Definition, test object and service boundary
Photobiological safety evaluation is a laboratory service that measures accessible ultraviolet, visible and infrared emission from a light source, luminaire or finished device under defined conditions, applies the relevant spectral weighting functions, and evaluates potential photochemical or thermal injury to the eye and skin. A result belongs to the tested sample, optical path, operating mode, viewing direction, distance and exposure duration. Electrically powered, broadband, incoherent sources are commonly evaluated within the photobiological safety framework for lamps and lamp systems, while products primarily emitting visible radiation, ultraviolet lamp products and image projectors may fall under dedicated parts of that framework.
The service is not light-product manufacturing, product certification, regulatory approval or a clinical study. It does not establish claims such as sleep improvement, eye protection or circadian benefit. Electrical safety, electromagnetic compatibility, flicker, colour performance, illuminance, visual comfort and non-visual biological effects have separate questions and metrics; completing this test does not cover them by implication. Lasers are normally classified under the laser-product safety framework rather than by substituting a risk group for an incoherent source. Human-subject research, therapeutic exposure and cumulative worker exposure in a real workplace also require separate protocols.
Defining the commission and test plan
Before accepting work, the parties should identify the sales or use jurisdiction, product family, applicable document and edition, whether a risk-group assignment or statement of conformity is requested, and whether the output is for design comparison, type evaluation, supply-chain acceptance or market documentation. The laboratory then reviews the contract against its accredited scope, method capability, measurement range and schedule. Whether a document is voluntary, incorporated into a mandatory product rule, or only a contractual specification must be checked in the relevant jurisdiction; an international-standard report is not automatic access to every market.
Technical inputs should include rated and foreseeable worst-case drive conditions, continuous or pulsed operation, dimming and software modes, optical accessories, warm-up and stability requirements, accessible directions, expected viewing distance and duration, intended users and reasonably foreseeable use. Lenses, reflectors, diffusers, filters and protective windows can change both the spectrum and its spatial distribution. Component data may be transferred to the finished system only when the applicable method permits it and the current, thermal state, optical coupling and assessment geometry meet the transfer conditions. Missing information should become a documented assumption or request for clarification, not an unrecorded choice of a favourable mode.
Measurement and hazard evaluation
A test plan first selects the relevant hazards, which may include actinic ultraviolet effects on skin and eye, near-ultraviolet ocular hazard, blue-light retinal hazard, retinal thermal hazard, infrared hazard to the eye and thermal hazard to skin. Spectral irradiance describes radiant power incident per unit area, while spectral radiance also describes direction and apparent source area; they are not interchangeable. The measured spectrum is weighted by the applicable action spectrum and compared with limits using the prescribed aperture, field of view, angular subtense and time basis. Illuminance, correlated colour temperature or a single blue peak cannot replace that evaluation.
The method should fix distance, orientation, field-of-view limits, receiving aperture, wavelength range and interval, integration time and stability criteria, and it should cover modes capable of producing the greatest accessible emission. Pulsed, scanned, modulated or temperature-dependent products require adequate temporal characterization so that an average does not hide a relevant peak. A risk group expresses how emission relates to limits under standardized time bases and behavioural assumptions; it is neither an accident probability nor a permanent label of absolute safety. Where a product-family standard specifies different geometry, conditions or manufacturing provisions, that more specific scope takes precedence.
Samples, workflow and deliverables
On receipt, the laboratory should record model, serial number or batch, hardware and firmware versions, accessories, physical condition, seals and the configuration supplied by the customer. The approved plan governs mounting and ageing, confirmation of input power, controls and thermal stability, exploratory scans to locate the worst accessible position, and then formal spectral and geometric measurements. If a sample is damaged, a safeguard is removed, or the evaluation intentionally departs from normal use, the reason and the limited applicability must be explicit; an engineering diagnostic configuration must not be presented as the result for a marketed product.
An auditable package normally identifies the commission and sample, standard and edition, dates, environment and operating conditions, equipment and calibration status, measurement geometry, raw or traceably processed spectra, weighted results, limit comparisons, measurement uncertainty, deviations and photographs. If a conformity statement is required, the decision rule addressing uncertainty should be agreed before testing and reported. The report should distinguish measured values, calculations, customer-supplied data and engineering assumptions. A bare word such as “pass,” or a certificate-like page showing only a risk group, is not enough for technical review.
Quality control and measurement uncertainty
Credible results require spectroradiometers, radiometers, apertures and field-of-view systems suited to the wavelength range and dynamic range, supported by traceable calibration of wavelength, spectral responsivity and radiometric quantities. Controls should address dark signal, ambient light, spectral stray light, wavelength error, bandwidth, detector non-linearity, saturation, polarization, distance, alignment, source non-uniformity and stability drift. Stray light is especially important when intense visible emission accompanies very weak ultraviolet emission, because it can inflate the apparent ultraviolet signal. A long decimal display is not evidence of small uncertainty.
The quality system should also cover authorized personnel, method verification or validation, intermediate equipment checks, reference sources, repeat measurements, validation of software and calculation sheets, protection of original records and review of anomalous results. Near a limit, uncertainty and the decision rule can change the conformity statement; the response may be additional repeats, improved geometry or a more suitable range, not rounding toward a preferred outcome. Accreditation supports competence only for the activities listed in its scope. A purchaser should confirm that the relevant standard, product and test are included rather than relying on the presence of an accreditation logo.
Data, operational safety and compliance limits
Testing can itself create intense ultraviolet, blue-light or infrared exposure. The laboratory should assess the work and prioritize enclosed beam paths, shielding, interlocks, remote operation, time limits and controlled access before selecting eye, face or skin protection for residual risk. Hot surfaces, high-voltage supplies, moving parts, sample rupture and ozone may require simultaneous controls. Personal protective equipment does not replace a feasible engineering control, and the condition and disposition of the sample after testing should be recorded.
Original spectra, photographs, firmware identifiers, calculation files and reports are technical evidence requiring access control, backup, version history, integrity checks and an agreed retention period. Routine bench testing normally needs no personal health data. If the commission expands to observations of participants, patient exposure or workplace monitoring, ethics review, informed consent, occupational-health duties and data-protection requirements may apply; authorization for an ordinary sample test is not enough. A product report likewise cannot replace an employer's assessment of multiple sources, reflections, distances and cumulative exposure over a work shift.
Interpreting evidence and purchasing the service
Report review should first establish whether the sample represents production, then examine the edition, hazards, worst-case mode, distance, field of view, angular subtense, exposure duration, uncertainty and decision rule. Results from different laboratories or products are comparable only when those conditions are comparable. A lower risk group or conformity with one limit does not establish safety at every distance and duration, with safeguards removed, or during servicing; nor does it prove sleep benefit, circadian compatibility or visual comfort. A higher group should be interpreted with the applicable product requirements, information and controls rather than used as an isolated alarm.
When buying the service, request the precise accredited scope, experience with similar sources, usable wavelength and dynamic ranges, close-distance capability, treatment of pulsed output, sample count, retest and change rules, depth of data delivery and subcontracting arrangements. Quotations should separate screening, formal measurements, additional modes, post-design retesting and any conformity statement so that a low price does not conceal a narrow scope. Changes to the source, driver, optics, enclosure opening, software power limit or supplier should trigger a documented review and, where necessary, retesting. The laboratory report is one item of evidence for risk and compliance decisions, not an official certification, permanent guarantee or proof of product efficacy.