What Dimming Compatibility Actually Depends On

Why this matters

The carton says dimmable, the control says it is made for LED loads, and the zone buzzes at half, steps instead of fading, drops out near the bottom and will not shut off at all. Nothing on that job is defective. Dimmability is not a property of a lamp, a driver or a control taken alone. It is a property of one specific combination of control device, driver and the circuit those two are sitting in, and only two things establish it: a manufacturer's tested-combination list, and a measurement in the installed fixture at the levels the space will actually run at. Everything else on the packaging is a claim about a component, and a component cannot be compatible by itself.

The four families, and where each one bites

The analog and digital ends fail in completely different places, which is why "it dims fine on the bench" travels so badly. On the digital end the addressing question belongs to the card on what a lighting control system is being asked to do, along with the dimming curve convention that every percentage below is quoted against.

The gate: five items, all five or you have not established compatibility

1. Protocol match, named on both devices. The driver's stated dimming input type against the control's stated output type. Not "both are dimmable."
2. The combination appears on a manufacturer-published tested-combination list, in a stated revision, or you test it yourself in the installed fixture.
3. The circuit is inside the control's load limits: minimum load, maximum load at the LED rating on that sheet rather than the incandescent rating, and the count of drivers within the control's absorbed-current or bus-power budget.
4. The wiring the control requires is present: a neutral where one is needed, a dedicated control pair where one is needed, run within the wiring method and circuit class the luminaire instructions and the adopted National Electrical Code permit.
5. Measured in the installed fixture at the lowest level the space will actually use: the output as a percentage of full at a stated control point on a stated curve, and the percent flicker at that same point at a stated frequency. The flicker card owns that metric.

Items 1 to 4 are paperwork and can be done before anyone leaves the shop. Item 5 cannot be done anywhere but the ceiling it is installed in, and it is the one that gets skipped.

Outcome one: four downlights and a dimmer that is already in the wall

A small conference room. Four recessed downlights at 12 W each, 48 W of connected LED load. The wall box holds an existing forward-phase dimmer on a two-wire switch leg with no neutral. The new drivers state a reverse-phase dimming input.

• Item 1, protocol. Forward-phase control against a reverse-phase-input driver. Fails.
• Item 2, tested list. The driver manufacturer's compatibility list at its current revision does not contain this control. Fails.
• Item 3, load limits. The dimmer's instruction sheet states a 25 W minimum for LED loads. 48 W against 25 W is inside it with 23 W of margin, so this item passes, and it is the item everyone checks. Note what the same sheet does not say: its incandescent wattage rating does not carry across to LED load, which is rated separately and much lower, and some controls cap driver count independently of wattage.
• Item 4, wiring. The controls that do appear on the driver's list require a neutral at the box, and this box has a two-wire switch leg. Fails as the room is wired.
• Item 5, measurement. Not reached; there is nothing to measure until 1, 2 and 4 are resolved.

Three of the four items evaluated fail, and the one that passes is the one that would have been quoted. Installing anyway buys the classic callback set: a snap-on partway up the slider, a step rather than a fade near the bottom, buzz from the box. The resolutions are to pull a neutral and use a control from the driver's list, or to select one on that list the manufacturer qualifies for two-wire installation. Both are hardware decisions and both are cheaper before the ceiling closes.

Outcome two: eighteen drivers on one analog channel

An open-plan floor. Eighteen 0-10 V drivers on one control channel, dedicated control pair, drivers and controller from the same published list.

• Item 1, protocol. 0-10 V driver input against a 0-10 V control output. Passes.
• Item 2, tested list. The pair appears on the controller manufacturer's list at revision 2 dated this year. Passes.
• Item 3, load limits. Each driver's datasheet states it pushes out 0.6 mA on the control pair. 18 x 0.6 = 10.8 mA against the channel's stated absorbed-current rating of 50 mA, leaving 39.2 mA of margin. On this budget alone the channel would hold 50 / 0.6 = 83 drivers, though the relay's load rating and the voltage drop on a long control pair will bind first, so the current budget is a ceiling and not a target. Passes.
• Item 4, wiring. Dedicated control pair, run in the method and circuit class the luminaire instructions and the adopted National Electrical Code permit. Passes.
• Item 5, measurement. Run in the installed fixtures, at one grid point, horizontal at 30 in above finished floor, as-found, with no light loss factor applied to any reading. All percentages below are measured output as a fraction of the full-output reading at that same point, on the linear curve the submittal states.

Correction line, error basis and character. Each percentage is a ratio of two readings on the same meter, so the meter's plus or minus 5 percent of reading is a multiplicative calibration factor that appears top and bottom and divides out. What survives is repeatability, an independent random spread of about 2 percent on each reading, combining in quadrature to about 2.8 percent on the ratio. At 3.0 V the measured 28 percent sits 2 points under the predicted 30, a relative deviation of 2 / 30 = 6.7 percent, outside that spread, so it is a real departure from the published curve rather than noise. Drivers usually depart from a nominal curve, and the measured value is what goes in the record.

At 1.0 V the measured 6 percent against a predicted 10 is a 40 percent relative deviation, and the percent flicker there is 21 percent at 120 Hz against 4 percent at full. That is the bottom-of-range behaviour item 5 exists to find.

Resolution, and it is not hardware. All five gate items are satisfiable with what is installed. The zone's low limit is programmed at 3.0 V, 28 percent measured output on the linear curve, 7 percent flicker at 120 Hz, and the record carries all three figures plus the date. Nobody in that space will ever ask for the bottom four percent, and the combination that was compatible on paper is now also usable in the room.

What voids a gate that already passed

• One driver replaced later with a different model. The tested combination was a pair, and changing half of it ends it. This is how a building that dimmed correctly for three years starts flickering after one warranty swap.
• Fixtures added to an analog channel. The absorbed-current budget is a running total nobody recomputes at the eleventh fixture.
• A mixed load on one control. LED sharing a control with a magnetic transformer or a motor load is a different circuit from the one on the list.
• A longer control run. Voltage drop on the control pair moves the level the far driver sees.

Checking your own figures

• Five gate items stated, and each case run against all five. Outcome one: items 1, 2 and 4 fail, item 3 passes at 48 W against a 25 W minimum, item 5 not reached, so 3 fails of the 4 evaluated. Outcome two: all 5 pass.
• The absorbed-current budget is printed with its source. 18 x 0.6 mA = 10.8 mA of a 50 mA rating, 39.2 mA margin; the 0.6 mA is a per-driver datasheet value, not a rule of thumb.
• The ceiling figure is labelled as a ceiling. 50 / 0.6 = 83 drivers on current budget alone, stated with the two limits that bind before it.
• Error basis and character named before any deviation is judged. Calibration divides out of a ratio; 2 percent repeatability on each of 2 readings combines in quadrature to 2.8 percent; the 6.7 percent and 40 percent deviations are compared against that.
• Every percentage carries its control point and its curve. 28 percent at 3.0 V, 6 percent at 1.0 V, both on the linear curve the submittal states, with the curve convention cited rather than redefined.
• Every flicker figure carries its frequency: 4, 7 and 21 percent, all at 120 Hz. Every reading carries its basis: horizontal at 30 in AFF, as-found, no light loss factor applied.

Hazards this work creates

• De-energize, lock and tag the branch circuit before opening a wall box, a driver enclosure or a control enclosure, under 29 CFR 1910.333(b)(2) for electrical work on utilization equipment, and prove dead with a meter checked on a known source before and after (NFPA 70E-2021, 120.5, binding through your employer's electrical safety program or your contract, not on its own).
• A 0-10 V pair is a low-voltage circuit and the fixture beside it is not. Keep the control conductors in the wiring method and circuit class the luminaire instructions and the adopted National Electrical Code require, so a line-voltage fault cannot appear on a control pair someone will handle as safe.
• Do not drive a zone to its low limit or to off in an occupied area to take these readings. Tell the occupants first, keep the adjacent zones at full, and take the low-end readings out of hours in any space with stairs, machinery or moving equipment.
• Never place any luminaire in an emergency or egress path under a dimming test. Those units answer to a listed device, and the sibling card on emergency lighting circuits owns why.
• Reaching an installed fixture to measure is work at height, on a properly set ladder or lift under 29 CFR 1910 Subpart D for general industry or 29 CFR 1926 Subpart X for construction.

References

• 29 CFR 1910.333(b)(2), electrical work on utilization equipment conductors; 29 CFR 1910 Subpart D and 29 CFR 1926 Subpart X for the height half
• NFPA 70E-2021, 120.5, as adopted through an employer electrical safety program or by contract
• NFPA 70 (National Electrical Code), Articles 410 and 725, in the edition your authority having jurisdiction has adopted, together with the luminaire's own listed installation instructions
• Manufacturer tested-combination lists, by revision, and driver and control datasheets, which own every product-specific value in this subject
• See related: What a Lighting Control System Is Being Asked To Do; What a Driver Is and Why It Fails First; What Flicker Is and Why People Notice What They Cannot See