LED Lighting Specifications: IP67 Is Not Better Than IP65

LED lighting specifications are the one place where a supplier can be entirely honest and still lose the order — because half the numbers on a typical spec sheet mean something different to the buyer than they do to the factory. The classic: a buyer asks for IP67 because they think it outranks IP65, and the fixture they receive fails the first hose-down.

It does not outrank it. IEC 60529 says so in writing. Below is every line a buyer's engineer actually checks, what each number legitimately permits, and which test report proves it.

LED Lighting Specifications: The Nine Lines Buyers Check

Line	What a buyer is deciding	Weak entry	Usable entry
Ingress protection	Can it survive their washdown and their puddle	IP67	IP65/IP67 (dual coded per IEC 60529)
Impact resistance	Vandalism, forklifts, sports halls	IP66/IK08	IK08 to IEC 62262 (separate line)
Luminous flux	Real fixture output	6000 lm LED	6000 lm luminaire, LM-79
Input power and efficacy	Energy compliance and running cost	50 W	48.6 W, 123 lm/W, LM-79
CCT and Duv	Whether two units look alike on a wall	3000 K	3000 K nominal, 7-step quadrangle
Colour rendering	Retail and showroom acceptance	CRI 80	Ra ≥ 80, R9 ≥ 0
Beam angle	Whether it lights what they aimed at	60°	60° beam (50%), 96° field (10%)
Lumen maintenance	Warranty exposure	50,000 h	L70 ≥ 50,000 h, B50, TM-21 from LM-80
Marks and documents	Whether it can legally be sold	CE	LVD + EMC + RoHS DoC, EU responsible person named

Everything after this section explains one of those rows.

IP Ratings: The Second Digit Stops Being Cumulative at 6

This is the single most expensive misunderstanding in LED export, and the standard states it explicitly. In IEC 60529, the first digit is cumulative — a rating only applies "if it also complies with all lower degrees of protection." The second digit is not, past 6:

"Up to and including second characteristic numeral 6, the designation implies compliance also with the requirements for all lower characteristic numerals… An enclosure designated with second characteristic numeral 7 or 8 only is considered unsuitable for exposure to water jets (designated by second characteristic numeral 5 or 6) and need not comply with requirements for numeral 5 or 6 unless it is dual coded."

The standard then prints the dual-code table, and uses two words worth putting straight into your quotations:

Passes water-jet test	Passes immersion test	Marking	IEC's own term
5	7	IPX5/IPX7	Versatile
6	7	IPX6/IPX7	Versatile
5	8	IPX5/IPX8	Versatile
6	8	IPX6/IPX8	Versatile
—	7	IPX7	Restricted
—	8	IPX8	Restricted

A "Restricted" enclosure is "considered suitable only for temporary or continuous immersion and unsuitable for exposure to water jets." So a car-park fixture cleaned with a pressure washer needs IP65 or IP66, and a bare IP67 is the wrong product no matter how large the number looks.

Three more IP facts that belong in your reply to any RFQ:

IP68 has no depth. The standard defines numeral 8 as immersion "under conditions which shall be agreed between manufacturer and user but which are more severe than for numeral 7." Any "IP68 = 2 m" claim is a private agreement, not an IEC requirement. State your own tested depth and time.
IP67 is not simply "1 metre." The test places the lowest point 1,000 mm under water for enclosures under 850 mm tall, but only the highest point 150 mm under water for enclosures 850 mm or taller, for 30 minutes. A tall bollard tested to IP67 was barely submerged.
IP tests are type tests. The standard says so directly: "The tests specified in this standard are type tests." An IP rating certifies that a representative sample passed, not that every unit in the container is sealed.

For reference, the jet tests differ by an order of magnitude in water volume: IPX5 uses a 6.3 mm nozzle at 12.5 l/min, IPX6 a 12.5 mm nozzle at 100 l/min, both from 2.5–3 m.

IK Is a Different Standard From IP

"IP66/IK08" written as one rating is a category error that appears on thousands of spec sheets. Impact resistance lives in IEC 62262, a separate document based on the CENELEC publication EN 50102, with its own code letters and its own table. IEC 60529's scope hands mechanical impact away explicitly, listing it among the "matters for the relevant product standard."

IK code	IK00	IK01	IK02	IK03	IK04	IK05	IK06	IK07	IK08	IK09	IK10	IK11
Impact energy (J)	not protected	0.14	0.2	0.35	0.5	0.7	1	2	5	10	20	50

Two corrections while you are updating the template: IK01 is 0.14 J, not 0.15 J — the figure most of the web prints is wrong — and IK11 (50 J) exists, added by the 2021 amendment. Also, IEC's ninth water numeral is written "IPX9"; the "K" in IP69K comes from the automotive lineage, so "IP69K per IEC 60529" is a citation that does not exist.

What "3000 K" Actually Permits

A nominal CCT is a label, not a measurement. Under ANSI/NEMA/ANSLG C78.377, each nominal CCT has a target CCT and a tolerance quadrangle around it. From the 2008 edition's Table 1:

Nominal CCT	Target CCT and tolerance	Target Duv and tolerance
2700 K	2725 ± 145 K	0.000 ± 0.006
3000 K	3045 ± 175 K	0.000 ± 0.006
3500 K	3465 ± 245 K	0.000 ± 0.006
4000 K	3985 ± 275 K	0.001 ± 0.006
5000 K	5028 ± 283 K	0.002 ± 0.006
6500 K	6530 ± 510 K	0.003 ± 0.006

So a fully compliant "3000 K" fixture can measure anywhere from 2,870 K to 3,220 K. Two compliant units side by side can be visibly different, and the buyer who complains is not wrong.

The reason is in the standard's own note: the quadrangles "correspond approximately to those of 7-step MacAdam ellipses." Tighter 3-step or 5-step binning is a voluntary manufacturer promise that goes beyond the ANSI bin — it is not what "meets ANSI C78.377" means. If your buyer is fitting a hotel corridor, sell them the tighter bin explicitly; if they only asked for "3000 K," tell them what that permits before the container ships. (Cite the current edition, C78.377-2024, and attribute the numbers above to the 2008 edition you can actually show.)

Colour is also two-dimensional. Duv, the distance from the blackbody locus, is why two "4000 K" fixtures can read green against pink. A CCT with no Duv is half a specification.

CRI 80 Tells You Nothing About Red

CRI is defined in CIE 13.3-1995, and "CRI" on a spec sheet means the general index Ra. R9, the saturated red sample, is reported separately — which is why serious procurement specifies both. The DesignLights Consortium's current technical requirements state them as two independent thresholds: indoor products need Ra ≥ 80 and R9 ≥ 0; outdoor and high-bay products need Ra ≥ 70, with high-bay allowing R9 ≥ −40, and outdoor products must report R9 regardless.

TM-30 has not replaced CRI. The same document offers TM-30 (IES Rf ≥ 70, Rg ≥ 89) and CIE 13.3 (Ra, R9) as alternative options — either satisfies the requirement. A supplier who claims CRI is obsolete is arguing with the buyer's actual spec.

Fixture Lumens Are Not Chip Lumens

The most common inflated number in the category. Luminaire efficacy is package efficacy multiplied by the luminaire's overall efficiency, and the losses are large. The US Department of Energy's own breakdown:

Product	LED package efficacy	Thermal	Driver	Optical	Overall efficiency	Luminaire efficacy
Troffer	200 lm/W	93%	88%	92%	75%	150 lm/W
Outdoor area light	175 lm/W	90%	92%	87%	72%	126 lm/W
A-lamp	200 lm/W	86%	84%	88%	65%	130 lm/W

DOE puts it bluntly elsewhere: thermal, driver and optical losses "can result in a decrease in efficacy of greater than 30%." LightingEurope lists the four ways this shows up as misrepresentation — module flux quoted instead of luminaire flux, data at 25 °C instead of actual in-fixture operating temperature, module power instead of total input power, and comparing built-in against remote control gear.

For concrete efficacy targets your buyer's incentive programme may impose, the DLC's current minimums are a useful floor: troffer 120 lm/W, linear ambient 125, high-bay 135, outdoor area and roadway 125, interior directional 95 — all measured to LM-79. The same document requires power factor ≥ 0.9, THD ≤ 20%, and a minimum five-year warranty, three lines missing from most factory spec sheets.

A practical note on presenting these numbers: buyers comparing five quotations open five product photos and maybe one PDF. Tools that lock the measured dimensions and rating callouts directly onto the product image — snap-to-edge measurement, then export at each marketplace's image size — put "IP65/IP67 · 126 lm/W · Ra 80 / R9 0" where the comparison is actually made. The reason to place them geometrically rather than generate them is simple: an AI image generator will happily produce a plausible-looking figure that is wrong, and a wrong rating on a product image is a claim, not a typo.

Beam Angle Is the 50% Point, Field Angle the 10%

Two different numbers for the same fixture, and the gap is not small. The IES definitions:

Beam angle — "the angle between the two directions for which the intensity is 50% of the maximum intensity."
Field angle — "the angle between the two directions for which the intensity is 10% of the maximum intensity."

DOE's worked example: a simple cosine distribution yields a beam angle of 120° and a field angle of 169°. There is also a live standards conflict — ANSI/IES RP-16 measures from maximum intensity while ANSI C78.379 uses centre-beam intensity, which diverge for an asymmetric beam. State which one your number is.

Which Report Proves What

Report	What it measures	What it cannot tell you
LM-79 (current: LM-79-24)	Optical and electrical performance of the finished luminaire or lamp: flux, input power, efficacy, colour, distribution	Anything about lifetime. It is "not applicable to LED packages, modules, or arrays"
LM-80 (current: LM-80-21)	Lumen maintenance of the LED package over time at set case temperatures	Nothing about your finished fixture's thermals or driver
TM-21 (current: TM-21-21)	Projects long-term flux maintenance from LM-80 data	It is a projection, not a measurement, and is capped in how far past the tested hours it may be reported

Two things to check before accepting an LM-79 report. First, it is a snapshot: it "does not address lifetime ratings, changing performance over time (e.g., lumen maintenance), or LED case temperature." Second, its contents depend on the apparatus — an integrating-sphere report contains no distribution data (no beam angle, no polar plot, no IES file), while a goniophotometer report often carries no colour data. Ask which was used.

Most Chinese spec sheets still cite LM-79-08, LM-80-08 and TM-21-11. Those are two editions stale, and a buyer's specifier will notice.

L70 Without a B Value Means Half the Fixtures

"L70 50,000 hours" is an incomplete claim. L is the flux threshold; B is the fraction of the population that has crossed it, and the industry default when B is omitted is B50 — half the fixtures. LightingEurope states it plainly: "Median Useful Life is the time at which 50% (B50) of a population of LED based luminaires are flux degraded… generally expressed as Lx so without the B50 notification."

B10 is a much stronger promise than B50 and costs more. And neither letter covers outright death: abrupt failures — a driver that simply stops — are tracked separately as a C value, so a population can meet L70/B50 while a share of units have failed completely. For a concrete benchmark, the DLC currently requires L70 ≥ 50,000 hours for white-light products, evidenced by LM-80 plus TM-21 with an in-situ temperature measurement, or by LM-84 plus TM-28.

Marks and Documents

Market	What is required	The line suppliers get wrong
EU	CE via LVD 2014/35/EU, EMC 2014/30/EU, RoHS 2011/65/EU (Annex I Category 5 = lighting), plus ecodesign and energy labelling	CE is self-declared — LVD states "there is no conformity assessment procedure in this Directive which requires the intervention of a notified body" — but the file still needs real test reports, kept 10 years
EU	An economic operator established in the Union, named on the product or packaging	Regulation (EU) 2019/1020 Article 4 has applied since 16 July 2021. No EU-based responsible person, no legal placing on the market
GB	CE accepted alongside or in place of UKCA, indefinitely, under SI 2024/696	There is no looming CE cut-off for Great Britain
US/Canada	Certification by an OSHA-recognised NRTL (UL 1598 for luminaires, UL 8750 for LED equipment)	UL is a company, not a legal category. Intertek's ETL mark is exactly as valid
US	FCC Part 15 Subpart B, via Supplier's Declaration of Conformity	A non-wireless luminaire has no FCC ID — those attach only to certification grants

An EU buyer who puts their own brand on your fixture becomes the legal manufacturer under LVD Article 10 — worth knowing before agreeing to a private-label deal.

Pre-Quotation Checklist for LED Lighting Specifications

Ingress rating dual coded where both jets and immersion apply (IP65/IP67, not IP67 alone)
Impact rating on its own line, cited to IEC 62262
Flux and power stated for the luminaire, not the LED module, with LM-79 attached
Efficacy calculated from those two numbers, not copied from the chip datasheet
CCT stated with Duv and the binning tolerance (7-step ANSI, or tighter if you actually bin tighter)
Ra and R9 both stated
Beam angle labelled as the 50% point, with field angle beside it
L70 stated with hours and B value, backed by LM-80 + TM-21
Power factor and THD stated
Test report editions current (LM-79-24, LM-80-21, TM-21-21)
EU responsible person named, or a clear note that the buyer must appoint one

Suppliers whose LED lighting specifications fill in every line of that list quote higher and argue less. It is the same discipline that makes a product spec sheet buyers read work, applied to a category where the numbers are unusually easy to fudge — and it pairs with the same problem in wiring, where two conversion conventions leave a 20% gap in AWG vs mm² cable specifications. For what a spec-labelled product image looks like, the annotated product examples are the short version.

FAQ

What should LED lighting specifications include for export buyers?

Nine lines: ingress protection (dual coded where both jets and immersion apply), impact rating cited to IEC 62262, luminaire flux and input power from an LM-79 report, efficacy calculated from those two, CCT with Duv and binning tolerance, Ra and R9, beam angle labelled as the 50% point, L70 with hours and a B value, and the marks plus the named EU responsible person. Anything less and the buyer's engineer fills the gaps with assumptions that favour them.

Is IP67 better than IP65?

No. They protect against different things, and IEC 60529 states that an enclosure marked with second numeral 7 or 8 alone "is considered unsuitable for exposure to water jets." IP65 covers water jets; IP67 covers temporary immersion. A product that must survive both is tested for both and marked with a dual code such as IP65/IP67, which the standard calls a "Versatile" application. A bare IP67 is "Restricted."

How far off can a 3000 K LED be and still be compliant?

Under ANSI C78.377's 2008 table, nominal 3000 K has a target of 3045 K with a tolerance of ±175 K — roughly 2,870 K to 3,220 K — inside a quadrangle that the standard says corresponds approximately to 7-step MacAdam ellipses. Tighter 3-step or 5-step binning is an additional voluntary claim, not part of meeting the ANSI nominal.

What is the difference between LM-79 and LM-80?

LM-79 measures the finished luminaire or lamp — flux, power, efficacy, colour and distribution — at a point in time, and is explicitly "not applicable to LED packages, modules, or arrays." LM-80 measures how the LED package's output decays over time at controlled case temperatures. Neither is a lifetime claim on its own; TM-21 is the method that projects an L-value from LM-80 data.

Does "50,000 hours" mean the fixture lasts 50,000 hours?

No. It normally means L70 at B50: the point at which half the population has dropped to 70% of initial output. Abrupt failures — drivers dying outright — are counted separately under a C value, so units can fail completely while the population still meets its L70/B50 figure. Ask for the B value; if the supplier cannot state one, assume B50.