|[ Main index » Bicycle components tests » (Dynamo) bicycle lighting analysis » Taillamps » Theory for taillamps||]|
At the moment this is additional to the section on what's good and bad in taillamps, on the main page. Perhaps I will move all that stuff from the main page here, and put a summary on the main page instead...
From the testing I did it is apparent that for long distance visibility the taillamp needs an almost-collimated beam. But you need more, for closer distance you want more light in various directions to the side, so that people clearly see you when they are passing. And sideways visibility is good for intersections, to warn other traffic that bicycles are in the area.
So in my view an almost-collimated beam (such as from the Spanninga 15 with incandescent bulb) together with a larger diffusely lit up surface (similar to the Spanninga Plateo or the Niteflux) would be the perfect taillamp.
Flashing is not good for several reasons, to best judge a cyclist's (change of) distance and to have an indication a cylist is turning, a steady taillamp is required.
In the Eurobike award advertisement page Philips says about their taillamp:
This could be close to what I think a taillamp should be like, so I'm looking forward to testing it... Update: I've tested the Philips taillamp and it's not quite as I expected above, but it does do what I want in a taillamp very well.
Flashing taillamps (light going on and off) are only good for one thing: To save battery life if they're battery powered. They make it very hard, almost impossible really, to estimate distance... Or rather, the change in distance, from which you can then also estimate distance. To see change in distance you need a continuous light coming from an illuminated area of reasonable size.
You could also consider a taillamp whose intensity goes from low to high and back to be flashing, but it would be better to describe it as having a cyclically variable intensity. This is probably better than true on/off flashing, but I'd have to do some tests to be certain of how well/badly this works. A problem with this is that to be really noticeable the high output must be similar to a car's brakelamp intensity, which again means being annoying in particular towards other cyclists.
Another option is a flash with a very short off time, say 1s on and 0.1s off. This would give more attention while possibly still allowing estimation of change of distance. I need to test how well/badly this works.
One consideration always remains: You do not need to be brighter than a car taillamp, nor have a flashing taillamp, to be seen/noticed!
A possible reason to indicate a bicycle is to indicate a speed differential on roads where both cars and bicycle are allowed ride, and where those cars may ride more than say 60 km/h. The reason I take 60 km/h is that in NL on shared roads cars may only ride 60 km/h, and the speed differential of slow bicycles (most people ride at 15 km/h) with cars going 60 km/h is not a problem here.
What was suggested to me is the following theory of how a very bright taillamp with light in a cone such as the one from Ktronik or taillamps from Dinotte can be useful in situations where drivers seem to have a hatred towards cyclists as in Australia (although whether it really helps toward attitude is questionable):
Theory (not mine): The light is pointed downwards which should make it impossible to claim that the cyclist was not visible (because a large section of the road in now reddishly lit up), and this reddish patch on the ground should serve as a passing guide the overtaking traffic.
I can see a problem with this, namely long distance visibility: If this is good then the lamp will be too bright at short distance because the light extends in a cone. Unless the cone is fully aimed down and long distance visibility can be achieved with the light reflected off of the road surface. Would that work? Possibly, I see this with headlamps from cars on some roads where I get the impression a moped is coming but it's reflected light off the road from the road parallel to the one I'm riding on. Here is a picture I made which shows such reflections off of the road from a moped and a car:
It would mean a lot of wasted light though, the same visibility could be achieved with far less power using an almost-collimated beam, so this way to achieve visibility is bad for dynamo driven taillamps.
This patch of red light on the ground as a passing guide is an interesting idea, and one I'm going to test... I will also test how well visible a cyclist is from the light bouncing off of the road surface.
I got a very interesting suggestion from someone in Rumania, this may actually work: A large illuminated taillamp with a flashing bit in that lamp. This should keep it possible to estimate distance, but the brightness may not become too high or it would mean there would still be a on/off effect, so brightness levels need to be set precisely, and adjusted to that of the brightness of the main light. although for places with many bicycles this is again suboptimal to say the least.
Here are some experiences of the laser light-lane (new URL: here). I had already seen it long ago (sometime in 2009, see e.g.: road.cc jan 2009), and it seems to be finally available in 2012. I don't believe it's a good idea as it means expecting drivers to look at the ground all the time. With the Dinotte tail-flood you would get bounce from the road from long distance as I've experienced from headlamps of cars, and short distance light is directly sent to your eyes as well from the part going up. With the laser light you probably will not get this long distance effect and for short distance it would mean drivers in cars need to look down. But perhaps the latter just works automatically.
So I think drivers in cars (need to!) keep a slightly higher view, just like I (need to) do on a bicycle. However, I would be interested in trying it out. I suspect it's more gimmick than useful, but I need to test it to make sure...
Another thing is that if it works, then I suspect it will only really work in areas with few bicycles.
But I have other reservations. When I first read about this I already thought about some possible problems:
and recently I thought of another 'what if' scenario:
Getting direct or almost direct light from a laser into your eyes is with all but very low power lasers a very bad thing, so considering what can go wrong, I can only consider even using something like this (if the concept works well) in case it is not dangerous to look into the light from this device, even at full power.
First a bit about the technical side: Brake lamps are easy to make on a bicycle. One way is to use the dynamo pulses: If the pulses slow down enough, a capacitor with extra stored energy (possibly the same one used for standlight) could be used to make the taillamp light up brighter. For this to work the dynamo must provide pulses that a microcontroller can count, so this works with a dynamo hub, roller dynamo or sidewall dynamo, but not with the Sunup generator. This is the way B&M's Braketec works. Another trivial way to implement a brake lamp is with an accelerometer. These devices are even simpler than gyroscopes which have also been available for a long time in miniature form which makes it possible to install these in a taillamp. When acceleration is sufficiently negative, the lamp can light up extra bright, again using for example a capacitor. Note that if you stop pedalling you will slow down which is a deceleration, but very gradually, probably not a good idea to make the brake lamp go on then, because it means brake lamps would go on all the time then! So you need a threshold value, this goes for the brakelamp using dynamo pulses as well.
Now the practical side: I've thought about the use of a brakelamp on a bicycle, and to be frank it seems to me rather pointless:
In 20 years of riding fast on road-type bikes through cities (and by fast I mean 30-40 km/h where possible, though the last few years I'm taking it a bit easier, usually 30km/h max) I have only once had a dangerous situation where a guy in front of me braked hard without a need, namely to go almost immediately to the left into a side street, perhaps he was trying out his MTB skills. This was late at night in the dark, but I could see what happened very well, I just braked and evaded. A brake lamp would not have helped.
As I wrote some time ago in emails (See below for example) a shape such as the Herrmans H-track or Philips Lumiring could be used to differentiate a bicycle from faster vehicles (car, truck or moped, etc.)
I spotted an example of a circular ring taillamp in a motorvehicle from quite long ago, but it is probably an exception. See this picture from the 1966 TV series Mission impossible (season 1, episode 27, at 15:14):
I guess it's actually more probable that the ring is a retro-reflector (in that case there must be a source of light coming from the direction of the filmcamera). A later scene shows the brake light which seems to be only the center section. I further heard that some recent cars use a ring of light with the brake lamp in the middle...
Some more searching and I found for example the Nissan 180SX has circular taillamps and either the originals or aftermarket ones from 2008 have the outer ring lighting up. Ditto for other brands such as shown here VW Jetta, 2008 so this type of design (as in the Hermanns and Philips Lumiring) is not new and gaining popularity as well, so probably not enough to differentiate bicycle from car/truck. Using 2 taillamps, one above the other, is possibly the best method yet. I received a suggestion of having a taillamp with flashing section in it, this is interesting and might work (I still don't think flashing is acceptable in areas with lots of bicycles).
As with reflectors mandated by StVZO, it could be good to have a second taillamp and as you always have available height as opposed to width (then you need e.g. panniers), and also because placing 2 taillamps vertically instead of horizontally makes them stand out in traffic as 'not a car/truck/moped' this could work well. It should also work better in judging distance compared to a single taillamp. I will do some experiments with this...
This is a problem not encountered in NL, but e.g. in Taiwan where there are lots of scooters with car-like taillamps where a standard taillamp such as the B&M Toplight line plus does not stand out and apparently barely gets noticed. In this case the person who told me about it says he wouldn't feel safe with a line plus, but a flashing taillamp stands out as 'a bicycle'. Also, though flashing distracts, there are so many strong distracting lights on the road that it is just some added noise. You do keep the problem of having no or very much diminished ability to estimate distance, but this might not be needed if cars just keep to one's left (or right in countries where they drive on the left) to overtake. Another comment made was that if people see a very strong (blinding, cone shaped) taillamp and/or flashing taillamp, the motorists and scooter riders can't accurately estimate your position on the road and they are more likely to go around it in a wide way (assuming they don't hit you from behind due to not being able to estimate distance in case of a flashing taillamp!).
Oh, I came an interesting page about driving safety (or unsafety :) ) in Taiwan, then I saw a link a another page as not everyone agrees with him, ah yes, that reminds of criticism of my lighting pages by 'experts' in newsgroups and forums ;-)
Anyway, if there are lots of bicycles, then I think flashing is not acceptable, and even if there are few bicycles, wouldn't a car-like taillamp be enough then? So what is needed is something I wanted for a long time, a fully lit up taillamp like the Plateo, perhaps even slightly bigger, and brighter as in a car. But perhaps on a bike, in situation like in Taiwan where traffic is chaotic and where many people (on scooters, motorists too?) don't really obey a lot of the traffic rules (such behaviour is extremely uncommon in NL) a flashing light might be the only way to 'get the scooters off your back' so to speak. Comments?
As I state in several places, brighter than a car's taillamp is annoying, but more light (without getting annoying!) is possible by having a larger illuminated surface instead of a higher light intensity. Having a large illuminated area is the way bicycle taillamps should be developed. Another thing to note is that even bicycles with many extremely bright taillamps get hit by motorists who claim they did not see such cyclists, just like ambulances with all lights get rammed by people claiming not to have seen them. The only way they couldn't have seen them is by being blind or looking down! Actually, claims by people 'I didn't see it' are more a copout to act as if they had no fault in causing the accident, such as making an incorrect estimate of distance to cyclicsts a motorist passed before turning right and cutting them off. Such people should get their licence taken away... What such examples show is very important: If someone is not paying attention, then it doesn't matter how much light your lamps put out, and flashing won't help either! If someone is paying attention you only need a very weak taillamp such as old incandescent taillamps in NL. These have always been considered good enough and not dangerous and there are a LOT of cyclists in NL. If such taillamps were not good enough something would have been done about them.
Here are various thoughts about taillamps which I will order into a cohesive whole later, but I want to share them already so here they are in almost raw form, with slight edits to provide more clarity and some extra information...
Someone questioned my views saying they are too much from the point of view of cyclists, and that flashing is good to differentiate cyclists from anything else including on the pavement. First this section about 'the need to know that a light belongs to a bicycle', my response is:
Why do you need to know it's a light that's not on the pavement? You can determine this by noticing the light is not straight ahead. But also, I have NEVER seen a red/white lamp on the pavement. Some pedestrians use LED armlamps when they are walking on bike lanes or on the public road, but not on pavement. So when do you get this in Poland?
Also note, suppose a guy with bicycle has a flashing taillamp. He stops riding and walks on the pavement with it, but doesn't switch off his flashing taillamp for example because he wants to get back on the road (or he just forgets to switch it off). Then you think he is on the road. So your suggestion that you need to distinguish lamps being on the road or pavement from the type of light used (flashing/non-flashing) does not work.
And to 'From the point of view of a driver in a car, a bicycle lamp should be strong and flashing' my response is:
I disagree that it needs to be flashing, and as to strength: There is a reason why there are limits to taillamp brightness in cars. Car-taillamps when braking are too bright for normal use, and this is known. I.e. if a car taillamp in normal use would be as bright as it is when braking, then it would be very very annoying. Annoying means distracting, which means less attention of other road users goes to everything else that's going on on the street, more is going towards this feeling of 'Goddam! That light is annoying'. This is how the brain works...
I've talked with someone else about motorcycles, scooters and behaviour of motorists (in cars). Motorists do not really make the distinction between cars and motorcycles/scooters in the city, because they can drive about as fast. However, motorcycles and scooters are much more dangerous for the riders, also, they cannot brake as hard. This means drivers in cars should take note of that but this doesn't really work.
It does work for bicycles, and I suppose one of the reasons is:
- There are a lot of them
- The speed difference is so great that that immediately gets them classified in the brain of the motorist as something to be careful about.
There are other psychological factors which could help, for example the fact that bicycles are used by children a lot (in NL).
My sister suggested that having more categories in one's head than a motorist seems to have may not work. It would mean having to think:
- This is a car, same as me.
- This is a bicycle, need to be careful going around it.
- This is a motorcycle, need to be careful as the driver is very exposed in case of an accident, cannot brake as fast as my car.
- This is a scooter, need to be careful as the driver is very exposed in case of an accident, cannot brake as fast as my car.
But motorists seem to have just 2 categories: Car-like, bicycle.
So things can get dangerous for motorcycles and scooters, but people on motorcycles and scooters all tend to get on the first row before a traffic light, this is a natural way for them to get ahead of traffic and avoid possibly dangerous situations as much as they can.
Pedestrians are a separate case as they almost never walk on the same road as motorists, they will only interact with motorists on crossings. What counts here is the attitude of the motorist in letting pedestrians go on zebra crossings, where they have the right of way (in NL).
Speed difference: Here in NL bicycles are on shared roads only where cars may drive 60 km/h. Other roads where you can ride 80 km/h or more are not shared. This means a speed difference with slow cyclists of 40-45 km/h. That's not so much as to be dangerous even with little light.
I think all in all the most important thing is driver education. I also believe as I said that letting children go to school on bicycles is an important factor in forcing drivers to take note. Note that here in NL drivers really look out for cyclists, they see what a cyclist does, they anticipate, etc. I suppose some practice is also part of it.
I suppose a way to distinguish bicycles from cars could be useful, but then, the little light is already a good indication. Reflectors in pedals are also incredibly useful. Since getting an Edelux I notice these with my bike from about 100 m already, so in cars this also helps a lot at night.
To that being 'annoying to drivers in cars isn't so bad, it only lasts a short while' my response is:
When I'm talking about lights being annoying, it's for these reasons:
1. There are a LOT of bicycles in the Netherlands. I get annoyed by these crappy taillamps all the time. Without good reason.
2. Cyclists are the ones who closely follow closely other cyclists, and strong point-sources of light are really bad for them, more so than for drivers in cars.
3. I think it's time that manufacturers learn the facts. The tests I've done clearly show that fully diffuse (non-cone) or taillamps with almost-collimated beam (most of the light is in an almost parallel bundle straight to the rear) are best for being seen, and also, they are not annoying. Making the beam annoying and in a cone is counterproductive. It's dangerous on short distance because it's blinding, and from large distance it's not as effective as an almost-collimated beam. The best is an almost-collimated beam with diffuse light added for more visibility when a car gets closer and goes a bit to the side to overtake.
About the need to know that a taillamp belongs to a bicycle:
The only situation where I can see being able to distinguish between bicycle and other vehicle would be useful is on roads where cars ride 80 km/h or more and where bicycles are allowed. In NL bicycles are only allowed on roads which are 60 km/h or less for cars. Riding on the edge of 'highways' as I've seen in documentaries about cycling in the US, just doesn't happen here. This is when I thought about the lamp design of Herrmans and the new one from Philips. These rings of light could be useful for that purpose, for those countries where bicycles can ride on such roads, perhaps even useful enough to make it a regulation, because with that ring of light these taillamps are different from any other light source on the road.
The more stimuli, the longer the reaction time of the brain. This means DRL can cause and probably will cause a degradation of safety at some point (when more vehicles use it). When everyone uses it there is either going to be a situation of being back to the start (light has no addition over no light as before) or of being worse due to too many distracting stimuli. I'm guessing it will be worse than with no lights. Lots of flashing lights will also have this effect. Cats, children on the sidewalk will all get overlooked. And no, it's not good to think "It's just a cat, let it get run over" because if that cat runs on the street and people respond too slowly (e.g. didn't notice the cat on the sidewalk, nor that it might go on the street) they can do silly stuff such a swerve into oncoming traffic. Ditto for children, noticing too late means a too late response, and then a wrong response which can have worse consequences.
DRL is a form of arms race. I'm not sure where I said it, perhaps candlepower forums (but it lost a lot of posts from end of 2010 to early 2011, so might be gone) in which I said this and also: "Nothing good comes from an arms race". I choose not to take part in that...
My views on headlamps and taillamps come from rational thinking about what's needed. This always means thinking ahead to what will happen if everyone has the same lights. What if all cyclists here in NL had flashing lights? What if they all had those nasty taillamps with point source? Note that there are a lot of cyclists in NL...
I prefer to think everyone is rational and not stupid and think it will be ok :) I know this is not the case, nor do I think everything will be ok when doing that. But my choices are based on the idea that people are rational, and that things will be ok. The reason is that I don't want to design stuff because of bad temporary situations, nor that people don't respect the rules. It's useless doing any sort of analysis on road safety for bicycles for India for example for that reason: Drivers are inept there. All of them. Education has to come first. This is also the problem with UK, Australia. So there you can say "jungle law rules". It still doesn't make it ok to consider the situation there and behaviour of those people normal. I can see myself selling or designing lights for such environments, but then I would explicitly say "This is for situations where other drivers are inept, crazy, antisocial, or all of that together, such as in country XXX". I already say something similar in my review of e.g. the Ktronik triple XP-G. Still, I would not use a lamp that I consider dangerous/distracting, even if it might increase my safety because in doing so I would make the road a more dangerous place as a whole. This 'I need to protect myself' argument is often nonsensical. You should not need a bike taillamp that is brighter than a car's taillamp. If you do so anyway, , then at least in cities you are endangering others by attracting too much attention. So that attitude is unacceptable for me except in special cases, such as Australia on roads with low numbers of cyclists and cars on roads where cars go very fast and you won't endanger anyone, as there are almost no distractions on it. But still, in that case the design principles I mentioned of an almost-collimated beam + wide diffuse beam hold. By using a beam in a cone shape you can blind the tailing traffic which might just home into you...
So, my views are based on what is best to see with headlamps, and to be seen with taillamps, without getting annoying. Getting annoying is counterproductive in lots of situations. The only thing you really need beside a well designed taillamp is education and good behaviour.
I asked someone about Australia, whether he thought a 3W taillamp would be useful as car drivers 'might be out to kill you': His response was interesting namely:
I think they are but I wonder how much difference a brighter tail light would make. Perhaps it just makes it harder for the driver to claim he didn't see you at the inquest.
That's an interesting point. I suppose as I thought it really comes back to driver education and attitude towards cyclists. This is where my idea of for example letting children cycle to school can make a big change as motorists then feel obliged to take note of cyclists...
I quote here what I wrote to someone wondering whether it is hard to make a lamp with good lighting pattern, slightly edited for clarification:
Well, it is apparently. For taillamps the funny things is that the old incandescents had good visibility, esp. a fairly collimated beam (the taillamps all have a lens...). Newer LED taillamps don't except for the Line plus and the Lumiring. Peter White notes on his page about LED taillamps that they put out plenty of light and thus don't need optics, which (again) shows he doesn't properly test as most taillamps' visibility is poor at long distance (and too bright at short distance). But this same phenomenon of thinking 'LEDs need no optics' occurs with the manufacturers! They (at least the Dutch manufacturers) made good incandescent taillamps, so why not good LED based ones? I think it's a case of luxury: LEDs give plenty of light so they assume visibility is good just from that. And what about being annoying? I'm not sure how that comes through their own reviewing, as I said it seems either they don't test, or they think "wow, that's bright! That's really visible! It's good!" when in fact it's a really poor taillamp. The people who did the optics and design for incandescent taillamps probably are all gone and working elsewhere or retired so they won't disagree with the views of the designers and 'testers' of the newer lamps who put little effort into actual optics.
For headlamps: Making an optic is not easy. This takes months of work. I know because I'm in contact with someone who is working on a new lighting system and the problems encountered are very clear: Using large surface area LEDs, making a strict cutoff is hard (when given the restriction of a fairly small reflector/lens, say 50mm or less).
Even without the original engineers/designers, I think it's disgraceful that manufacturers who have been in the field of bicycle lighting for decades, make such a pig's ear of taillamps.
|To email me go to the email page|
Last modified: Tue Dec 18 07:10:02 CET 2012