How to Read a Luminaire Cut Sheet

Why this matters

A cut sheet is a sales document that contains laboratory results. Both halves are on the same page in the same typeface, and nothing marks the boundary. Read it as one document and you will compare a family's best configuration against a competitor's honest one, pick the wrong product, and find out at turnover when the meter comes out. The skill is not knowing more numbers. It is sorting the numbers on the page into three piles before any of them is allowed into a comparison.

Three passes, and what each is allowed to decide

Work the sheet three times, and let each pass answer one question only.

• Identity. Which object am I actually buying? Nothing else can be settled until this is.
• Measured. Which numbers are traceable to a named test, on a named sample, at a named condition? Only these may be compared between products, and only when the conditions match.
• Claimed. Which numbers are projections, family maxima, category words or marketing roundings? These may inform a question to the manufacturer. They may not carry a decision.

The order matters. A measured number attached to the wrong ordering code is worse than no number, because it looks defensible.

Pass one: identity

A catalog number on a cut sheet is usually a family, not a product. The ordering matrix underneath it multiplies out into dozens or hundreds of real objects, and the headline figures on the front page belong to whichever one flatters the family most, which is typically the highest color temperature, the lowest color rendering, and the highest drive current available.

Write down the full ordering code the job needs, every field of it, then re-read the sheet asking whether each headline number applies to that code. Where the sheet gives a table by configuration, use the row. Where it does not, the sheet has not answered the question and the manufacturer has to.

Two identity traps worth naming. First, a marking on a cut sheet is not a listing: a wet-location or ingress-protection claim binds through the product listing and the label on the housing, and through the electrical code your authority having jurisdiction has adopted, in the edition it adopted. Second, the cut sheet and the photometric file can disagree, and when they do the file plus its test report governs, because that is the measured object.

Pass two: measured

Four questions, asked of every number that survives pass one.

1. What test produced it? For a solid-state luminaire's output, input power, chromaticity and rendering, that is an absolute photometric and electrical test of the finished product. For lumen maintenance, it is a long-duration test of the LED package.
2. At what condition? Drive current, ambient temperature, and orientation. A lumen figure taken at a 25 C bench ambient does not survive a ceiling that runs hotter, and the manufacturer's derating curve is a separate document you have to ask for. If the mounting location's ambient is in question, confirm it at height from a properly rated lift or platform with fall protection appropriate to the equipment rather than reaching from a stepladder, and take the reading with the space in normal operation.
3. On which sample? A test report describes the unit that went in the instrument.
4. Is this the same condition as the competitor's? Two honest numbers taken at different drive currents are not a comparison.

The one number that gets a pass of its own is efficacy, because both its numerator and its denominator move with configuration; the efficacy card covers why the headline figure stopped discriminating between products.

Pass three: claimed

The tells are short and reliable.

• "Up to" in front of a lumen figure means it belongs to the family maximum, not to your code.
• A lifetime figure is a projection, not a measurement. The measured input is the package lumen maintenance test, and the projection method that turns it into an L70 or L90 hour figure caps what may be reported at six times the tested duration. A projection quoted past that limit is an extrapolation, and it should be labeled as one.
• L70 and L90 are different criteria and cannot be compared to each other at all. L70 is the point at which output has fallen to 70 percent of initial; L90 is 90 percent. A shorter L90 is a stricter claim, not a worse product.
• Category words such as high output, premium optics or superior uniformity carry no basis and route to nothing. Strike them.
• A warranty term is a commercial promise, not a performance measurement, and what it covers is a question for whoever signs the contract.

The marked-up sheet: two candidates

Same job, same room, two cut sheets.

Candidate A, as printed: up to 15,000 lm, 150 lm/W, L70 greater than 100,000 hours, available 3000K to 5000K, 80 CRI.

Candidate B, as printed: 13,200 lm delivered, 132 lm/W, L90 60,000 hours, 4000K, 80 CRI.

On the sheets, A leads on both output and efficacy and appears to last longer. Now run the passes.

Identity. The job needs 4000K at 80 CRI at the mid drive setting. A's headline belongs to the 5000K, highest-drive code. B published the code it sells.

Measured, pulled from each manufacturer's test report for the exact ordering code:

Claimed, struck: A's 15,000 lm ("up to", family maximum) and A's 100,000-hour figure, which sits at 1.67 times the 60,000-hour reportable limit its own test duration supports.

What the markup changed

The ranking flipped in two places and held in one.

• Output. As printed, A led 15,000 to 13,200. As tested for the codes being bought, B leads 13,200 to 11,600, and A's advertised figure overstates its own tested code by 15,000 / 11,600 = a factor of 1.29, or 29 percent.
• Efficacy. As printed, A led 150 to 132. As tested, A still leads but only 138.1 to 132.0, and its own advertised efficacy overstates its tested code by 8.6 percent. The lead survived; its size did not.
• Lifetime. No comparison is available. One claim is L70 and the other L90, so they measure different things, and only B's sits inside what its test duration supports. What you can say is that A's claim needs either a longer test duration or an explicit extrapolation statement before it means anything.

What would change this conclusion: if the two luminaires have materially different distributions, the tested lumen totals stop being the right comparison, because what matters is what each puts on the plane in this room. That is a different calculation and the efficacy card owns it. Nothing in this markup ranks the two on delivered light; it only establishes which numbers are real enough to feed into that step.

When the report does not exist

The passes assume you can get a test report for the ordering code. Sometimes you cannot, and the answer is not to fall back on the cut sheet.

Escalate once, in writing, asking for the absolute test report for the exact code and the ambient derating curve. A manufacturer selling into commercial work has both, and the request is routine. If what comes back is a report for a different code in the family, that is a pass-three claim wearing a pass-two jacket, and it goes in the struck pile with the "up to" figures.

If nothing comes back, the product is unspecified. Two honest paths: substitute a product whose numbers you can obtain, or carry the risk explicitly in the proposal, naming which figures are unverified and that the delivered level may differ from the calculation. Do not quietly use the sheet figures in a calculation the customer will later measure against. The measurement will happen, and the calculation is the document you will be handed.

One more document sits alongside the other two on many jobs. A rebate or utility program's qualified products list is a third source of numbers, with its own submission rules and its own listed values, and where it disagrees with the manufacturer's report the program's listing is what the incentive pays against, in the edition of the program rules in force on your submission date.

Checking your own figures

• Every compared number traced to the ordering code. A's compared figures are 11,600 lm and 84 W for the 4000K mid-drive code, not the 15,000 lm family maximum.
• Conditions matched before comparing. Both tested at 25 C ambient. Had one been at 25 C and the other at 40 C, the pair would have been struck rather than adjusted.
• Efficacy recomputed from the pair, not copied. 11,600 / 84 = 138.1 lm/W; 13,200 / 100 = 132.0 lm/W. Neither headline figure was used.
• The projection limit printed against the claim. Test duration 10,000 h, reportable limit 60,000 h, A's claim greater than 100,000 h, ratio 1.67, so it exceeds the limit and was struck. B's claim is 60,000 h, ratio 1.00, so it sits at the limit and stands.
• Criteria not blended. L70 and L90 were compared to nothing, because they are different thresholds.
• Overstatement percentages recomputed, not paraphrased. 15,000 / 11,600 = 1.29; 150 / 138.1 = 1.086. Both quoted with their bases.
• Nothing in the example used a class the passes did not define. Every figure in the table is a pass-two measured value or a pass-three claim explicitly marked as struck.

References

• IES LM-79, the electrical and photometric measurement method for solid-state lighting products, which is the test behind a finished luminaire's output, input power and chromaticity
• IES LM-80 and IES TM-21, the package lumen maintenance test and the projection method, in the editions the manufacturer's report names, including the limit on reported projection duration
• Manufacturer test reports and ambient temperature derating curves for the exact ordering code, which own every measured value on a cut sheet
• The product listing and label, and the electrical code as adopted by your authority having jurisdiction in the edition it adopted, which is what a location or ingress rating binds through
• See related: What a Photometric File Actually Describes; What Efficacy Is and Why It Stopped Being the Question; What Color Rendering Actually Measures