When a client calls me to say the lights are flickering, dimming is jumping around, or half the fixtures on a run just stopped working, my first instinct used to be the same every time: great, another batch of bad LEDs.
I'd call the supplier. I'd request a return authorization. I'd start mentally preparing the argument about who eats the cost. And then, more often than not, I'd find out the LEDs were fine. The problem was sitting quietly in a metal box, mounted inside the ceiling, doing exactly what it was told—except it had been told the wrong thing.
I've been doing procurement and project management for lighting installations for seven years now. In that time, I've documented eleven significant screw-ups that cost our shop somewhere around $18,000 in rework, rush fees, and discounts we had to eat. If I'm being honest, about seven of those eleven traced back to the same root cause: we treated the power supply as a checkbox item instead of a design decision.
The Surface Problem: Everything Looks Like an LED Issue
Here's what it looks like from the ground level. You've got a hotel project—say, 86 rooms, each with a chandelier bedroom setup, recessed downlights, and a few accent spots. The fixtures go in. Power gets connected. The electrician hits the switch.
And then the complaints start.
"The lights are buzzing." "The dimmer doesn't go all the way down." "Two fixtures in 412 are dead already and we just installed them six weeks ago."
Your first move—my first move, anyway—is to blame the LED chip. Cheap Chinese diodes, you think. Or the dimmer. Or the wiring. Or the phase of the moon.
Here's what you need to know: the LED is usually the last thing that's actually broken. And if you've ever spent three days chasing a flicker that turned out to be a driver issue, you know exactly how much that misdiagnosis costs you.
The Deeper Problem: We Were Speccing Power Supplies Like We Were Ordering Copy Paper
The real issue isn't that power supplies fail. It's that we kept choosing them based on a single line item—voltage and current—and then moving on. Like buying a car because it has four wheels.
Let me break down the specific mistakes we made, because I think they'll sound painfully familiar to anyone who manages lighting projects at scale.
1. We treated the 80% load rule as a suggestion
The rule of thumb in the industry is that you shouldn't run a power supply past 80% of its rated capacity for continuous loads. That's not a suggestion—it's basic thermal management. But when you're ordering 200 units and the 60W model is cheaper than the 100W model, it's really easy to justify the math. "We're only pulling 51 watts. It's a 60-watt supply. We're fine."
You're not fine. That supply is going to run hot. It's going to age faster. And if it's mounted in a closed ceiling cavity—which it always is—the ambient temperature inside that cavity might be 15 degrees higher than the spec sheet's test conditions. Now you're running a 60W supply at 51W in an environment that's effectively derating it to 45W. That supply is going to fail. It's not a matter of if. It's a matter of when.
2. We assumed "3-in-1 dimming" meant "works with any dimmer"
This one hurt. Mean Well's 3-in-1 dimming feature—resistor, PWM, and voltage—is incredibly useful. But it's not automatic. Out of the box, the driver expects a specific input type. If your dimmer sends a PWM signal but the driver is set to resistor mode, you get flicker. Or nothing. Or both.
We learned this the hard way on a boutique hotel job in September 2023. The spec called for the Mean Well 12V 5A power supply (the LRS-60-12) across all guest room lighting. Good unit, solid reputation. But we didn't verify the dimming protocol match. The result: 86 rooms with LED strips that flickered like a haunted house on the first dimming scene test.
3. We ignored the 48V conversation until it was too late
For long runs—corridors, lobby features, landscape lighting—you need to consider voltage drop. That's where the Mean Well 48V power supply comes in. Higher voltage means less current over the same wire gauge, which means less drop. Makes sense, right?
Here's the part we missed: every single fixture on that 48V run needs a step-down converter or a driver that can handle 48V input. And those converters have their own efficiency curves and thermal profiles. Our corridor lighting ran on 48V. The converters at each fixture were undersized. Six months later, we were replacing 40% of them.
4. We let the brand do the thinking for us
This is the subtle one, and honestly the one I'm most embarrassed about. When you see Mean Well on the label, there's a tendency to relax. They're a known quantity. They've been around forever. Their stuff works.
But "works" and "works in your specific application" are two different things. A brand reputation doesn't replace a compatibility check. It doesn't replace a thermal calculation. It doesn't tell you whether the driver's minimum load matches your fixture's standby draw.
I have mixed feelings about this realization. On one hand, it feels disloyal to a brand that's generally served us well. On the other, I've seen too many projects fail because someone trusted the logo instead of the spec sheet. Trust is good. Verify is better.
The Cost: $4,200, Three Weeks, and a Very Awkward Client Call
So here's what that September 2023 hotel project actually cost us.
The flickering dimmer issue—86 rooms, all with LED strips that needed replacement drivers because we'd ordered the wrong dimming configuration. That was $2,800 in replacement units, plus the labor to swap them, which meant pulling down sections of ceiling in every single room. Add another $1,400 for the electrical contractor's time.
Total direct cost: $4,200. Plus three weeks of delay. Plus the moment when the hotel's operations manager walked into a room during the testing phase, saw the flicker, and asked me—very politely—whether we actually knew what we were doing.
That moment right there? That's the one that stuck. Because when a client sees flickering lights, they don't think "ah, the power supply's dimming protocol wasn't configured correctly." They think "these people don't know what they're doing."
And they're not wrong to think that. The output is the brand. If the output is inconsistent, the brand is inconsistent. You can't explain your way out of a flickering light. The product speaks louder than the spec sheet.
We spent the next two months re-testing every driver and dimmer combination on the project. And then we spent another two weeks building a checklist that we now refuse to skip.
The Fix: A Four-Question Process That We Actually Follow
I'm not going to give you a white paper on power supply selection. There are actual engineers for that. What I can give you is the shop-floor process that has prevented every one of these issues since we started using it in January 2024.
Question one: What's the dimming protocol? Not "does it dim?" but "what signal does the dimmer send, and what signal does the driver expect?" We verify this in writing before any purchase order goes out. No exceptions.
Question two: What's the real ambient temperature? Not the room temperature. The temperature inside the enclosure, inside the ceiling, on the hottest day of summer. We derate accordingly. If we're above 80% load in that environment, we go up a size.
Question three: What's the full chain? For 48V runs, we map every converter, every driver, every fixture. If we can't verify the efficiency and thermal behavior at each step, we don't use that configuration.
Question four: Who signs off? It's not the project manager. It's not the procurement person. It's the person who actually understands the electrical design. We didn't have a formal approval chain for power supply specifications before 2024. That cost us when we realized nobody had actually checked the dimming configuration on 86 rooms. Now we do.
Here's what I'd argue is the most important part: we stopped thinking of the power supply as a component and started thinking of it as infrastructure. The LEDs are what people see. The driver is what makes it work. If you get the driver wrong, the LEDs—no matter how good they are—will look terrible.
And if you're wondering about the beam angle question on those downlights—whether it should be 24 degrees, 36 degrees, or 60 degrees—yes, that matters too. A 24° LED downlight will throw a tight, dramatic pool of light. A 36° is more forgiving for general lighting. The spotlight background you're trying to create with accent lighting will look completely different depending on the beam angle and the driver's dimming smoothness. We spent a week debating beam angles on that hotel project. In hindsight, we should have spent that week on driver selection. Beam angle affects the look. Driver selection affects whether the fixture works at all.
So that's my story. $4,200, three weeks of my life, and a very polite but devastating question from a client. If you take anything from this, take the checklist. Trust me on this one: the power supply is not the part you get to skip.
