TECHNICAL NOTE · CONTROL & DIMMING

Dimming, compared: ELV / TRIAC, 0–10V, 1–10V and DALI-2

Almost every dimming failure in a hotel is decided months earlier, when someone writes the word “dimmable” on a drawing and nobody asks which kind.

WEILUMEN engineering · 2026-10-20 · 11 min read

Dimming is not a feature you add to a light. It is an interface between three parties who rarely meet: the driver inside the fixture, the control gear in the wall or the ceiling, and the person who will stand in the room at 11pm and decide the lighting is wrong. This note compares the four methods used in hotel work in North America and Europe — phase control (forward and reverse), 0–10V, 1–10V and DALI-2 — on wiring, cost, dimming range, flicker and what each is actually good for.

Why the protocol gets decided before the drawings leave the office

The dimming method fixes three things that cannot be changed later without rework: how many wires run into the furniture, how much space the driver needs, and who is allowed to commission the room. A headboard detailed for a two-wire phase dimmer cannot accept a DALI bus without pulling a new cable through a finished wall.

It also fixes the price. On the same 24V linear light, the driver cost difference between a plain on/off unit and a DALI-2 certified one is roughly 25–50%. On a 400-room project, that is not a rounding error — and on a 40-room boutique hotel it may be the difference between winning and losing the job.

The rule we work to: decide the protocol when the furniture drawing is issued, not when the lighting schedule is priced.

Phase control: forward (TRIAC) and reverse (ELV)

Phase-control dimmers sit in the mains supply and chop part of every half-cycle. The lamp sees less energy, so it dims. There are two ways to chop, and in North America they are usually called forward phase and reverse phase; in Europe, leading edge and trailing edge.

Forward phase (TRIAC / leading edge). The switch closes at the start of each half-cycle and opens part-way through, cutting the front of the waveform. It is the oldest and cheapest method, works with resistive loads (incandescent, halogen) and with magnetic transformers (MLV), and — critically for retrofit — it usually works on two wires, with no neutral at the switch. That single fact keeps it alive in every renovation of a pre-1990s building.

The catch is that a TRIAC fires into whatever the load presents at that instant. With a capacitive LED driver the inrush can be violent: audible buzz from the driver, visible shimmer at the bottom of the range, or a driver that simply refuses to start below 20%. Most forward-phase dimmers also need a minimum load — typically 25–40W — and a single 8W LED headboard light will fall below it.

Reverse phase (ELV / trailing edge). A MOSFET or IGBT switches off at the end of each half-cycle instead of switching on at the start, cutting the back of the waveform. Because the circuit is already conducting when it switches, inrush is far gentler and the driver does not sing. This is the correct method for electronic transformers and for the capacitive input of a dimmable LED driver, which is what almost every hotel fixture now contains.

Reverse phase is quieter, dims lower, and tolerates small loads better. It costs more, generates more heat in the dimmer, and in most designs needs a neutral at the switch position — which is exactly what a 1980s guest room does not have.

Naming trap, worth a paragraph in any training document. In the North American trade, ELV dimmer means a dimmer intended for electronic low voltage loads — that is, a reverse-phase dimmer. It has nothing to do with SELV, the safety extra-low voltage class (24V, isolated) defined in IEC 60598. A purchasing department that reads “ELV dimmer” as “safe low voltage” will order the wrong part. If you write ELV anywhere in a specification, write reverse phase (trailing edge) next to it.

What to demand from the supplier. A driver compatibility list, by dimmer model, not a reassuring sentence. Phase-control dimming is the only method where the driver and the dimmer must be tested together; there is no standard that guarantees a random pairing will work. In practice, a project should name the exact wall dimmer in the lighting schedule and require the fixture supplier to confirm compatibility in writing, on a sample, before the order is released.

0–10V and 1–10V: analogue, one-way, cheap and reliable

Here the dimmer does not touch the mains at all. A separate low-voltage pair carries a DC control signal — 0 to 10V — and the driver inside the fixture reads it: 10V means full output, 0V means minimum. The origin is IEC 60929 Annex E, written for dimmable fluorescent ballasts; LED drivers inherited it because it was already there, already cheap, and already understood by every electrical contractor in North America.

Who sources, who sinks. A 0–10V controller sources a voltage; each driver sinks a small current (commonly 0.4–1.5mA). One controller channel therefore drives a limited number of drivers — a typical output is rated 50mA, which is roughly 30–50 fixtures. Exceed it and the voltage sags: the lights stop reaching full output long before anyone notices why.

0–10V does not usually switch off. At 0V most drivers go to 1–10%, not to zero. To actually turn the room dark you need a relay or contactor to break the mains, or a driver specified with a true off position. This is the single most common acceptance failure in hotel rooms: the client presses “off” at night and the ceiling still glows.

1–10V is the near-relative. The signal range is offset by one volt, and the convention is that anything below about 1V is off, so the driver does switch the output down to zero. It is more common on European drivers, less so in North America. The two are not interchangeable in the same way that 0–10V and 0–10V are, and mixing them on one control channel produces a floor of light on half the fixtures.

Polarity and wiring. Classic 0–10V is polarised and, because the signal is referenced to the driver's common, the negative leg is not the same as earth. Reversed polarity gives either no dimming or a fixture stuck at one level. Newer polarity-free 0–10V drivers accept the pair either way and remove an entire class of site error. Shielding is normally unnecessary for runs under 100m in a dry interior, but the control pair must not share a conduit with dimmed mains without regard to induced noise — cheap unshielded cable next to a phase-cut circuit is how a 0–10V room ends up with a visible shimmer at 40%.

Where it wins. Simple, protocol-free, no commissioning software, no addressing, tolerant of long cable runs and of a contractor who has never seen the product before. In a guest room — one circuit, one dimmer, no scenes worth programming — 0–10V is usually the right answer and the cheapest reliable one.

DALI-2: addressable, bidirectional, and more work to commission

DALI-2 (IEC 62386, with the 2 denoting the certified generation tested by DiiA) is a digital bus: two wires, no polarity, roughly 16V DC, carrying a 250mA supply and low-speed data at 1200 bit/s. Because it is a bus and not a signal, topology is free — star, chain, tree — and up to 64 devices share one line, each with its own address.

What that buys you. Every fixture can be addressed individually (64 short addresses), assigned to one of 16 groups, and recalled into one of 16 scenes. It reports back: driver fault, lamp failure, operating hours, and on D4i luminaires (IEC 62386-250 to -253) actual power and diagnostics over the same pair. The dimming curve is logarithmic, matching how the eye perceives light, which is why a DALI fixture can go down to 0.1–1% and still feel like a smooth fade rather than a cliff.

What it costs. Certified drivers are more expensive; the bus needs a power supply and correct current budgeting; and commissioning requires software and someone trained to use it. Addresses must be planned, groups and scenes documented, and the file handed over — otherwise the next technician re-addresses the floor and wipes the settings.

A DALI bus supplies 250mA. Add up the quiescent current of every device on the line before adding the 64th — the symptom of an overloaded bus is not a clean failure but a handful of fixtures that ignore commands, which reads like a software bug and wastes a day.

Where it wins. Anywhere the space changes purpose during the day: ballrooms, meeting rooms, restaurants, lobbies, cove and facade lighting with time-based scenes, and anywhere the operator wants to know a fitting has failed before a guest reports it. In a single guest room it is usually more protocol than the room needs — unless the whole property is on one head-end and the room has to appear in that system.

The four, side by side

 Forward phase (TRIAC)Reverse phase (ELV)0–10V / 1–10VDALI-2
Wires to the fixture2 (mains only)2–3 (neutral at switch)mains + 2 controlmains + 2 bus (no polarity)
Directionone-wayone-wayone-waytwo-way, with feedback
Addressablenonono (per channel only)yes, 64 per line
Typical floor5–10%1–3%1–10% (0V ≠ off)0.1–1%
Small loadspoor (min. 25–40W)goodexcellentexcellent
Buzz / shimmer riskhighlowlow, if cable is cleanlow
Commissioning needednonenonenone (set the driver)yes — software and training
Relative driver cost1.01.21.11.3–1.6
Works in a 1980s retrofityes, on existing 2-wireonly with a neutralneeds new control pairneeds new bus

Cost multiples are indicative and vary by wattage, protocol certification and order volume. They are meant for early-stage budgeting, not for a purchase order.

Flicker is decided by the driver, not by the protocol name

No dimming method is flicker-free by definition. What matters is what happens inside the driver when it is asked for 5%:

  1. Phase control — if the chopped waveform is not clean, the driver's input capacitor sees an irregular supply and the output ripples at twice mains frequency (100 or 120Hz). Visible to some people at the bottom of the range, and always visible on a phone camera.
  2. 0–10V — the signal is smooth, so any flicker comes from the driver's own output stage. A cheap unit dims by low-frequency PWM (a few hundred Hz) to reach 1%; a good one keeps the output current continuous or pushes PWM above 3kHz.
  3. DALI-2 — the digital command is exact, but the driver still decides how it delivers 0.1%. Certification does not guarantee a good curve at the bottom end; ask for the measured curve, not the datasheet floor.

The two numbers worth writing into a specification are Pst LM ≤ 1.0 and SVM ≤ 0.4, the metrics used in the EU ecodesign regulation for light sources. IEEE 1789 is the other common reference. Neither is a guarantee of comfort, but both rule out the worst offenders, and they are objective enough to put in a contract.

Quick site test that needs no instruments: set the room to its lowest usable level and film it at 1/1000s or in slow motion. Bands that crawl across the frame mean the output is modulating. Do it on every dimming type on the project — it takes two minutes and settles most arguments.

Which one, by space

Protocol choice is a cost decision as much as a technical one. The table below is the split we use on hotel projects; it assumes a mid-to-upscale property and a 24V platform in the furniture.

Space / fixtureSpecifyReason
Guest room headboard reading light0–10V, or integrated PWM touch dimmingOne circuit, no scenes, no commissioning. Integrated touch removes the control pair entirely.
Wardrobe interior, drawer, safeDoor contact / PIR with fixed-level outputNobody dims a wardrobe. A sensor switch is cheaper and more reliable than any protocol.
Vanity and mirror light0–10V, or reverse phase on a wall dimmerNeeds a smooth low end for night use; reverse phase only if the wall box has a neutral.
Corridor, cove, plinthDALI-2Long runs, many fixtures, time schedules, and lamp-failure reporting worth having.
Ballroom / function roomDALI-2Scene recall, 1% floor for AV presentations, per-fixture addressing when the room subdivides.
Restaurant, lobby, barDALI-2, or 0–10V per zoneScene changes across the day. Zone-based 0–10V is acceptable if the scenes are few and fixed.
Facade and landscapeDALI-2 or 0–10V, with an astronomical clockSchedule-driven; the protocol matters less than the time control and the surge protection.
Renovation with existing wiringReverse phase if a neutral exists, otherwise forward phaseNo new cabling. Accept the higher floor and specify a compatible driver.

The pattern: rooms get analogue, public areas get digital. A guest room has one occupant, one scene and no facilities team watching it. A ballroom has four partitions, twelve scenes, an AV contractor and a duty engineer who needs to know which of 96 downlights has failed. Paying for DALI-2 in the bedroom and 0–10V in the ballroom is the classic way to spend the budget in the wrong order.

How to write it into the specification

“Dimmable” on its own is not a specification; it is an invitation to substitute. Three sentences that close the gap, usable as they are:

  • Dimming: 1–100% via 0–10V (current sink), 10V = 100% output, 1V = 1%. Driver to be listed on the project dimmer compatibility schedule. 0–10V pair to be polarity-free or clearly marked.
  • Dimming: DALI-2 (IEC 62386-101/102/207), logarithmic curve, minimum 1% output. Driver to hold DALI-2 certification and to report lamp failure and operating hours. Bus power budget to be submitted with the shop drawings.
  • Dimming: reverse phase (trailing edge) only. Forward-phase TRIAC dimming is not accepted for this fixture. Compatibility with the specified wall dimmer to be demonstrated on a sample before release.

Add two project-level clauses and most dimming disputes disappear: no visible flicker at any level between 10% and 100%, and a documented off state (relay, contactor, or a driver that switches to zero). Then state who commissions the scenes and who hands over the file.

Six mistakes we see on site

  1. “Dimmable” with no protocol. The contractor orders the cheapest dimmable driver, which turns out to be forward-phase, and it buzzes behind a headboard at 30%.
  2. ELV read as “low voltage”. A reverse-phase dimmer is bought for a 24V SELV circuit, or a 24V driver is expected to accept a mains dimmer. Both end in a return.
  3. 0–10V assumed to switch off. The room never goes fully dark and the snag list grows by 400 items, one per room.
  4. Control pair run beside dimmed mains, unshielded. A shimmer that only appears at certain levels, and a week of everyone blaming the driver.
  5. DALI bus over its current budget. Some fixtures ignore commands; the fault looks like software, so nobody checks the milliamp sum for two days.
  6. No dimming floor in the spec. “Dims to 1%” is a claim; “1% measured, Pst LM ≤ 1.0” is a requirement. Only the second one survives handover.

Specifier's checklist

  • Dimming protocol stated per fixture type, not once for the whole project
  • Driver and dimmer tested together, on a sample, before the order is released
  • Neutral available at every switch position that needs reverse phase
  • Control pair separated from dimmed mains, or shielded where they share a conduit
  • 0–10V channel current budget checked against the number of drivers per channel
  • DALI bus current budget checked, with addresses, groups and scenes documented at handover
  • Lowest usable level measured, and filmed at 1/1000s, on every dimming type on the job
  • A defined off state — relay, contactor or driver — written into the specification

Questions we get asked

Can a 0–10V driver be dimmed by a phase-cut dimmer?
No. A 0–10V driver expects a DC signal on a separate pair; it has no dimming input that reads a chopped mains waveform. Connecting one to a TRIAC dimmer gives either full output, flicker, or a dead driver. If the room already has a phase dimmer in the wall, the fixture must be specified with a phase-cut driver instead.
Is DALI-2 worth it in a guest room?
Rarely on its own. A single room has one scene and one occupant, and 0–10V or integrated touch dimming does the job for a third of the driver cost. It becomes worthwhile when the property runs one head-end and the room must appear in it, or when the operator wants lamp-failure reporting from the rooms without sending someone to walk the corridors.
Why does my LED buzz only at certain dimmer positions?
Because the driver is being switched into a part of its input range where the internal inductor or capacitor is mechanically resonating. It is a compatibility problem between that dimmer and that driver, not a fault in either product on its own. The fix is a different driver from the dimmer maker's compatibility list, or a change to reverse phase — not a replacement of the same part.
Can 0–10V and 1–10V share one control channel?
They can be wired together, but they will not behave together: at 10V both give full output, while at the bottom of the range the 1–10V units switch off and the 0–10V units sit at their minimum. On a project, keep them on separate channels or standardise on one.
How far can a dimming control cable run?
0–10V is comfortable to about 100m in a dry interior with decent cable; beyond that, voltage drop on the signal pair starts to shift the level and the installer should calculate it like any other circuit. DALI-2 is specified to 300m of bus at 1200 bit/s. Phase control is limited by mains voltage drop and by the dimmer's own minimum load, not by signal integrity.
What dimming range should a hotel bedroom actually have?
A reading light needs a genuinely low floor — 1–3% — because it will be used at night next to someone asleep. General room lighting can stop at 5–10% without complaint. Write the two requirements separately; a single “1–100%” clause across every fixture either over-specifies the ceiling or under-specifies the headboard.

Send the furniture drawing. We will answer with a lighting solution.

Cut-out drawings, photometric files and samples for the lines that fit your detail.

Request a Quote

MORE TECHNICAL NOTES