What a Photometric File Actually Describes

Why this matters

In a lighting calculation exactly one input was measured in a laboratory. Everything else is a model: the room, the reflectances, the loss factors, the furniture. That one measured input is the photometric file, and because it is the only hard thing in the stack, an error inside it propagates cleanly into every footcandle the report prints and never announces itself. A file that has been scaled by the wrong lumen figure produces a calculation that is internally consistent, professionally formatted, and off by half.

The call

A shop retrofit a small warehouse office and mezzanine, replacing an older lamp-based troffer layout one for one. The calculation supplied with the proposal predicted 35 fc maintained, horizontal, at 2.5 ft above finished floor. At turnover the customer's own meter read 22 fc at the same points, and the shop was told to make up the difference.

The first correction was not a defect, and it made things worse

Before chasing the fixtures, the two numbers had to be put on the same basis. The 35 fc was maintained, carrying a light loss factor of 0.85. The 22 fc was read at turnover on fixtures that had been energized for a week, so it is effectively an initial reading. A maintained figure already contains its loss factors, so the comparable prediction is a re-basing upward:

35 fc / 0.85 = 41.2 fc initial predicted.

Against 41.2 fc, the measured 22 fc is 46.6 percent short, not the 37.1 percent short that 35 against 22 suggests. The correction ran against the shop, which is why it is the first one to make: an honest basis check that only ever helps you is not a basis check.

The second correction was the file

Three things were eliminated first, in this order, because each is faster to rule out than the one after it.

Driver output. The fixtures were confirmed running at full output with no dimming preset engaged, read at the control station rather than at the fixture. Nothing to find.

Loss factors applied twice. The calculation's own report listed a single 0.85 and no second application. Nothing to find.

Room surfaces. The as-built ceiling and wall finishes matched the reflectance triple the calculation had assumed closely enough that the coefficient of utilization would not move by more than a few percent. Nothing to find.

That left the file. Reading the housing label required opening the branch circuit at the panel, locking it out under 29 CFR 1910.333(b)(2), and proving the fixture dead with the live-dead-live sequence of NFPA 70E-2021, 120.5, in the edition the employer's electrical safety program adopts, with the housing reached from a fitted platform rather than the top step of a stepladder. The label matched the catalog number in the proposal.

The file did not. It was relative photometry: a distribution measured on a test source and published with its candela values normalized to an assumed lamp output, which the calculation software then scales by whatever lamp lumen figure you type in. The figure typed in was 5,000 lumens per luminaire, taken from the LED module's package rating. The finished luminaire's own absolute test report gave 2,700 lumens delivered out of the fixture.

That ratio is 2,700 / 5,000 = 0.540, and it decomposes into two things the module rating never contained:

• The luminaire drives the module at roughly 70 percent of the current the package rating was measured at, and because LED output rises sublinearly with current, that yields about 0.75 of rated flux rather than 0.70.
• Optical losses through the lens and the thermal condition inside the assembled housing account for about 0.72 of what the module produces in place.

0.75 x 0.72 = 0.54, and 41.2 fc initial x 0.540 = 22.2 fc, against 22 fc measured. The calculation was arithmetically flawless and its single measured input had been multiplied by a number that belonged to a different object.

What a photometric file contains

A photometric file in the IES LM-63 format, in whichever edition of that format the file was written to, is a candela table plus a short header of test facts. Concretely:

• A candela value for every combination of vertical and horizontal angle on a measured grid, which is the entire optical content of the file.
• The photometric type, which says how those angles are oriented relative to the luminaire.
• Number of lamps, lumens per lamp, and a candela multiplier, which together decide whether and how the table gets scaled.
• Input watts at the tested condition, and a ballast or driver factor.
• Luminous opening dimensions, which is what a spacing check needs for the five-times point-source boundary the inverse-square card owns.
• Test lab, report number and date, and a tilt field.

Everything a calculation reports past that is produced by the software's room model, not by the file.

What it does not contain, and who owns each

• The room. Coefficient of utilization is computed from this file plus a room geometry and a reflectance triple. Owned by the delivered-illuminance calculation, not by the file.
• Maintenance. No loss factor of any kind is in a photometric file. Every candela value in it is initial.
• Color. Chromaticity, correlated color temperature and rendering come from the electrical and photometric test report for the finished luminaire, not from the geometry file, even where a file carries them as text keywords.
• Glare as a judgement. A file supports an outdoor luminaire classification, because that is derived from the candela table alone. An interior discomfort-glare index is not in the file, because it needs the room, the background luminance and an observer position, which the glare card covers.
• Sample variation and production tolerance. The file describes the sample that went in the goniophotometer.
• Dimmed behavior. The table is one operating point. What the distribution and output do at 30 percent dim is a separate question for the manufacturer.

The absolute versus relative fork, stated before you use either

• Absolute photometry tests the whole finished luminaire as one object. The candela values are already the real ones. In the LM-63 format this is conventionally flagged by a lumens-per-lamp field of -1, and when you see it, no lamp lumen scaling is permitted. Typing a lumen figure into the software's lamp field on an absolute file either does nothing or corrupts the result, depending on the tool.
• Relative photometry measures the distribution on a reference source and publishes candela per assumed lamp lumens, so the table must be scaled, and the only defensible scaling figure is the delivered output of the actual luminaire from its own test report.

The trap is that both files open the same way and look identical in a viewer. Check the lumens-per-lamp field before you touch the multiplier, every time.

How it was closed

The shop pulled the manufacturer's absolute test report for the assembled luminaire, re-ran the calculation with the file that matched it, and got 22.4 fc initial, or 19.0 fc maintained at the same 0.85 loss factor. That is the honest capacity of what was installed. The gap to the specified level was then a real quantity to negotiate over rather than an argument about whether the meter was lying, and it was closed by adding luminaires, because no setting on the ones already up there was going to find another 45 percent.

Checking your own figures

• Basis matched before comparison. Prediction 35 fc maintained re-based to 41.2 fc initial at LLF 0.85 before being set against the 22 fc turnover reading. Comparing 35 to 22 would have reported 37.1 percent short against the true 46.6 percent.
• Photometry type identified before scaling. File was relative, so scaling was required; had the lumens-per-lamp field read -1, the 5,000 entry would have been struck outright.
• The scaling ratio closes to the measurement. 41.2 fc x 0.540 = 22.2 fc predicted against 22 fc measured, inside one percent, so the cause accounts for the whole gap and no second cause is needed.
• The ratio decomposes rather than being asserted. 0.75 drive-current term x 0.72 optical and thermal term = 0.540, both sourced to the manufacturer's report for the finished luminaire.
• Conventions printed. Every illuminance in this case carries plane (horizontal), height (2.5 ft above finished floor), and basis (initial or maintained), including the closing 22.4 fc initial and 19.0 fc maintained pair, which are the same number re-based once, not two measurements.
• No loss factor applied twice. The 0.85 appears exactly twice in the arithmetic, once dividing to reach initial and once multiplying to return to maintained, and never stacked on a figure that already carried it.

References

• IES LM-63, the photometric data file format, in the edition the file was written to, which binds through the file itself rather than as a regulation
• IES LM-79, the electrical and photometric measurement method for solid-state lighting products, which is the test that produces a finished luminaire's absolute output, chromaticity and input watts
• Manufacturer photometric test report and absolute test report for the specific catalog number, which own every measured value in a calculation
• 29 CFR 1910.333(b)(2), de-energizing and lockout for work on electrical circuits, and NFPA 70E-2021, 120.5, for the live-dead-live verification sequence in the edition your employer's electrical safety program adopts
• See related: What the Inverse Square Law Does to a Lighting Layout; Why Lumens and Lux Are Not Interchangeable; How to Read a Luminaire Cut Sheet