What a Driver Is and Why It Fails First

Why this matters

Most of the lighting work a shop bills after a retrofit is driver work, and most driver replacements are ordered against one number off the old label. That gets the wrong part on the truck often enough to matter, because a driver is a match on three separate ratings, not one, and the two that get skipped are the ones that decide whether the fixture starts cold and how many years the new part lasts in the place you are putting it. A driver's rated hours are stated at a marked case temperature point, not at room ambient, and that single convention is what turns a thirteen-year part into a part that is outside its rating on the day you install it.

What it actually is

A driver is a power converter. Line voltage alternating current goes in, and a regulated constant current comes out, at whatever voltage the load requires within a stated window. That is the opposite of the supply most people picture. A constant-voltage supply holds 12 V or 24 V and lets the load draw what it wants, which is what feeds tape with its own on-board current limiting. A constant-current driver holds the current and lets the voltage float, because an LED's output tracks current and its forward voltage does not stay put.

Three ratings define a match, and all three have to be satisfied by the same part:

1. Output current, in milliamps, matched to the board's design drive current. Many drivers are field-selectable or factory-programmed across a range, so this is a setting as well as a rating.
2. Output voltage compliance window, in volts. The board's string voltage at the operating current must fall inside this window at every temperature the fixture sees, including a cold start.
3. Rated life at a stated case temperature, in hours at a stated Tc, plus a marked maximum Tc that is a different number and a hard limit.

A fourth item is not a rating but is part of the match: the dimming input type, if the circuit dims. What compatibility depends on belongs to the sibling card on dimming compatibility.

Why the driver is the short pole

The usual life-limiting component inside a driver is an electrolytic capacitor, and electrolytic life is governed by the loss of electrolyte through the seal, which is a chemical rate. The rule of thumb that follows is that life roughly halves for each 10 C rise in the capacitor's temperature, at constant ripple current and constant voltage stress, and both of those held-constant conditions matter: raise the ripple by overloading the driver and the internal rise goes up without the ambient moving at all.

Two consequences the lamp era did not have. First, the board and the driver run on different clocks: the board fades on a projection the lumen depreciation card owns, with no step change at the end, while the driver goes from working to dark in one event. Second, switching cycles now cost the driver rather than the source. Fluorescent lamp life was the casualty of frequent switching, which is why long occupancy delays were once defensible; with a solid-state source that argument mostly evaporates, but the inrush charging the driver's input capacitance at every switch-on is a stress the driver absorbs. Where a zone switches many times an hour, the switching count belongs in the driver selection, not in a debate about source life.

The convention that travels: Tc, not ambient

A driver's rated hours are stated at a case temperature measured at a marked point on the driver's own case, at a stated output loading. The mark is there because the number is meaningless anywhere else. Ambient is not Tc: Tc is ambient plus the rise the driver generates in the thermal path it is mounted in, so the same driver in the same room reads a different Tc bolted to a sheet-metal housing than it does in an insulated plenum.

Two numbers come with it and they are not interchangeable:

• The Tc the life figure was stated at. Move away from it and the life figure has to be re-based, not adjusted by subtraction.
• The marked maximum Tc. Past this the life curve does not apply at all, the driver's own thermal protection may fold back output or shut down, and no arithmetic you do is meaningful.

Field measurement of Tc means landing a thermocouple on an energized driver, usually at height, for a value the manufacturer already characterized in that housing. Ask for the thermal data and the ambient derating curve instead.

What the failure looks like from the floor

That last row is the one worth enforcing as a rule. A driver that fails repeatedly in one location is reporting its surroundings. Establish why before you order a fourth.

Worked example: the replacement match sheet, filled in

A discontinued recessed luminaire in a plenum above an insulated ceiling. The board is one series string of 24 LEDs at a 700 mA design drive current. The forward-voltage figures below are read from the board's data, not assumed: 2.85 V per LED at 700 mA at operating temperature, rising to 3.05 V per LED at a cold start.

Candidate A: 700 mA, compliance 50 to 100 V, rated 50,000 h at Tc 75 C, marked maximum Tc 85 C. Candidate B: 700 mA, compliance 25 to 70 V, rated 50,000 h at Tc 75 C, marked maximum Tc 85 C.

Compliance check. A contains 68.4 V and 73.2 V. B contains 68.4 V but not 73.2 V, so B runs the string at reduced current until the LEDs warm and the forward voltage drops into its window. That is the third row of the signature table, arriving as a warranty complaint about a fixture that "takes a minute to come up." B is struck on the cold figure, and the cold figure is the one a match sheet built from a warm measurement never contains.

Correction line, thermal, and it is a re-basing. The 50,000 h already contains a Tc of 75 C. The measured plenum runs 15 C above the ambient the fixture was characterized at, and with the thermal path unchanged that puts the projected Tc at 75 + 15 = 90 C. That is 5 C past candidate A's marked maximum of 85 C, so no life arithmetic is permitted on it: the part is outside its rating on installation day, and the correct output of the sheet is a stop, not a number.

What the two ways out cost. Remote-mount the driver outside the plenum, where ambient returns to 25 C and Tc to 75 C, and the rated 50,000 h stands: 50,000 / 3,744 = 13.4 years. Or select a driver whose life is stated at 90 C with a marked maximum above it, and take that manufacturer's figure at its own stated Tc.

Where the rule of thumb is allowed to run. Suppose the plenum had measured 30 C instead, putting Tc at 80 C, inside the 85 C maximum. The halving rule at constant ripple current and voltage stress gives 2 raised to 5/10, a factor of 1.41, so 50,000 / 1.41 = 35,400 h, which is 35,400 / 3,744 = 9.4 years against 13.4. That is a screening estimate. Where the manufacturer's Tc-versus-life curve disagrees with it, the curve governs and the rule of thumb comes out, because one is measured on that product and the other is a chemistry generalization.

Which clock governs. The board's projection report puts its flux criterion well past 20 years at 3,744 h/yr, so at 13.4 years the remote-mounted driver is the shorter pole and the maintenance plan is built on it, with spares held by ordering code. The projection method behind the board figure belongs to the lumen depreciation card.

Checking your own figures

• Compliance checked at the cold figure, not the hot one. 73.2 V against A's 50 to 100 V window: inside. Against B's 25 to 70 V window: 73.2 exceeds 70, outside by 3.2 V. B struck.
• String arithmetic printed both ways. 24 x 2.85 = 68.4 V, 24 x 3.05 = 73.2 V.
• The thermal correction is a re-basing and the base is named. 50,000 h at Tc 75 C, re-based to a projected Tc of 90 C, which is 5 C above the 85 C marked maximum, so the figure is voided rather than reduced.
• No life number is quoted for the out-of-rating case. Zero hour figures appear against the 90 C projection.
• The rule of thumb carries its held-constant variables and its span. Factor 1.41 at a 5 C rise, at constant ripple current and constant voltage stress, applied only to the 80 C case that sits inside the marked maximum.
• Operating hours recomputed. 12 x 6 x 52 = 3,744 h/yr; 50,000 / 3,744 = 13.4 years; 35,400 / 3,744 = 9.4 years.
• Three ratings named, three ratings checked on both candidates, plus the dimming input, which is named as part of the match and routed to the sibling card rather than resolved here.

Hazards this work creates

• A driver's output can be well over 100 V direct current, and an unloaded constant-current output rises toward the top of its compliance window. Do not meter a running driver output to diagnose a fixture; diagnose by substitution with the circuit de-energized.
• De-energize, lock and tag the branch circuit before opening a luminaire or a driver enclosure, under 29 CFR 1910.333(b)(2) for electrical work on utilization equipment (29 CFR 1910.147 expressly excludes this work from its scope), 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).
• A driver holds charge after disconnection. Wait the discharge time the instructions state, then prove dead at the driver's own output terminals, not only at the input.
• Measuring plenum ambient and mounting a driver above a ceiling 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, never standing on the grid or on a duct.
• A 40 C plenum is a heat exposure for the person in it, so schedule that work for unoccupied hours with the air handling running and with water available, rather than at the end of a shift.

References

• 29 CFR 1910.333(b)(2), electrical work on utilization equipment conductors; 29 CFR 1926.417 is the construction counterpart for lockout and tagging of circuits
• 29 CFR 1910 Subpart D for general industry ladders and walking-working surfaces; 29 CFR 1926 Subpart X for construction
• NFPA 70E-2021, 120.5, as adopted through an employer electrical safety program or by contract
• Manufacturer driver datasheet and Tc-versus-life curve, and the luminaire's thermal characterization data, which own every product-specific value in this subject
• See related: What Lumen Depreciation and Lifetime Ratings Mean; What Dimming Compatibility Actually Depends On; What Flicker Is and Why People Notice What They Cannot See