Here's the number you keep seeing on the box: 8 hours of light. You read it like a promise. Then you run the string for the first evening in the cold, and by hour five it's a half-dead, dim, flickering thing that you'd mistake for a broken one. Nobody warned you. The box didn't. The box just printed a number that is true exactly once - on a lab bench, at 20C, with a brand-new cell that will never ship in your carton.

I've been pulling these boards apart for over a decade, and the battery-runtime lie is one of the most quietly profitable things in the whole category. Not because the lights are bad. Because the number is real, and the conditions are fake.

The number is real. The conditions are a lie.

Let's do the math the way a controller-board engineer does it, because that's how the spec was made. A typical 2x AAA battery string - the kind draping your roofline or a fence - runs the controller and the LEDs off two cells in series. Total current draw is usually in the 25-40mA range once the driver, the mode logic, and the LEDs all share the rail. A fresh alkaline AAA is rated around 1000-1200mAh on the datasheet.

Do the division: 1200mAh / 30mA = 40 hours. Forty hours. So why does the box say 8?

Because the manufacturer never rates capacity at the load. Datasheet capacity is measured at a trickle, at room temperature, down to a generous cutoff. Under real continuous draw, inside a plastic housing that traps heat and then freezes, at a cutoff voltage the board actually triggers, the usable number collapses. The "8 hours" is the honest floor the supplier agreed to guarantee - the rest is your risk to absorb. They sold you the floor and called it the ceiling.

The part nobody tells you: it doesn't die, it dims into the dead

This is the piece that makes people think their string is defective when it isn't. LEDs don't switch off at the bottom of the battery's life. They dim. As a cell discharges past about 1.1V each, the rail sags, the current through each LED drops, and the string goes from bright to a grey, lifeless half-glow long before it truly dies.

So the "it died suddenly" story is wrong. What actually happened: hour four it started dimming, hour five it looked dead, hour six it was technically still putting out light you could barely see from a meter. You walk the house at night and see a dark roof and conclude the battery died. It didn't die. It was dim for two hours and you just never noticed from inside the room.

That's a controller-design choice, not a battery fault. The board's low-voltage cutoff is set high enough to protect the cells and the brand, and the LEDs below that line are simply no longer bright enough to matter. The last 25-30% of the cell's actual capacity is, for all practical purposes, wasted.

Cold is the quiet killer - and it's not a myth

Now the number you can actually trust, because it's physics: alkaline chemistry hates the cold. A cell's usable capacity is strongly temperature-dependent. Below roughly 5C, an alkaline AAA can lose a substantial fraction of its available capacity at load, and in hard freeze the string that gave you eight hours on the porch in July will hand you four - or worse, a dead board the moment it's unplugged and re-plugged, because the chemistry has gone sluggish enough to trip the cutoff almost immediately.

This is exactly why battery strings feel "flaky" outdoors and fine in a garage test. The garage test was the lie. The roof is the truth.

The fix that actually works: lithium cells for outdoor strings. Lithium holds its voltage and capacity far better at low temperature than alkaline, and because the cutoff is a voltage trip, the flatter discharge curve means the string stays bright to the end instead of dimming into the dead. If it's a rechargeable NiMH pack, expect the opposite - NiMH self-discharges 1-2% a day and sags under cold load, which is why a rechargeable set that "used to last all night" dies by Christmas week. Either top up the pack on the shelf or accept the runtime drop.

The three settings that eat your runtime without your permission

Runtime is not a fixed property of the product. It's a function of the mode you chose. Three levers:

  • Mode matters more than you think. A steady "on" mode is the most efficient. A fast-chasing or bright-twinkle mode pulses current harder and the board works harder, and the perceived brightness is higher but the runtime per cell-hour is worse. If your box claims 8 hours, check the spec line - it's almost always measured on a single slow mode, not the flashy one you actually run.
  • Connected strings multiply the draw. That 9-meter string you can daisy-chain into the neighbor's line? Each connected set adds its current to the same rail. Two strings on one battery source is roughly half the runtime of one. The "hours of light" spec assumes one string.
  • The timer and the dimmer. If your set has a dimmer, running it dim saves runtime but you've already lost the whole point of buying lights. Running it bright on a short battery is the trap most people fall into - they turn it up because it looks dark, and the dark is the battery dying.

What a spec you can actually trust looks like

When you're buying - or when a client asks why their set died - the difference between a real spec and a marketing number is the conditions. A defensible runtime statement tells you the cell chemistry, the cell count, the measured load, the temperature, the mode, and the cutoff. What you usually get is one number and a smile.

For a real installation, my rule is simple: buy the set rated at the hours you actually want, in the temperature you'll actually use it, and run one string per battery source. Want a roofline glowing all night at -5C? Don't buy an "8 hours" alkaline set - that number was never going to survive January. Buy a lithium-powered line or a mains-fed run for the permanent pieces.

That's the line I spec for the permanent roofline work - mains-fed, no runtime anxiety, and the brightness doesn't sag: this connectable warm-white icicle line. And for the places a real flame makes sense but you can't risk it - a mantel, a memorial table, a church pew - I spec the battery cells as lithium and keep the draw low so the dim-to-dead window stays wide: these battery-operated LED candles run clean because the whole point is they stay bright to the end.

The one-line takeaway

The number on the box is a floor measured in a warm room. Your roof is a cold room. Buy for the temperature, buy the chemistry that holds voltage in the cold, and stop reading "8 hours" like a promise. It was never a promise. It was a spec sheet with the fine print on the other side of the carton.