The Chandelier That Broke Me (and Why I Now Spec Nichia LEDs)

It was a Thursday afternoon in March 2022. I was standing in a high-end residential showroom, looking at a sample for a custom fabric chandelier we'd been designing for three months. The frame was hand-polished brass, the fabric a silk blend from a mill outside Milan. It was, by any standard, going to be beautiful.

I made the mistake of thinking 'beautiful' was enough.

The client, a boutique hotel chain, had approved the design. The budget was set. The timeline was aggressive, but manageable. The only real decision left was the light source. The spec called for 40 warm-white LED modules. The question was: which ones?

I chose the cheapest option with a high lumen-per-watt rating. It was $0.88 per module. The Nichia alternative? $1.75. The procurement team was happy. My project budget looked great. I felt smart. (Ugh.)

The First Sign Something Was Off

The chandelier was installed in June 2022. Sixteen feet wide, suspended over the hotel's grand lobby. The fabric panels diffused the light beautifully—at least, that's what I told myself when I saw the photos.

Then I visited the site in person.

The light looked… flat. Lifeless. The silk fabric, which had this incredible depth under the halogen lamps in our studio, looked like cheap polyester under the LEDs. The shadows were harsh in the wrong places. The color temperature felt sterile, not warm.

I compared it to a smaller, older chandelier in the hotel's bar that used halogen bulbs. The bar chandelier felt alive. The new one felt clinical. Side by side—or rather, in the same building—the contrast was stark.

Seeing our $47,000 chandelier vs. a $12,000 one from the 1990s made me realize that brightness numbers don't predict how a space will feel.

When the Problem Got Expensive

The hotel's design director called me two weeks later. 'The light,' she said, 'it's making the fabric look wrong. The colors seem washed out. Can we fix it?'

I knew exactly what the problem was. The LEDs I'd chosen had a Color Rendering Index (CRI) of about 80. The Nichia 519a modules, which I'd dismissed as 'overkill,' had a CRI north of 95. The difference? The 519a would render the deep reds and greens in the fabric accurately. The cheap modules were basically smearing a gray filter over everything.

The 'fix' was a complete re-lamping. 40 modules removed, 40 new ones installed. Module cost: $1.75 each. But the total cost of the redo included: $380 in labor, a $1,200 fee for a boom lift rental, and the cost of my credibility. The hotel delayed the grand opening by one week. That week cost them, conservatively, $8,000 in lost room revenue. They didn't send me the bill for that last part (thankfully). They just sent me a very polite, very firm email about who would pay for the re-lamping.

The Lesson That Stuck

It's tempting to think you can compare LEDs by looking at lumens and price per unit. But the 'cheaper per lumen' advice ignores one critical variable: how the light makes people feel.

After the re-lamping, the chandelier looked like what it was supposed to be. The fabric had texture and depth. The brass frame caught warm highlights. The space felt inviting rather than institutional.

Here's what I learned, in three specific things:

1. CRI matters more than lumen output for decorative fixtures. You can have 1000 lumens that look terrible, or 800 lumens that look incredible. The 519a's high CRI and R9 value (deep red rendering) are non-negotiable for fabric and warm metals.
2. Never spec a light source without seeing it in the fixture. The module's datasheet said 'warm white 3000K.' But two different modules at 3000K can look completely different. Ask for a sample, put it behind your fabric, and look at it. (I keep a Nichia 519a sample in a box on my desk now—my 'control' sample).
3. The $0.87 difference per module wasn't savings; it was deferred cost. The 'savings' on the original purchase evaporated the moment we had to redo the work. Total cost of ownership (i.e., not just the unit price but installation, disappointment, and rework) is the only number that matters.

The question isn't 'how much does a good LED cost?' It's 'what is the cost of a bad one?'

For me, the answer was $1,580 out of pocket plus a damaged relationship with a client I'd worked for years to win. That was Q3 2022. I've specced Nichia 519a modules on every major project since—fabric, metal, or glass. The upfront cost is higher. But I haven't had a single callback about light quality. There's something satisfying about that.

According to USPS pricing effective January 2025, a First-Class Mail letter costs $0.73. That's cheap. A Nichia 519a LED module? About $1.75. Also cheap, in the grand scheme of things. The alternative—a re-lamping invoice—is much, much more expensive. (Prices as of this writing; verify current rates.)