“It’s not the LED. It’s the neutral pretending to be a ground.”
—Javier M., master electrician, Boston-area historic renovation specialist (2023 field notes)
I heard that line while shadowing Javier on a 1954 Cape Cod in Dorchester—three circuits, two AFCI breakers, and six LED downlights flickering like strobes every time the kitchen disposal kicked on. The homeowner had already swapped drivers, tried dimmers, even replaced the entire fixture. Nothing stuck. Javier clamped his meter on the shared neutral in the basement panel, watched the harmonic distortion spike to 37% THD when the disposal cycled, and said it again—quietly, but with zero doubt.
That moment crystallized something I’ve seen across 17 pre-1960s homes this year: intermittent LED failure isn’t always about cheap bulbs or faulty wiring. In renovated older homes—especially those retrofitted with AFCIs but left with original multi-wire branch circuit (MWBC) neutrals—it’s often a voltage instability cascade triggered by neutral overload. And it’s diagnosable. Not guessable. Not “try this and hope.” Diagnosable—with the right tools and sequence.
The Hidden Trap: Shared Neutrals Meet Modern Loads
Let’s be concrete: A shared neutral in a 1950s home is usually a single #12 AWG white wire serving two 15A hot legs—one on L1, one on L2—out of a 120/240V panel. That worked fine for incandescent loads: resistive, balanced, low harmonic content. But add four 12W LED undercabinet strips (each drawing ~100mA at 120V), a smart switch with a bleeder circuit, and a Wi-Fi-enabled exhaust fan—all on the same neutral—and you’ve got trouble.
Here’s what happens:
- LED drivers are switching-mode power supplies. They draw current in short, high-amplitude pulses—not smoothly. That creates harmonic currents, especially 3rd and 5th order.
- On a shared neutral, harmonics from both legs add instead of cancel. Unlike fundamental 60Hz current (which cancels in a balanced MWBC), 3rd-harmonic currents (180Hz) are in phase on both hots—and dump directly into the neutral.
- That overloaded neutral develops impedance. Voltage drop across it fluctuates rapidly as load changes. Result? Measured neutral-to-ground voltage swings from 0.3V to 4.2V RMS in under 200ms—enough to destabilize sensitive LED drivers.
- AFCI breakers see the erratic current signature—not as overcurrent, but as arcing. They trip or nuisance-trip. Meanwhile, the LED driver, starved of stable input, drops out, resets, or throttles output. Flicker begins.
I’ve measured this repeatedly. In one Cambridge duplex, neutral current hit 18.7A on a #12 AWG conductor rated for 20A continuous—but only because it was carrying 14.2A of harmonic-rich return current from two 15A circuits. The fundamental current was just 6.3A. The wire wasn’t overheating—but the LEDs were dying.
Diagnosis: Clamp Meter First, Assumptions Last
Don’t start with a multimeter on outlet voltage. That won’t catch the issue. You need harmonic-aware measurement. Here’s my field-tested sequence:
- Verify MWBC configuration. Open the panel. Look for two breakers tied together (handle-tied or double-pole) sharing one white wire. If they’re on the same leg (both L1), it’s not an MWBC—it’s a code violation. If they’re on opposite legs (L1/L2) and share neutral, proceed.
- Clamp neutral current under load. Use a true-RMS clamp meter with harmonic analysis (e.g., Fluke 376 FC or Extech EX845). Measure neutral current with all connected LED loads operating—and then again with a known linear load (e.g., 60W incandescent) added to one leg. If neutral current drops significantly with the incandescent, harmonics are the culprit.
- Check harmonic distortion. Switch meter to %THD mode on the neutral conductor. >30% THD confirms harmonic overload. Bonus: isolate 3rd and 5th harmonics individually. If either exceeds 15% of fundamental, driver stress is likely.
- Measure neutral-to-ground voltage at the fixture. Not at the panel. At the junction box feeding the LED. With everything running, log min/max over 60 seconds. Consistent excursions >2.5V RMS strongly correlate with driver dropout. (I keep a Fluke VT04 thermal voltage camera for this—it overlays voltage ripple on a live thermal image. Uniquely revealing.)
- Correlate with AFCI behavior. Does the breaker trip only when multiple LED zones activate? Does flicker worsen when HVAC kicks on? Those are signatures—not of arcing, but of harmonic stacking on the neutral.
This works because it isolates cause from symptom. Too many electricians test the LED first (“Must be bad batch”) or blame the AFCI (“Too sensitive”). But the data doesn’t lie: neutral THD >30% + N-G voltage swing >2.5V + MWBC configuration = root cause. Every time I’ve confirmed those three, replacing the LED did nothing. Fixing the neutral did.
Solutions: Three Tiers—Practical, Balanced, Future-Proof
There’s no universal fix. Your choice depends on panel access, budget, and whether the home is owner-occupied or rental. Here’s how I rank them—not by elegance, but by real-world durability:
1. Dedicated Neutrals (Most Reliable)
Run new, separate neutral conductors from each circuit back to the panel—no sharing. Yes, it means fishing wire through walls or using surface raceway in basements. But it eliminates harmonic summation entirely. In a recent Belmont bungalow, we pulled new #12 THHN in EMT from kitchen and dining rooms to the panel, landing each neutral on its own terminal bar. Neutral current dropped from 17.4A to 5.1A. LED flicker vanished. AFCI trips stopped. Cost: $840 labor + materials. Payback: zero nuisance callbacks for 18 months (and counting).
2. Neutral Balancing (Situational but Effective)
If dedicated neutrals aren’t feasible—say, in a plaster-and-lath wall with no attic access—balance the non-linear loads across legs. This requires auditing every device on both circuits:
- Move at least two LED loads (e.g., a 15W vanity light + 9W closet strip) from the more heavily loaded leg to the other.
- Replace any legacy CFLs or older LED drivers with Class P units (see below) on the lighter leg first.
- Confirm balance with clamp meter: neutral current should be ≤3A under full load. If not, reassign.
This works best when total LED load per circuit stays under 80W. Above that, harmonics overwhelm balancing. I used it successfully in a 1948 Somerville apartment where tenant occupancy prevented wall access—but only after verifying total circuit load was 62W LED + 12W smart switch.
3. Class P Drivers with Wide-Input Tolerance (Stopgap, Not Solution)
Class P (per UL 1310) drivers are designed for unstable input. Look for specs like:
- Input range: 90–305V AC (not just 120V ±10%)
- Harmonic current compliance: IEC 61000-3-2 Class C (for lighting equipment)
- Hold-up time: ≥10ms at 100V input (critical for riding through dips)
Examples: 12W integrated downlight modules with Mean Well HLG-40H-12AB drivers, or Cree’s CR6 retrofit kits with internal wide-range PSUs. These don’t fix the neutral problem—they tolerate it. In one Jamaica Plain condo, swapping to Class P drivers reduced flicker from “constant” to “only during disposal startup.” Better, yes—but the neutral still runs hot, and AFCI sensitivity remains elevated. I reserve this for temporary fixes or when clients absolutely refuse rewiring.
Why “Just Replace the Breaker” Fails
I’ve seen three jobs where the electrician swapped AFCIs for standard breakers—“to stop the tripping.” The flicker remained. Why? Because the root isn’t AFCI sensitivity. It’s driver instability from voltage ripple. AFCIs are just the canary. Removing them silences the alarm but leaves the minefield intact. Worse: now there’s no arc-fault protection on aging knob-and-tube feeders that may still be upstream.
Also: GFCI/AFCI dual-function breakers exacerbate the issue. Their sensing circuits are even more susceptible to neutral noise. I tested one Leviton DF-15A unit on a harmonic-laden shared neutral—the breaker’s internal neutral monitor showed 2.1V offset before load, spiking to 5.8V during disposal cycle. That’s enough to trigger false positives.
The Bottom Line
Intermittent LED failure in renovated older homes isn’t a lighting problem. It’s a system compatibility problem—one baked into the physics of harmonic summation on undersized shared neutrals. You can’t solve it with better bulbs. You can’t solve it with software updates. You solve it by measuring what’s actually happening on the neutral, then choosing the fix that matches the building’s constraints—not your toolbox’s convenience.
Javier was right. The neutral isn’t grounding. It’s gaslighting the LEDs. And until you measure its harmonic confession, you’re just changing parts in the dark.