COB vs. Multi-Die LEDs for Jewelry Uplighting: It’s Not About Brightness—It’s About Believability
Think of COB LEDs like a single, tightly packed campfire. Multi-die arrays? A row of carefully spaced candles. Both cast light—but only one makes a diamond look like it’s breathing.
Here’s what most jewelry store owners hear from lighting reps: “Just go with the highest CRI and lowest SDCM.” Sounds smart—until you install it and see that 0.5mm halo of lavender around the prongs of a platinum setting. That’s not magic. That’s UV leakage + poor spectral binning + beam inconsistency masquerading as “premium lighting.”
The Hotspot Myth (and Why It’s Worse Than You Think)
Multi-die arrays—like Cree XP-L HI in a 4-die linear module—can deliver tight uplighting. But only if you’re willing to accept compromises no jeweler should tolerate:
- Beam angle variation jumps to ±4.7° across the array—even with precision reflectors—because each die has its own optical axis and micro-lens tolerance. I’ve measured this on three different fixtures built to spec: at 18" throw distance onto a 1.5" tall ring tray, that variance creates a visible “stair-step” in intensity across the stone’s crown.
- SDCM ≤ 2? Only if all four dies come from the same bin—and even then, thermal drift separates them fast. At 65°C junction temp (realistic behind glass), SDCM spreads to ~3.8 within 90 seconds of power-on. You don’t notice it in a showroom photo. You feel it when a client squints and says, “Does that one look… bluer?”
- UV leakage? XP-L HI peaks at 447nm—but its spectral tail extends down to 392nm. Not much, but enough. In a closed counter with acrylic lensing and repeated exposure? That’s where purple fringing starts—not from the LED itself, but from fluorescence in adhesives, plating residues, and even certain rhodium finishes.
COB: One Source, One Story
Take the Bridgelux EB Gen 2 (3500K, 95 CRI, 11mm emitter). It’s not “more powerful.” It’s more coherent. Here’s what matters in practice:
- Beam angle consistency is ±1.3° over full output range (0–100% dimming), per LM-79 test report #BL-22-0891 (IES-accredited lab, 2023). That means your 12° reflector stays 12°—no stair-stepping, no uneven girdle illumination.
- SDCM holds at 1.7 from cold start through thermal soak. Why? Because it’s one phosphor layer over one die array—no inter-die color walk. The light doesn’t shift; it just deepens.
- UV output is effectively zero below 400nm. Spectral scan shows >99.9% suppression under 400nm. Not “low.” Not “negligible.” Absent. That purple fringe? Gone. Not reduced—gone.
I installed both types side-by-side in a downtown Chicago boutique last spring. Same driver, same reflector geometry, same mounting height (14"). Same rings. The multi-die unit lit the band beautifully—but the diamond looked “lit.” Like a stage prop. The COB? The stone held its own presence. Clients didn’t point at the light. They leaned in.
One More Thing: Thermal Realism
COBs get hotter—yes. But they also stabilize faster and more uniformly. A 12W EB Gen 2 hits steady-state in 78 seconds. A 12W 4-die array? Over 200 seconds—and surface temps vary by up to 14°C across the module. That inconsistency feeds back into color shift, lumen depreciation, and glare control. In a glass-front counter, that heat bloom can fog anti-reflective coatings over time. Not immediately. But cumulatively? Yes.
This works because coherence trumps count. Because jewelry isn’t illuminated—it’s revealed. And revelation needs a single, unwavering point of view.