Specifying Colour Quality Beyond CRI: TM-30 and IEC 62471 for Hospitality and Gallery Lighting

For two decades a single number, CRI Ra, decided whether an LED luminaire was "good enough" for a hotel lobby or an exhibition wall. That habit is now the most common cause of expensive lighting complaints: a fixture can score Ra 90 and still render skin tones grey, shift reds visibly, or fail a photobiological safety review at close viewing distance. This note sets out a practical, submittal-oriented specification method built on IES TM-30 and IEC 62471, written for specifiers, MEP consultants and contractors who have to approve a luminaire schedule rather than read a textbook.

1. What CRI actually measures, and why it fails on site

CIE 13.3 CRI Ra is the mean of eight special colour rendering indices R1-R8, all low-saturation pastel samples. Three structural weaknesses matter in projects:

- **Averaging hides failure.** A source with one badly rendered hue can still average above Ra 90. R9 (saturated red) is excluded from Ra and is routinely negative in otherwise "high CRI" products.
- **Only fidelity, no direction.** CRI penalises any deviation from the reference, but says nothing about whether a hue became more or less saturated. Two fixtures with identical Ra can make the same timber veneer look warm or washed out.
- **Outdated reference and samples.** The reference illuminant switches between Planckian and a daylight phase at 5000 K, and the eight samples are not the materials a hospitality interior actually contains.

The practical consequence: Ra cannot adjudicate a mock-up dispute. TM-30 can.

2. TM-30-15 / TM-30-20 anatomy for a submittal

IES TM-30 evaluates a source against 99 colour evaluation samples (CES) and reports two headline values plus a hue-resolved breakdown.

- **Rf (fidelity index, 0-100):** the successor concept to Ra, computed with 99 CES and a colour-space-uniform metric. Rf 90 is roughly comparable to Ra 90 in intent but far harder to game.
- **Rg (gamut index):** average saturation shift relative to the reference. Rg 100 means no average change; Rg < 100 desaturates, Rg > 100 oversaturates.
- **16 hue bins:** local Rf and hue-angle / chroma shifts per bin, plus the colour vector graphic. This is where you find "reds rotate toward orange" or "blues desaturate" before installation.
- **Annex E / preference designations (TM-30-20):** when a specification targets visual preference rather than strict fidelity, the Rf/Rg combination can be checked against the specified preference ranges.

A usable TM-30 report therefore contains Rf, Rg, the 16-bin local values, the vector graphic, CCT and Duv. A one-line "TM-30 compliant" claim in a datasheet is not a report. Where a project needs an independent reference for photometric format and test-method conventions, the [architectural facade luminaires](https://www.topenglighting.com/) submittal package we issue includes the LM-79 report, IES file and TM-30 report as separate documents, which is the minimum set a reviewer should demand.

3. Setting defensible targets by application

Targets should be written into the luminaire schedule, not left to a sample approval meeting.

| Application | Suggested Rf | Suggested Rg | Additional gate |
| --- | --- | --- | --- |
| Guest rooms, spa, circulation | ≥ 90 | 95-105 | Duv within ±0.003 of the specified locus |
| Lobby, reception, F&B | ≥ 92 | 98-105 | Local hue-bin Rf ≥ 85 in bins covering skin and timber |
| Retail / food display | ≥ 90 | 100-110 | Check bin 1-3 chroma shift against the actual merchandise |
| Galleries, artwork, conservation | ≥ 95 | 97-103 | IEC 62471 risk group + UV/IR content + lux-hour budget |
| Facade / landscape | ≥ 80 | 90-110 | Glare and uplight limits take precedence over colour metrics |

Two additional cautions. First, Rg above ~110 tends to look artificial on human skin even when Rf is high; do not use "vivid" LEDs in guest rooms. Second, high Rf does not guarantee consistency between production batches — binning and a macadam-step tolerance still need to be specified separately.

4. Photobiological safety: IEC 62471 and IEC/TR 62778

IEC 62471 classifies lamps and luminaires into risk groups (RG0 exempt, RG1 low, RG2 moderate, RG3 high) for several hazards; for white LEDs the binding one is usually blue-light retinal hazard, evaluated over a defined radiance and exposure duration. IEC/TR 62778 applies the same framework specifically to LED sources and chains, and is what most LED module and luminaire declarations now reference.

Points that repeatedly cause submittals to be rejected:

- **Measurement distance and geometry.** Blue-light hazard is radiance-based and depends on the stated distance (commonly 200 mm for lamps, or the luminaire's own assessment position). A declaration valid at 500 mm is not automatically valid for a reading light 200 mm from a guest's eye.
- **The final luminaire, not the bare LED.** Diffusers, lenses and reflectors change radiance. A RG1 LED package inside a tight-beam downlight can push the assembled luminaire into RG2.
- **Accessible emission and child occupancy.** RG2 fixtures require labelling and installation restrictions. For guest rooms, bedside reading lights, cove lighting at low mounting height and any interactive installation, specify RG0/RG1 at the actual viewing distance.
- **UV and IR for conservation.** Galleries should additionally request spectral power distribution data to confirm negligible UV content below 400 nm, independent of the visible-light risk group.

Ask for the IEC 62471 / IEC TR 62778 report with the exact test distance, the tested luminaire configuration (including optic and drive current) and the risk group, then check it against the reflected drawing. A "complies with IEC 62471" tick box is not evidence.

5. The submittal checklist that actually closes review comments

A complete luminaire submittal for a colour-critical hospitality or gallery project should contain:

1. LM-79 report from an accredited lab (total flux, efficacy, distribution, CCT, Duv, CRI including R9).
2. TM-30-20 report: Rf, Rg, 16 hue bins, colour vector graphic.
3. IES LM-63 photometric file (and, where BIM is used, the TM-33 / XML equivalent) with the correct orientation and ballast/driver factor.
4. IEC 62471 / IEC TR 62778 declaration with test distance and assembled-luminaire configuration.
5. LED binning statement: chromaticity macadam steps, initial and maintained, plus the supplier's batch-consistency policy.
6. Driver data: flicker percent / PstLM / SVM, dimming protocol and minimum dim level, inrush current.
7. Lumen maintenance evidence: LM-80 dataset and TM-21 projection with the reported ISTMT temperature, plus warranty terms that state the L-value and hours.
8. Sample policy: two production-unit samples from the actual batch, not hand-selected lab units.

Items 1-3 and 5 are what determine whether the mock-up matches the calculation; items 4, 6 and 7 are what determine whether the installation is safe and durable. Treat a missing item as a hold point, not a clarification to be resolved after order placement.

6. Batch consistency and on-site verification

Colour complaints rarely originate in the design; they originate in a second shipment. Three controls are cheap and effective:

- **Specify one chromaticity bin per project zone**, and require the supplier to declare the bin code on each carton. Mixing adjacent 3-step bins across a continuous cove line is visible at night.
- **Measure a sample at handover** with a calibrated handheld spectrometer: CCT, Duv, Rf, Rg and illuminance at a defined reference point. Record the readings against the approved mock-up.
- **Verify dimmed behaviour.** Some drivers shift CCT and Duv at low output, and TM-30 values measured at full output do not describe a 10 % evening scene. Measure at full output and at the scene's actual dim level.

Keep the approved mock-up luminaire, its measurement sheet and the batch code together as the site acceptance record. A signed visual approval with no numbers is not reproducible when a replacement batch arrives eighteen months later.

7. Common failure modes

- **"Ra 95" specified with no TM-30 or R9** — skin and timber complaints, no contractual basis to reject.
- **Rg ignored** — oversaturated LEDs that make food and artwork look unnatural despite high Rf.
- **Risk group declared for the LED package only** — assembled downlight fails review at the actual viewing distance.
- **TM-30 report from a different CCT / optic family** — values do not describe the delivered product.
- **No flicker data** — visible banding on camera and guest discomfort at low dim levels.
- **Samples from a different bin than production** — a perfect mock-up followed by a mismatched installation.

Each of these is preventable at the submittal stage and expensive to fix after handover. Write the metrics, the test conditions and the evidence into the specification, then hold the line during review.

8. Summary

CRI Ra is a screening number, not a specification. For colour-critical hospitality, retail and gallery work, specify Rf with an Rg window, add hue-bin or skin-tone requirements where the material palette demands it, gate photobiological safety with IEC 62471 / IEC TR 62778 at the real viewing distance, and require the full LM-79 / LM-80 / TM-21 / flicker evidence set. Verify at handover with a spectrometer at both full and scene dim levels, and tie the approved sample to a declared production bin. That sequence converts "the lighting looks wrong" from an unresolvable aesthetic argument into a checkable line item.

Prepared by Jack Chen, Topeng Lighting. Technical notes only; project-specific limits should always be confirmed against the local code and the lighting designer's schedule.