It’s Bike Lights Season. Why Counting Lumens is a Waste of Your Time

Winter is swiftly approaching in the Northern Hemisphere, forcing dedicated MAMILs to seek out lights to keep their rides going, and forcing less-dedicated MAMILs to seek out the comfort of a sunset cocktail at 3PM.

This time of year brings many discussions on forums about which light is best, which has the most lumens, which lasts longest, and so forth. The posts you may read will all have one thing in common: the writers have no idea what they are talking about. None of these authors could pick a lumen out of a lineup, much less know what to do with it. This, dear reader, is why this article exists: it’s time to take the bullshit out of bike lighting. These opinions are formed because all the bike light manufacturers tell you how their lumens are more, or better, or strawberry flavored, or have more gears, or something.

Let’s start with the most bandied-about and least understood value: the lumen. What do we know about it? Is having more better? The answer is: Maybe. If we’re going to use it, we’d better know what it is, right: The answer is simple- a lumen is 1 candela per unit solid angle, or steradian.

Duh.

Everyone knows that.

If, on  the other hand, you have no idea what a steradian is, imagine a sphere with a radius of 1 unit. Now trace out an area on the inside of that sphere that is a circle with an area equal to the square of the radius, or 1 square unit. Next, if you put a light source of 1 candela at the center of the sphere, the amount of light that falls in that area is one lumen. Well, you say to yourself, (because no one in their right mind listens to you anyway) that’s obvious.

credit: https://commons.wikimedia.org/wiki/User:AndyAnderson

Of course that’s it. So all I need is a candela. How much light is that? Oddly, it’s about how much light a candle gives off. This in itself begs the question: Beeswax? Scented? Tapered? It’s based on a simple definition:  the fixed numerical value of the luminous efficacy of monochromatic radiation of frequency 540 × 1012 Hz, Kcd, to be 683 when expressed in the unit lm W−1, which is equal to cd sr W−1, or cd sr kg−1 m−2 s3, where the kilogram, metre and second are defined in terms of hc and ΔνCs.

Again, duh. That’s so obvious. I remember the Sesame Street segment on this, it was one of my favorites.

Basically, it’s about as much light as a candle made out of spermaceti (sperm whale brain stuff) made in the mid-19th century. That’s what the candela was based on, burning sperm whale head goo. If that’s not science, I don’t know what is.

Are you still with me, or are you blinded by the (bike) light? Because we’re going in deeper.

Let’s say we have 1000 lumens in our light. Technically, that’s a whole shitload of steradians, and a big pile of candelas. Obviously, that tells us how well the light performs compared to some other bike light with 1000lm. Or, actually, not. This lumen value tells us nothing whatsoever about where the light is going and what it will do when it gets there.

You could think of it this way, because everyone can imagine this: I have a gallon of water. How deep is it?

Well, if it’s in a bud vase, pretty deep. If it’s in a horse trough, less so. What we have in the marketing of bike lights is all sorts of manufacturers talking about gallons, not vases and troughs. We’re going to need another set of values, one that tells us what those lumens are doing once they get out of the light (for the time being, we will be skipping over effciency, which is how many escape vs how many get trapped inside the light fixture, or efficacy, which is how many lumens per watt we are getting out of the system).

Enter center-beam candlepower (CBCP) and beam angle. These are the values that tell us how deep the vessel that holds our gallon is. It’s also a pair of numbers you’ll never find on a bike light manufacturer’s site. You’ll find it on many non-bike lighting websites, particularly for reflector-type light bulbs, like MR-16s and PARs. In a sense, it’s a way to use an angle and an intensity to visualize the light coming from a fixture as a cone of light, and gives you a sense of how wide or narrow that cone is and how tall (intense) it is. CBCP is a simple method to use, and has been superceded somewhat, but for the general cone-shaped light pattern we’re discussing here, it’s plenty accurate.

A beam of light can have many shapes, and optical engineers will use reflectors and collimnators (lens systems) to shape that beam. It may be round, like our cone, or oval, rectangular, and so forth. In each shape of that beam, the same amount of light is in it, regardless of its form.

Imagine a flashlight that has a zoom feature- that is, you can change it from spot to flood. These have existed for decades, most people have one around the house. If you put it in spot mode and stand a given distance from a wall, you see a bunch of light in a small pool. Without moving closer, zoom it to flood. Now you have a bunch of light in a big pool, but it’s not as bright. Here’s the part that matters: the same amount of light is falling inside of each pool, but the bigger pool doesn’t look as bright because the light is spread out over the area of the pool. It’s our horse trough, while the spot is our bud vase.

Why does this matter? After all, the bike light guys are telling me I have 1000lm, so that’s enough info, right? Well, what if you wanted a flood light to illuminate the ground in front of the bike, and a spot light so you could see exactly what obstacles are in front of you? Then, you might want two 1000lm lights, but how can you tell which is the spot and which is the flood?

The answer, based on what is generally published is: you can’t. Fortunately, road.cc has done a remarkable job comparing lights using photographs and a set. This is very well done, but from a numbers perspective, something is missing: the measured amount of light on the ground. This is not to fault road.cc’s staff. Rather, it’s a point that we need to discuss because some companies publish a number of lux (lx) that their light puts out.

And this, dear reader, is where you need a good set of hip waders, because the bullshit gets real deep, real quick.

Go back up to the graphic above, where we see the unit solid angle, or steradian. See that area on the inside of the sphere? If the radius is a meter, then there is 1 lux of light (candela per square meter) in that area. If the radius is a foot, then there is 1 footcandle of light (candela per square foot) in that area. There are 10.76 square feet in a square meter, so a FC value is about one-eleventyth of a lux. Or, 55lx = 5FC. Roughly. So, keep in mind, that the higher lx value does not mean it’s brighter than the same light source in FC. This like how 39″ is not actually longer than 1 meter. And yes, it’s also one lumen per unit of area. Because you weren’t confused enough. Now, 1lx isn’t a whole lot of light, it’s about what you can measure on the ground under a full moon. This means our sphere above is a pretty dim bulb.

Still with me? Good, because it’s going to get a bit more complex. When we talk lux or FC values, we are measuring incident light, or light falling on a surface. A passable analogy is a rain gauge- it measures how much rain fell in a spot. To measure the light falling on a surface, we will use a light meter. It has a silicon chip that detects the light falling on it, and gives us a value, such as 100lx, or 48FC, or whatever.

Ok, you say, that’s cool. We have a number of lumens in our light fixture, we have designed our light fixture to send those lumens out in a direction at an angle instead of a 360-degree sphere, and we know that there is another value, CBCP that can help us predict the number that the light meter will read, so that means, finally, we can tell which of these two 1000lm lights is brighter. Huzzah!

Nope.

We also need to know the distance from the light source to the place where we put the meter. Light falls off by the square of the distance (from the source to the location of the measurement). This picture explains the concept:

source: https://commons.wikimedia.org/wiki/File:Inverse_square_law.svg

Same amount of light, spread out over a larger area. OK, so now what? I have a light, I have a light meter, I have a tape measure, so how can I tell which one is better? After all, this bike light manufacturer shows me a chart of lux, and that one says their light has 1000lm, so I should be able to tell which is a better light, right?

Again, nope.

Here’s the problem with the light meter and the values of both lx and FC: they both measure incident light, or the light falling on a surface. Our eyes don’t see that. We see reflected light, or the light that hits the surface and bounces back towards us. Sit at a dark wooden desk, with nothing on it, and look at how bright it is. Now, put a clean, white piece of paper there. Which is brighter to look at? You didn’t change the light falling on the surface, you changed the surface reflectivity. So a light that is fantastic on snow with your fatbike may be utterly useless on fresh blacktop in summer, because snow is about 99% reflective and blacktop is black because it absorbs all light. On top of that, you didn’t see the light moving through the air, like you saw the rain. You only saw the light bounce back.  So think of it this way- what if you were trying to tell how hard it was raining, but all you could see was the splashes of the rain, not the raindrops themselves. You’d really struggle to tell how hard it was raining if you were looking at grass and trying to see splashes, as opposed to looking at the splashes off pavement.

The bullshit part is that in all of the bike light pages I have found that publish a lux value, none of them state a distance from the light source. So you, as the rider, don’t know if that 50lx is 1 meter in front of your bars or 10. Also, they don’t say if the light meter is perpendicular to the light source or tilted, this will greatly affect the measured value, because the angle of incidence will change the calculation of incident light, reducing it by the cosine of the angle. Again, since you don’t know the measurement conditions, you can’t really know if the presented number is valid or pulled directly out of someone’s ass.

So the simple unit of a lux or a footcandle is this: a theoretical invisible unit of measure we cannot percieve, based on the amount of light produced by burning whale goo two centuries ago. Yeah, that’s got a ton of context in our daily lives. Here’s the whole marketing concept: someone can publish a number, based on flaming cetacean brains, and that will convince me to buy their light instead of a different one, because that’s a highly valid metric if I’ve ever seen one. Best sales pitch ever!

Ok, you say, I think I get the concept of how these parts of lighting are measured. Let’s go back to your 1000lm bike headlight. 1000lm is a lot of light, and it’s pointed ahead, and you’re riding a road or trail you know, so it’s plenty bright. And, since a lumen and a lux are pretty much the same thing, that means your 1000lm light gives me 1000lx, so that’s pretty good.

Nope.

Let’s keep in mind that if a lumen = a lux, it only does that at a unit solid angle, or, say, 1 meter from the idealized light source. These bike lights aren’t an idealized theoretical source, so don’t equate the concept of a candela per square unit and a measurable value from a physical light with an optical train that meets a specific design criteria.

Further, none of these bike light companies actually tell you clearly if you get that 1000lm at the low, medium, or high setting of the battery. We have to factor in efficacy, or lumens per watt, plus the watt-hour rating of the battery. The higher we pull power off the battery in order to get more light out, the faster we use it up. LEDs can operate in the order of 120-160lm/W, so depending on your battery and control chip design, that 1000lm output may only be good for 30-60 minutes. It may last 6 hours at 100lm, but that’s a 90% reduction in the amount of light.

Let’s come back to the beam, or shape of the light as it comes out of the fixture. More bike light manufacturers are making lights that are compliant with StVZO. This is a code, commonly used in the EU (and other areas) that was originally designed for automobile headlights. They’re not doing this because they’re super concerned about beam performance; they’re doing it so they can sell more lights in more countries. That’s not to say optical control is bad- it isn’t. As a favor, I read the StVZO code texts so you don’t have to. That’s what I do, I read codes about lighting, and sometimes I write them. I do that to save you the hassle. You’re welcome.

Here’s what StVZO does, in a nutshell. It limits the amount of light above a certain horizontal angle, in an effort to keep light from hitting an oncoming driver in the eyes. Basically, if you imagine our cone of light, pointed horizontally, we then slice the top of it off so that the top is flat and pretty much parallel to the ground. That’s about it. What it is basically saying is that the light emitted will be in a wide, flat fan instead of a cone, and generally won’t blind an oncoming observer. Some countries also have restrictions on flash rate of bike lights, but these are not spelled out in StVZO as they are specific to individual countries. It gets used by marketing people to give you the impression that these lights are fantastic, and better than the other guy’s, but in reality, it just means the light is intended to 1) not burn someone’s retinas so they run you over (assuming the light is positioned properly), and 2) be legal to sell in Yerp.

What’s never mentioned is this: there’s nothing from any of these lighting manufacturers that limits or controls how you place the light on your bars or helmet. You can point that StVZO-compliant light up at a higher angle and blind a driver, and there’s nothing the bike light manufacturer can do about it.

One thing StVZO tries to address (and really doesn’t, beyond blink rates) is daytime running lights for bikes. Most countries (and US States) mandate lights or reflectors after dark. Some allow blinking, some require it, some prohibit it. That’s why your light has modes. Trek has been instrumental in leading the push for DRLs, and in my opinion, their 200-series DRLS are fantastic and the best on the market. Again, we have a lumen count happening here, as these are 200lm lights, but they are designed very differently than a headlight you’d use for night riding on the road or trail. They are designed to have a high surface luminance, or brightness, when viewed from a distance. The upshot is that they work extremely well, so buy a set and reduce your risk of being squished.

This type of light is the same category as brake lights or stoplights: a light that you are supposed to see as a bright thing instead of a light that is designed to proeject light onto a surface. Think about it- a stoplight doesn’t project light outwards on the ground, it is bright enough to see in sunlight because it has high surface brightness (luminance). Brake lights are the same, whether newer LED-type or older filament-type. The idea is to magnify a light source over a larger surface area, making it easy to see. This type of optic is basically a complete opposite to the type of optic that is designed to throw light forward so you can see where you’re going. This is a be-seen light, not a see-with light.

So, where are we here? Can we finally say one bike light is better than another? Sadly, the answer is nope again. Basically, all the efforts we’ve now put into defining a lumen, a candela, CBCP, beam angle, lux and footcandles, and so forth, has been a waste of time, effort, and brain goo.

Well, crap, you say. I need to go ride, but I don’t know what light to get. I’m screwed. (Well, yes, sort of.) Thanks a lot, MAMIL-ass, you gave me a thousand words on why a thousand lumens is a crock. Now what do I do?

Honestly, go to a bike shop and try out the lights they have. Turn them on, point them at the wall, and try to compare them. Use the road.cc page. Ask your friends. Do a bit of research. Consider a wider beam, like an StVZO-compliant pattern, on your bars, and a spot beam on your helmet. You can do this- after all, you have a computer at work and we know you don’t use it for actual productivity. I won’t fink you out.

Keep this in mind though: if you read a lux or footcandle value on a bike light site, it’s a big brown flag waving at you. This value is useless for many reasons we won’t go into here (cosine correction, surface reflectivity, color temperature, spectral distribution, correlated color temperature, color rendition index, human visual response, perspecacity, etc). The lumen value is somewhat useful, but only if you keep in mind that is a raw number- it does not take into account the beam angle and shape, how many lumens escape the light fixture, what battery setting you’re using, and so forth. It is merely a gross starting point, not an actual metric you can put to use.

Are all bike lights useless? Not at all, but their websites may be. Let’s hope road.cc keeps their comparison page alive, and let’s hope you’ve become somewhat more educated about the light that’s in that light you want to buy.

As a side note, in the architectural and theatrical lighting worlds, we have all of these datapoints and metrics available to use as well as many more, including files we can use for computer modeling and rendering. The lights produce much better beams, have more optical systems available, and are entirely calcuable and predictable. This is fundamental data that is easy to find and use. The fact that it basically doesn’t exist in the bike lighting world is a pain in the ass. Maybe that industry doesn’t think its users aren’t sophisitcated enough to use it, or maybe since there’s no pressure from the users there’s no need to publish it. Whatever the case, not having valid data is a detriment to all cyclists who may ride after the sun sets.