Colour temperature consistency is a specification, not a preference. On a chandelier carrying twenty to eighty lamps, one lamp a shade warmer than its neighbours is visible from the far side of the room. Two numbers decide whether a set of lamps matches — CCT and SDCM — and two more decide how the fixture behaves on a dimmer.
Why a 40-lamp chandelier makes colour temperature a specification
A pendant is judged on its own. A chandelier is judged against itself, and that difference changes what has to be written down. The human visual system is far more sensitive to a mismatch between two adjacent white surfaces than to the absolute colour of either one, so once a fixture carries twenty, forty or eighty lamps, the eye stops reading brightness and starts reading difference — a difference nobody notices across a room is obvious across 30 cm. The mechanism is ordinary rather than exotic: lamps arrive from different production batches, individual lamps sit at slightly different temperatures inside the same frame, and a long chain can deliver marginally different current at its far end. Any one of those produces a fixture that reads as patched rather than lit. The practical consequence is that colour temperature has to be settled before production, alongside diameter, drop and finish — not chosen from a photograph after the crate has shipped.
Three things make this harder on a chandelier than on a single pendant.
- The reference is next to the subject. Two lamps 30 cm apart are compared directly, which is the hardest viewing condition there is. Move one lamp to another room and even a five-step difference passes unnoticed.
- The count multiplies the risk. A twenty-lamp fixture gives twenty chances of drawing a lamp from a different bin. A matched batch is the only reliable answer, and it has to be planned at order stage.
- Glass does not hide it. Crystal and art glass transmit and reflect the light they are given. A warm lamp behind clear crystal, next to a neutral one, shows as a colour split across the fixture.
The two numbers that decide it: CCT and SDCM
Two specifications control whether a set of lamps matches, and they measure different things. The first is CCT, correlated colour temperature — the familiar 2700 K, 3000 K, 4000 K. The second is SDCM, standard deviation of colour matching, which describes how far a lamp's actual chromaticity may sit from the target before the difference becomes visible. CCT on its own is a weak guarantee. ANSI C78.377, the standard that defines white for LED products, sets eight nominal CCTs from 2700 K to 6500 K and gives each one an acceptance quadrangle. For nominal 3000 K the target sits at xy (0.4338, 0.4030) — about 3045 K on the Planckian locus — inside a region running from roughly 2870 K to 3220 K with Duv of plus or minus 0.006. Measured in MacAdam steps, that quadrangle is seven to ten steps wide. The usual summary is that CCT gets you into the neighbourhood; SDCM keeps everyone in the same house.
| Binning | What it means when two lamps sit side by side | Where it is normally specified |
|---|---|---|
| 1 SDCM | Below the threshold a trained observer can resolve | Metrology, museum and reference lighting |
| 3 SDCM | Not distinguishable by roughly 95% of observers | The usual professional threshold |
| 5 SDCM | The difference shows when the two fixtures are compared directly | General room lighting, fixtures seen one at a time |
| Unbinned, "ANSI only" | Up to 7–10 steps from the target — see the note below | No colour guarantee at all |
Duv is the perpendicular distance from the Planckian locus, and it captures the green-to-pink tint that CCT alone cannot describe. Two lamps can both be labelled 3000 K and still differ visibly, because one sits above the locus and one below it. That is why a supplier who can quote CCT but not SDCM or Duv has not answered the colour question; they have only named the neighbourhood.
Two practical notes follow from the table. First, the comparison only matters where the two sources are seen together — which, on a chandelier, is always. Second, the acceptance limit should be written into the order, not assumed: a purchase specification that says "3000 K" and nothing else will be satisfied by two lamps that are visibly different. Where the scheme is colour-critical, the limit is usually tightened to 3 SDCM or better; where fixtures are seen one at a time, 5 SDCM is normally sufficient.
What our own catalogue publishes — and what it does not
This is the part most lighting suppliers leave out, so here it is in full. On 5 October 2026 we read the descriptions of all 893 product models listed as active on palerill.com, and counted the models stating each specification. Supply voltage and certification are close to universal: 829 of the 893 models — 92.8% — state 110–240 V supply together with CE, RoHS and ETL or UL certification. Colour temperature is mentioned by 596 models, but only 163 of them state a kelvin figure anywhere in the text, and only 79 place a value directly beside the words "colour temperature" in the specification block. Luminous flux, SDCM binning, a dimming protocol and any flicker metric appear in none of the 893. That is not a boast. It is the list of questions to put to us, and to any other supplier, before a scheme is signed off.
| Specification stated in the product description | Models | Share of 893 |
|---|---|---|
| Supply voltage 110–240 V (US/CA and EU) | 829 | 92.8% |
| Certification: CE · RoHS · ETL / UL | 829 | 92.8% |
| Colour temperature mentioned at all | 596 | 66.7% |
| A kelvin figure stated anywhere | 163 | 18.3% |
| — 3000 K | 96 | 10.7% |
| — 4000 K | 35 | 3.9% |
| — 6000 K | 49 | 5.5% |
| — 2700 K | 1 | 0.1% |
| CRI (Ra) figure | 21 | 2.4% |
| Full-spectrum lighting | 50 | 5.6% |
| 3-colour dimming | 106 | 11.9% |
| Stepless dimming | 9 | 1.0% |
| Continuous dimming | 5 | 0.6% |
| Flicker-free stated as a claim | 5 | 0.6% |
| Luminous flux (lumens) | 0 | 0% |
| SDCM or MacAdam bin | 0 | 0% |
| Dimming protocol (triac / 0–10 V / DALI) | 0 | 0% |
| Flicker metric (Hz, PstLM or SVM) | 0 | 0% |
The kelvin rows overlap: a model that offers a choice of 3000 K, 4000 K and 6000 K is counted in each of those three rows, which is why they sum to more than 163. The reading is a text search of the product descriptions as they are stored in our admin, not a photometric measurement of a delivered fixture — the distinction matters and is repeated in the sources section at the end.
Dimming: why the colour moves, and why it can flicker
Dimming is where a fixture that looks correct at full output starts to look wrong. There are two ways to reduce the light and they behave differently. Pulse-width modulation switches the lamp current fully on and off at a fixed frequency and varies the duty cycle, so the LED stays at its rated current and its operating point barely moves. Constant current reduction — also called analogue or CCR dimming — simply reduces the forward current; the blue pump wavelength shifts by two to four nanometres, phosphor conversion efficiency changes, and the white point drifts warmer. Published comparisons put the difference at roughly plus or minus 50 K across the dimming range for PWM, against a 200 to 600 K warm shift for analogue CCR, with CRI falling five to fifteen points by the time output reaches 10%. Better drivers use both, switching to high-frequency PWM below about 10%.
| PWM dimming | Analogue CCR dimming | |
|---|---|---|
| Mechanism | Fixed current, duty cycle varies | Continuous forward current reduced |
| CCT shift, 100% to 1% | ±50 K | 200–600 K warmer |
| CRI at 10% output | Within 2 points of full output | Drops 5–15 points |
| Flicker risk | Depends entirely on switching frequency | Low — output is DC |
| Where it suits | Colour-critical work, deep dimming | Silent, flicker-free residential dimming |
Flicker: the limits you can write into a specification
Flicker is a measurable quantity, so it can be specified rather than promised. IEEE 1789-2015, the recommended practice for modulating current in LED products, ties the allowable modulation depth to the frequency at which it repeats.
| Frequency band | Low-risk limit | No-observable-effect limit |
|---|---|---|
| Up to 90 Hz | 0.025 × f | 0.01 × f |
| 90 Hz to 1250 Hz | 0.08 × f | 0.0333 × f |
| Above 1250 Hz | No restriction | No restriction |
At 120 Hz — the ripple frequency on a 60 Hz supply, which is what a North American installation produces — the low-risk limit is 9.6% modulation and the no-observable-effect limit is 4.0%. On a 50 Hz European supply the ripple repeats at 100 Hz, where the same limits are 8.0% and 3.3%. Two further metrics appear in European procurement documents: PstLM, defined in IEC TR 61547-1, where values below 1.0 are considered acceptable, and the stroboscopic visibility measure SVM, defined in IEC TR 63158. European ecodesign rules for light sources require PstLM of 1.0 or below and SVM of 0.4 or below. A fixture can dim perfectly smoothly and still fail a camera test, so if the installation will be filmed, the criterion to specify is the measured modulation percentage at the lowest dimming level the room will actually use — not at full output.
What to send us so we can match a batch
Colour matching is decided by three pieces of information, and none of them is a photograph. Send the destination and the supply voltage, because 110–240 V covers both North American and European mains but the dimming hardware does not follow automatically. Send the wall control you intend to use, because a fixture built for a trailing-edge dimmer is not interchangeable with one built for 0–10 V or DALI, and that decision cannot be retrofitted on site. Send the CCT and the tolerance you need, because 3000 K to 3 SDCM and 3000 K to unbinned ANSI are different products with the same label. Add the room and the surface the fixture will be seen against if the scheme is colour-critical: a warm white that reads correctly against pale stone reads differently against a dark coffered ceiling. Fixtures ordered together for one project are produced as a matched batch rather than as separate orders, which is the single most effective way to prevent a colour split.
- Destination and supply — country, mains voltage and frequency, and the ceiling it will be fixed to.
- Dimming control — the dimmer already on the wall, or the protocol you intend to specify.
- Colour target — nominal CCT, and the SDCM or Δu'v' limit you will accept at handover.
- Quantity and phasing — the number of identical fixtures the project needs, and when each one is installed.
- Viewing condition — the room type and the adjacent surfaces, if the scheme is colour-critical.
You can also read how we handle voltage, CSA and UL requirements for North America, and how matched batches and staged delivery work on a hospitality project.
When one colour temperature is not the right choice
There are situations where demanding a single CCT across a whole scheme is the wrong answer, and it is better to say so before a specification is written than to discover it at commissioning. The five below come up most often, and only the first is about the fixture itself; the other four are about the room. None of them means the specification was a mistake — they mean it has to change shape. In each case the alternative is a deliberate decision rather than a compromise, and it is worth writing down which one you have taken. On a project with more than one room, it is also worth checking whether the brief genuinely calls for one colour temperature everywhere, or whether it has inherited that requirement from a 3000 K template without being re-examined.
- Accent and general lighting in one fixture. Where a chandelier is expected to do both jobs, one CCT makes it do neither well. Split the circuit rather than the colour.
- A daylight-dominated room. A 3000 K fixture under a north-facing glazed wall reads noticeably warmer in daylight than the same fixture reads after dark, because the adaptation state of the eye changes.
- A room where the finish is the subject. Where the interior depends on careful material colour — stone, veneer, textiles — a lower CCT with a high CRI usually serves the scheme better than a neutral one.
- Mixed generations on site. Adding two matching fixtures to an installation that already has five older ones rarely matches, whatever the label says. Specify a full replacement set or accept the difference.
- A photographic or broadcast environment. Here the flicker limit, not the colour temperature, is the binding constraint, and it usually forces a different driver rather than a different CCT.
FAQ
What is the difference between CCT and SDCM? CCT names the colour of the white light — 3000 K, 4000 K. SDCM describes how far a lamp may sit from that target, in MacAdam steps, before the difference becomes visible when two lamps are compared side by side. CCT is the neighbourhood; SDCM is the house.
Is 3000 K or 4000 K better for a chandelier? Neither is better in principle. 3000 K suits residential and hospitality interiors and is the value most often specified for crystal; 4000 K suits spaces where task lighting or neutrality matters. What decides the result is not the number but whether every lamp in the fixture sits within the same tolerance of it.
Why does my LED chandelier flicker on camera but not to the eye? Because a phone camera samples light at a different rate from the eye, so modulation that is invisible in person appears as rolling bands on screen. The underlying cause is a driver passing low-frequency modulation through to the lamps — most often mains ripple at twice the supply frequency.
Can all the lamps for one chandelier come from a single batch? Yes, and on repeated or colour-critical work they should. Ask for it at order stage, because a matched batch is planned before production rather than selected afterwards.
Does dimming change the colour of the light? It can. Analogue dimming reduces the forward current, which shifts the white point warmer as output falls — published figures put that at 200 to 600 K by the bottom of the range. Pulse-width modulation holds the colour close to stable, within roughly plus or minus 50 K, because the lamps stay at their rated current.
Sources and numbers
Every figure in this article comes from one of two places: a read of our own catalogue, or a published standard. Nothing here is an estimate dressed up as a measurement, and nothing is quoted from another lighting supplier's marketing. The catalogue figures were produced on 5 October 2026 by reading the stored product descriptions of all 893 active models through our admin interface and counting matches for each specification phrase; the counting script and its output are retained with the project record. The standards figures come from the published documents named in the table, and the dimming comparison comes from industry-published test data rather than from our own laboratory. Where a number describes behaviour that depends on the driver, the lamp and the dimmer together, that dependency is stated rather than averaged away.
| Figure in this article | Value | Source | Status |
|---|---|---|---|
| Active product models | 893 | Admin read, 2026-10-05 | Measured |
| Models stating 110–240 V and CE/RoHS/ETL/UL | 829 (92.8%) | As above | Measured |
| Models stating any kelvin figure | 163 (18.3%) | As above | Measured |
| Models stating 3000 K / 4000 K / 6000 K / 2700 K | 96 / 35 / 49 / 1 | As above | Measured |
| Models stating a CRI (Ra) figure | 21 | As above | Measured |
| Models stating 3-colour dimming / stepless / continuous | 106 / 9 / 5 | As above | Measured |
| Models stating lumens, SDCM, dimming protocol or flicker metric | 0 | As above | Measured |
| Nominal CCTs and quadrangles; 3000 K target xy (0.4338, 0.4030) | — | ANSI C78.377 | Published standard |
| Quadrangle width in MacAdam steps (7–10) | — | ANSI C78.377 analysis | Published analysis |
| 3 SDCM ≈ imperceptible to ~95% of observers | — | Industry convention | Published interpretation |
| Flicker limits by frequency band | — | IEEE 1789-2015 | Published recommended practice |
| 120 Hz: 9.6% low-risk / 4.0% no-effect; 100 Hz: 8.0% / 3.3% | — | IEEE 1789-2015 | Calculated from the published limits |
| PstLM ≤ 1.0, SVM ≤ 0.4 | — | IEC TR 61547-1 / IEC TR 63158; EU ecodesign | Published requirement |
| PWM ±50 K; CCR 200–600 K warmer; CRI −5 to −15 points | — | Industry-published driver comparisons | Published test data |
What this page does not claim
The catalogue figures are a text read of product descriptions, not photometric measurements. A model described as 3000 K has not been measured with an integrating sphere for this page, and the absence of an SDCM figure in a description means the value is not published — not that the fixture fails any particular limit. The standards figures are recommended practices and published limits, not certifications that PALERILL holds: we do not claim IEC 61547-1 or IEC 63158 test reports on every model, and the page should not be read as doing so. The dimming comparison describes behaviour that depends on the driver, the lamp and the dimmer as a system, so treat the ranges as design guidance rather than as guaranteed values for a specific combination. Colour coordinates drift as a fixture warms, so any measurement quoted between two parties should state that it was taken at thermal equilibrium, at the nominal drive current, and on which instrument. Where a figure above matters to your project, ask us for it in writing against the exact configuration you intend to order.