The single most useful thing I ever learned from the measured-output community is this: a flashlight’s runtime isn’t a number, it’s a graph. Two lights can both claim “1,000 lumens” and behave completely differently in your hand over the next ten minutes, because what matters isn’t the starting point — it’s the shape of the curve as the light sheds output. Once you understand step-down, you stop getting fooled.
Why brightness can’t just stay high
High output makes heat. A lot of it. Push an emitter hard and it gets hot fast — too hot to hold, and hot enough to damage itself if nothing intervenes. A small flashlight has almost nowhere to dump that heat. So every light that runs high output has to deal with the heat somehow, and how it does that is the whole story.
There are two honest ways to manage it, and one dishonest one:
- Regulation + a managed step-down (good). The light holds a steady, flat output for as long as it safely can, then deliberately drops to a sustainable level once it heats up or the cell drains — a controlled step, not a collapse.
- Direct drive (cheap). The emitter is fed straight from the cell with little regulation. It starts bright as the cell is full, then sags continuously as the voltage drops. No flat line — just a slide downhill from the first second.
- The lie (the cheap “90,000 lumen” torch). Flash huge for a few seconds for the photo and the box, then fall off a cliff because nothing in it can hold that output and the tiny cell can’t feed it.
What the curves actually look like
Picture brightness on the up axis, time across the bottom.
A good regulated light draws a flat plateau — it holds, say, 800 lumens dead level for a stretch — then makes a clean step down to a sustainable level (maybe 300) and holds that flat for a long time. The graph looks like a tabletop with a step in it. You get predictable, usable light, and you can feel the step coming because it’s gentle and announced.
A direct-drive light draws a slope from the very start. Bright at second one, noticeably dimmer a few minutes later, sliding down the whole way as the cell drains. No plateau. You’re always on the downhill.
The fantasy torch draws a spike: a tall, thin peak in the first seconds, then a near-vertical drop to a dim glow it can actually sustain — which is a fraction of the headline. That spike is the number on the box. The dim glow is the light you actually own.
This is exactly why the ANSI FL-1 runtime-to-10% figure matters and a bare lumen number doesn’t. FL-1 runtime measures how long until output falls to 10% of the start. It anchors the graph to something honest. A light that quotes 1,000 lumens and a 2-hour runtime-to-10% is showing you a real tabletop. A light that quotes a giant lumen number with no runtime is hiding the cliff.
How to read a runtime graph (and a spec sheet)
When the community or a maker publishes an actual runtime graph, here’s what I look for:
- Is there a flat plateau, or just a slope? A flat regulated section means the light holds its output. A continuous slope means you’re on direct drive, always fading. I’ll take the plateau.
- How high is the sustained level after step-down? The plateau the light settles into for the long haul matters more than the peak. A light that sustains 600 lumens beats one that peaks at 2,000 and settles at 200.
- When does the step happen? A light that holds high for several minutes before stepping is genuinely useful for bursts. One that steps in 30 seconds is mostly marketing its peak.
- Where’s the 10% point? That’s your real working runtime, and it’s the FL-1 number to compare across lights.
If there’s no graph, the FL-1 lumens-plus-runtime pairing is the next best thing. Lumens alone, with no runtime, is the spec sheet hiding the curve from you.
Why the flat line wins
Here’s the practical payoff. When you’re actually using a light — under a hood, on a trail, working a job — you don’t experience the peak. You experience the level the light holds while you work. A flat, regulated 600 lumens for two hours is calm, predictable, and there when you reach the end of the job. A light that’s always sliding downhill leaves you wondering, halfway through, why it’s so much dimmer than when you turned it on.
A big starting number with a bad curve is a worse light than a modest starting number with a good one. Runtime is a graph. Buy the shape, not the spike.
If you want the lights that hold a real plateau — regulated output, honest FL-1 runtime, and the cells and chargers that feed them — the best 18650 flashlights guide is where I lay out the format that gets the curve right.