You blame the fixtures. The fixtures are usually innocent.
You specified GE Lighting. Or you specified a comparable brand. The spec sheets looked clean, the electrician signed off on the wiring diagram, and for the first few weeks everything looked fine. Then the calls start coming in.
Lights flickering at 40% dim. Can light motion sensors triggering in empty rooms. Fixtures going dark on circuits that tested fine at commissioning. On one job last year, a hotel corridor with shade chandelier sconces and recessed downlights started losing 30% of its output at low dim—three months after handover, with guests already checked in.
The first instinct is to blame the fixtures. It's the visible component, it's the thing that failed, and it's the thing your client can point at. I've sat in those meetings. I get why that happens.
But in four years of reviewing lighting packages before they reach a site—roughly 200+ SKUs a year across commercial, industrial, roadway, and sports projects—I can count on one hand the number of failures that were actually the fixture's fault. The rest? Every single one traced back to something between the components.
What's actually happening when the lights misbehave
People think a good fixture produces good lighting. Actually, a good fixture plus mismatched control components produces lighting that behaves unpredictably—and the unpredictability only shows up under real load, real dimming curves, and real time.
I see two patterns again and again.
Pattern one: dimming compatibility that passes on paper but fails in the field
One spec sheet says "compatible with 0-10V dimming." Another says "provides 0-10V control output." On paper, that's a match. In reality, 0-10V is an analog low-voltage signal, and the actual working voltage varies by manufacturer, and the driver's threshold window is narrower than most controller output tolerances.
When you introduce a GE Lighting autotransformer into the loop—say, to step voltage for a mixed-generation installation—the interaction gets even more fragile. I've seen an autotransformer add enough inductance to shift the dimming curve just outside the driver's usable range. Nothing was broken. Everything was "compatible." The lights still flickered.
The assumption runs the other way: "If each component meets its spec, the system will work." The reality is that a system is only as good as its worst interface, and interface behavior is almost never on the spec sheet.
Pattern two: sensors that behave differently together than alone
Can light motion sensors are a common source of this. A fixture with integrated emergency backup has a standby current draw that can be meaningfully different from the standard sensor circuit the manufacturer tested against. I rejected a pallet of recessed cans in early 2024 because the voltage drop between them and the specified sensors starved the sensor during specific dim states—not all dim states, just some. You'd walk into a room, the lights would be off, and they'd snap on the moment you stepped out.
Those are the calls that ruin a project manager's week. The client thinks the system is haunted. The electrician thinks the sensor is defective. The sensor manufacturer blames the driver. And everyone is technically right, because nobody tested the whole chain together.
What this actually costs you
Here's the part most project owners underestimate.
A 64,000 sq ft warehouse lighting project I was involved with used a GE Lighting/dimming setup with autotransformer stepping between the dimmer racks and the driver banks. The plans reviewer approved it because every individual component listed the right specs. Fast forward to post-install: about 30% of the fixtures flickered at low dim. The electrician swapped sensors. Swapped dimmers. Swapped fixtures. Nothing fixed it.
The bill: roughly $18,000 in diagnostic electrician hours over six months, $6,400 in replacement parts, and—this is the one that hurt—a three-week delay on the certificate of occupancy. The warehouse couldn't open. The tenant couldn't move in. Revenue delayed at $4,000 a day.
Looking back, I should have pushed harder for system-level testing before install. At the time, it seemed redundant—every component passed its spec check, the lighting agent said they'd used the same brand family before, and the schedule was already tight. But given what I knew then, my choice was reasonable. That's the trap. The reasonable choice is often the expensive one.
The cheapest thing you'll do on the next project
So here's what I actually recommend now, and it's not glamorous.
Test the system, not the components. Rent a test rig. Put your actual dimmer, your actual sensor, your actual driver, and your actual fixture on it. Run it through the full dim curve under load. A day of testing typically runs $600–$900. That's less than one electrician callback.
Get the compatibility claim in writing—with test data. A vendor saying "yes, it works" is not data. A dimming compatibility curve under the exact load you're installing is data. Ask for it. If they don't have it, that's the answer.
Clarify who owns the interface. In your contract, the sentence "supplier is responsible for system-level performance" needs to exist. Otherwise, when the sensors ghost-trigger, the fixture vendor blames the sensor vendor, the sensor vendor blames the dimmer manufacturer, and you're the one paying the electrician to referee.
I have mixed feelings about how much of this job is now verification work rather than specification work. On one hand, it feels like an indictment of the industry—that we can't trust component-level claims, and we have to build our own test infrastructure just to specify a system that works. On the other, the projects where we do this verification close cleaner, ship faster, and generate fewer callbacks. The certainty is worth the overhead.
Which, honestly, is the whole point. It's not that rush margins or expedited shipping from the fixture supplier save you. It's that pre-install verification removes the randomness. When your project is on a deadline—and every commercial lighting project is on a deadline—"we think it'll work" is the most expensive sentence in the bid. The fixtures were never going to save you. The system was.