Well, here we are again. Waking up in the dark, riding home in the dark... it's lights time.
We've been busy getting some of this year's crop and the reviews will be going up through the winter. In the meantime, however, we thought we'd share our beam testing data with you so you have something to be going on with. We took all the lights that came into the office before the Big Test deadline – about 40 of them – and put them through their paces. The great thing about lights is that it isn't just subjective: you can measure the beam and take directly comparable photos of what it looks like, so we did. The results are available in the big road.cc light comparator at the bottom of the page. Rear lights are coming soon.
What did you do?
We've collected lots of beam data so you can compare and contrast the different lights. Light manufacturers use a number of different metrics to describe light output. Top of the pile right now is lumens, which is a measure of the total output of the light across the whole beam. Some cheaper lights use candlepower, candela or lux, which are measurements of the brightest part of the beam at a set distance. We've used lux here, but measured at a number of points across the width of the beam. That gives an indication of the brightness of the beam at the centre, the amount of peripheral light and the throw of the beam. Specifically, we measured the lux value of the beam at two metres distance, in 10cm increments from the centre of the beam to 1m from the centre, giving eleven readings.
This year we've also included data on the shape of the beam. Putting the brightest part of the beam at the centre, we measured the output at thirty-degree increments around the beam, at a distance of 50cm from the centre. That gives you a good idea of the pattern of the beam; most are more or less round, but some have more interesting characteristics.
To get a good idea of what the beam looks like, we set up a bike on a rig so that we could photograph the beams of all the different lights in a comparable way. We used a tunnel this year, because it has the advantage of always being dry and pitch black down there which means that we should be able to more easily add to this test when more lights come in; last year we found with outdoor shots that replicating the rig wasn't easy, as different conditions above gorund mean differing levels of ambient light and reflection from surfaces, even in the same spot. Each of the beam shots you can see above was taken using the same settings on the camera. Specifically, they're all shot from directly above the saddle, using a 28mm lens on a Canon EOS1100D (effective 45mm), shooting for 2s at f22 on ISO3200. If you fancy doing some of your own. So as much as they can be, they're directly comparable to one another. If one looks brighter than another, that's because it was. The two reflective jackets are at a distance of 15m and 30m, respectively, from the light. The reflective strips down the centre are at 2.8m intervals.
Is that it, then?
No, of course not. You can look through out Buyer's Guide for more information on what kind of lights will suit your riding. A super-bright beam isn't much use if the light ends up in a hedge after the first pothole, or fizzles out when it starts raining. We'll be subjecting all the lights to the rigours of the road.cc testing process and when we're happy that we've thrashed them they'll each get a full review. We'll include the comparison tool in each review too. In the meantime, we thought you'd like to see how they fared.
A word about logs
The graph displaying the beam data uses a logarithmic scale to display the output of the lights. If you understand or care about such things, here's why:
Firstly, light beams follow an inverse square law regarding the strength of the light at increasing distance, because they're illuminating a two-dimensional plane. So at twice the distance, the light beam is spread over four times the area. Consequently, a light that is measured as twice as bright at its centre won't let you see twice as far. The logarithmic scale produces a more realistic comparison because of this.
Secondly, the variations in the amount of peripheral light, though much smaller than the variations in the centre, make a big difference to how much peripheral vision you get. The logarithmic scale amplifies these differences relative to the centre of the beam, so it's easier to see which unit is putting out more light at the sides.
A word about the non-circular beam patterns
Some of the lights on test don't have a uniform circular beam pattern, with more light along the centre of the beam. Because of this, the beam values on the long graph are a bit inflated because there's more light concentrated in the axis we're measuring, and less illuminating the tree canopy. It doesn't skew the data hugely though, and the beam graph in conjunction with the beam shot and beam shape should give you the whole story.
The comparator is below. Have fun!
If you have a nice big screen you can click here for the widescreen version (1400x1000px)
Add new comment
52 comments
Great comparative test - sets a new standard - please add the full exposure range though
Dave,
Thanks. That also explains why other "furren" but popular lights like Ay-Up are absent.
Nice coding to get the slick horizontal comparison!
It's called the Electron micro 5 and it retails at a certain bike shop chain starting with E, for £17.99
nice paying of $8 to get someone else's code, more like
we think we might just have a faulty one, lezyne are sending us another for comparison.
it's possible that we had it on the wrong setting but the beam shot and the beam graph are both lower than the power drive, and they were done separately, it's unlikely that we got it wrong both times on that one light.
certainly the super drive is *meant* to be brighter so we'll see what gives when we get another sample
I've just recently bought the Magicshine. It's an awesome light but I was worried that it would blind oncoming motorists so I've added a bit of reflective plastic as a cowling to limit the beam a bit. This makes me much more comfortable using it as the last thing i want is for drivers to not see me because I've blinded them
Extremely thorough. Brilliant, even.
Funny thing Paul, if the driver can't see they are meant to slow down.
that's what i ended up doing with mine last year too.
Thanks for this. Really a very helpful test!
I've got a couple of the Ebay/DealExtreme Magicshine clones - one with a smooth mirror and one with light orange-peel effect mirror. I have one of them on the bar, one on a head mount (when off road only).
They are completely and totally amazing. Amazing enough that in the forest last night - under trees in the pitch dark - I was still bombing along at over 20mph in places although it was a bit muddy and I was taking it easy.
The smooth mirror is a bit better on road, it 'throws' better, but off-road the orange peel one is brilliant as it gives a smoother, wider field. £30 or so each and if one breaks I'll buy another - I can't imagine why anyone buys these £200-plus ones!
Well done Road.CC it's a great test and should be rolled out every year or a permanent web feature. As a reward all Road.CC staff may have Christmas day off.
And forcing oncoming traffic to slow down by shining a blinding light in their eyes is acceptable how?
Yes shame to miss this as it is meant to be one of the few lights to be designed for the road more like a car or motorcycle light
BUT
great article and presentation! very useful.
I am a Sustrans Ranger and a few of us have this light . It is easily as good if not better than those 2 shown.
This is fantastic! Many thanks for the extremely hard work that must have gone into developing this. This must be the world's only bible for bike light choice.
Thoughts for further tests:
1) Would be awesome if possible to add some lights from previous years to act as a useful benchmark for those of us buying new lights (appreciate that this will depend largely on what you have lying around at any point in time, but that may be better than nothing)
2) Agree about the photo of the light from in front of the bike so you can see how much you will blind/be invisible to oncoming traffic
Amazing stuff. Hope this lives and grows for many years!
We did test the Trelock 950 last year, which is very similar in concept. Awesome light.
Agreed!
Although the bracket on the dyno version is a piece of s**t, the light itself is excellent, good switch, nice wide beam, good for seeing both verges on dark country lanes, and enough throw for normal speeds.
As a back-up I use one of the silly bright chinese "Cree" torches which has a narrow beam but enough throw for full speed descents. Makes a brilliant combination and for less money total than, say, a Strada.
Excellent. I've got the Super Drive (ordered it two days before the XL's were announced - arggggh!) and I'm very happy with it. Possibly not quite as bright as my 3yo Magic Shine, but much more practical, so I actually use it.
Seems unlikely then - at least on my version the brightest setting is the first setting in the cycle.
Of course now every road.cc-reading Lezyne owner is going to be worrying that they got a dud too - me included!
Where's the latest version of the Hope Vision One? As far as I'm aware this is still a very popular light choice, and has been in at least one previous roadcc test. Really disappointed that I can't compare the Lezyne models with the updated Hope Vision One.
OK, you guys did a nice job with the photos and the side by side stuff, but you completely missed the boat in your comment "Some cheaper lights use candlepower, candela or lux, which are measurements of the brightest part of the beam at a set distance." You correctly state that light falls off per the square of the distance (Inverse Square Law), but what you missed is that the candela value is the number that is plugged into the ISL to determine how much light is on a surface at a given distance.
Lumens are not an accurate way to compare projected light. It's like this: I have a gallon of water, how deep is it? You can't say unless you know how big your bucket is. If these were LEDs that were to be used in table lamps (spreading light in all directions with no optics), lumen count is enough to compare them side-by-side.
What very few LED bike light manufacturers do is publish candela or candlepower values, which can be used to calculate footcandles (lumens/sq ft) or lux (per square meter) at a given distance. These numbers are really the only value that should be used, and we need to demand that manufacturers publish them. Instead, unfortunately, you discounted them in your introduction.
Think of candela or candlepower like water pressure from your garden hose. If you can set the spray to wide fan or narrow stream, the flow rate has not changed but the distribution has. A spot and a flood may have the same lumen package but will distribute their light very differently.
Unfortunately, by discounting this value, your article supports the manufacturers that publish only initial (not actual) lumens, and does not increase pressure on them to publish photometric data that they have.
+1 on the lack of hope lights. I thought Exposure might have been included too. Fantastic test all the same. Any idea when the rear light test will be with us? I'm poised to buy some better rear end illumination!
Pages