validated at your ambient. Size, fin area and contact flatness must be matched to wattage and to the ambient temperature of the site. The Qatar case failed here: 120W of LED on a 200x60x37mm heat sink, when the same family's 50-60W module already uses 300x70x40mm.

Then ask for the documents that prove it. ISTMT gives you Tc measured at a declared test ambient, usually 25°C, in a stated mounting condition. It is not a site measurement. If the site runs at 45°C or 50°C, the Tc you see in the field is higher, so ask for the derating calculation from the ISTMT point to your ambient, and the resulting Tj.

LM-80 describes the LED package at a fixed case temperature over the tested hours. It says nothing about the module. TM-21 turns LM-80 data into an L70 or L90 projection. LM-82 describes how electrical and photometric characteristics change with temperature. LM-84 covers lumen and colour maintenance at module and luminaire level, and TM-28 projects it.

5. Sealing and Mechanical: IP66 Is a Test Result, Not a Sticker

IP66 per IEC 60529 means dust tight and protected against strong water jets. It is a result of a test on a specific design, not a property of a material.

Ask about the parts that decide it:

• Gasket material and compression. Silicone for high temperature, EPDM where chemical resistance matters, foam only for dry interiors.
• Screw type and torque. Stainless screws for coastal sites, defined torque, and a thread design that does not crack the lens.
• Pressure equalization. A vent stops the module from breathing moisture through the gasket as temperature cycles.
• Impact rating. IK08 per IEC 62262 is a common requirement for road lighting.
• Salt spray. ISO 9227 testing matters for coastal and island projects. Our PC-lens waterproof module uses a silicone gasket and targets exactly these replacement jobs.

6. Driver, Surge and Electrical Compatibility

The driver is usually the first part to fail, and it is the part buyers negotiate hardest on price. That is a bad combination.

• Surge protection: 10kV / 10kA is a sensible baseline for road lighting. Areas with overhead lines and frequent storms need more, and the SPD should be on the line side of the driver.
• Isolation and safety class: Class I or Class II must match the housing and the earthing scheme.
• Input range: AC 100-240V covers most markets. North American projects may need 277V.
• Power quality: PF above 0.9 and THD below 15% are normal tender expectations. Our 30W kit ships with an isolated constant-current driver rated PF above 0.9.
• Thermal derating: a driver rated 60W at 25°C may only deliver 40W at 60°C. Ask for the derating curve, not the headline wattage.

7. Color, Lifetime and Compliance: The Part Most Datasheets Skip

Engineers evaluating a bid look at color and lifetime before they look at price.

• CCT and CRI: state the nominal CCT (for example 3000K, 4000K, 5700-6300K) and the minimum CRI, usually 70 or 80.
• Color consistency: ask for SDCM. Three-step MacAdam is the normal expectation for a project, five-step is a compromise. Mixed bins show up as a street with visible color patches.
• Lifetime evidence: the document set in section 4 — LM-80 with TM-21 at package level, LM-84 with TM-28 at module level, and ISTMT for the measured Tc.Seven specifications, each with a document behind it. That is the difference between a datasheet and a product you can defend at handover.
• Warranty: three years is common for a component sold to a luminaire maker. Five years is increasingly the entry ticket for road tenders and for DLC-listed products, and the warranty has to survive the project's defect liability period. Check who pays labor and freight, not only who replaces the part.

A Specification Sheet You Can Send to Any Supplier

Copy this into your RFQ. It removes most of the room for different interpretations.

• Efficacy: module-level lm/W, at stated drive current and Tc
• Distribution: IES file, type, road class result, BUG or TI
• Optics: material, measured transmittance, gasket material
• Thermal: heat sink size, Rth, ISTMT Tc at declared ambient, derated Tj at site ambient, LM-80 and TM-21
• Sealing: IP rating and test report, IK rating, salt spray
• Electrical: surge level, isolation class, PF, THD, derating curve
• Color: CCT, CRI, SDCM
• Commercial: warranty term and coverage, spare ratio, lead time

Price: Compare at the Same Boundary

A 30W SKD kit with lens, aluminium PCB, gasket and driver sits in a different price band from a bare lens, and a finished luminaire sits in a third band. Quoting them against each other is where buyers lose money.

Compare per watt and per lumen at a defined boundary, and add the cost of what is missing: heat sink, driver, housing, assembly, testing, freight. A quote that looks 20% cheaper often has 20% of the content removed.

FAQ

Is a 200 lm/W claim a red flag? Not by itself. Ask at what current, at what temperature, and at which boundary. Then ask for the LM-79 or LM-80 report behind it.

PMMA or PC for street light lenses? PMMA for maximum transmittance and UV stability in a cooler, sealed design with a silicone gasket. UV-stabilized PC for higher heat, impact or coastal conditions. Both are legitimate; the failure comes from mismatching material and environment.

Is 1.6mm aluminium PCB enough? It depends on copper thickness, dielectric conductivity and the wattage per area. Ask for the calculated Tj and the measured Tc, not a board thickness rule.

Thermal pad or grease? Judge by interface thermal resistance, bond line thickness and long-term stability. Pads are more consistent in production; grease performs well when the process is controlled.

Is a three-year warranty enough? For a component, it is common. For a road tender or a DLC-listed project, expect five years and confirm that labor and freight are covered.

Final Word

Seven specifications, each with a document behind it. That is the difference between a datasheet and a product you can defend at handover.

Sunshineopto has been designing and manufacturing custom LED optics and PCB modules in Shenzhen since 2011, with free LED PCB Booard design on every enquiry. 

Please tell me the application environment and relevant requirements for your lighting fixtures. 

We will send back a specification proposal with the reports attached, usually within one working day.