One of the challenges randonneurs face is night riding. While it is theoretically possible to complete the various brevets entirely in daylight, virtually all randonneurs spend some brevet time in darkness and the subject of bicycle lights is a frequently debated subject of conversation in many randonneuring clubs. While there are probably as many solutions to the lighting problem as there are randonneurs, I'm not going to discuss all the options here. Interested readers are encouraged to explore the many options available in systems using halogen bulbs, linear voltage regulators, various battery packs and/or generators. A wide range of systems have been successfully used in events like PBP and BMB and quick web search should reveal a wealth of information.

The question I intend to answer here is this: Can the current generation of LEDs provide sufficient light for randonneuring or bicycle commuting applications? The answer to this question is ultimately a very subjective one, since each rider has his or her own definitions of what constitutes "sufficient light" but for me the short answer is "yes". The long answer will constitute the bulk of this report.

Light Emitting Diodes (LEDs) have been around for years and while the red LEDs are almost universally used in bicycle tail lights, it has only been in the past few years that newer generation white LEDs have been available for use in flashlights, head lamps and bicycle front lights. These LEDs still aren't capable of putting out the amount of light that a halogen or krypton bulb can produce, but they do offer several advantages over a traditional filament bulb.

LEDs are solid-state devices and under normal conditions they virtually never burn out. Over-volting an LED will destroy it, but this doesn't happen with a properly designed circuit. Unlike filament bulbs, LEDs are unaffected by jarring shocks, such as those encountered in riding over cobblestones. Finally, LEDs draw very little power. This allows the creation of small lights that have very long run-times. For example, my helmet light, a Princeton Tec Matrix, uses two AA cells to drive three white LEDs. With a fresh set of alkaline batteries, the light will run for 40 hours. With lithium AAs, the run time is over 100 hours!

I have used the Princeton Tec Matrix head lamp for the past two years and I've found that it is very good as a helmet light. It casts good light for reading cue sheets and lighting road signs and route markers. It also provides sufficient light for repairing a punctured tire or making other roadside repairs. It is also very good for catching the attention of motorists at intersections. The fact that the beam from a helmet light goes where you point your gaze makes it a very useful addition to a night rider's lighting system.

I have experimented with other helmet lights, both incandescent and LED and find myself always returning to the Matrix. Higher powered incandescent systems require external battery packs, suffer from shorter run-times or both. Smaller LED head lamps, like the Petzl Tikka or Zipka are lighter and have decent run-times but several factors make me favor the Princeton Tec Matrix over the Petzls. First, the Matrix casts a better beam. Second, the Matrix uses a solid-state circuit to increase the voltage from 2 AA cells to power the three LEDs, whereas the Petzls use 3 AAA cells to power their three LEDs. This leads to problems that I'll discuss in detail a bit further on. Third, the Matrix has a pivoting mount that I find more useful that the Petzl's fixed angle. Fourth, the Priceton Tec is far more rugged and better sealed than the Petzl (although I have never had any problem with the Petzl failing and I've used my Tikka on some very wet rides).

One of the main strengths of the Petzl lights is their extreme light weight. Unfortunately while the Petzl lights weigh less, their unsophisticated circuitry means that the brightness of the light drops off as the batteries drain. Alkaline batteries have a sharply sloped discharge curve, so this means that for much of the life of the batteries, the Petzls are putting out less light than the LEDs are capable of emitting. The Princeton Tec Matrix, on the other hand, puts out strong, steady light for almost the entire life of it's batteries. The problem with the Petzls would be far less if lithium AAA cells were available. Lithium cells have a discharge curve that is fairly flat for the bulk of their life before ending in a very steep drop. Using Lithium cells to power simple LED circuits gives far better light for the bulk of the time that running the same simple circuit with alkaline cells. Unfortunately, to the best of my knowledge, no company is currently manufacturing lithium AAA cells.

In the autumn of 2001 Cateye began producing the HL-EL100 bicycle headlight. Like the Petzl Tikka, the HL-EL100 lacks any sophisticated LED driver circuitry. The lamp powers it's three white LEDs directly off 4 AA batteries with a 15 ohm resistor as a current limiter. However, since the Cateye uses AA cells, lithium batteries can be used in the light. While the lithium cells are more costly than alkalines, their better discharge curve and lighter weight make them a good choice for powering this lamp.

In December of 2001 I began an experiment to use LED lights for my bicycle commuting and randonneuring training rides. My forward bike lighting consisted of and Cateye EL-HL100 paired with a Cateye HL-MC200 Micro Halogen II 2.4 watt halogen light. I also had the Princeton Tec Matrix mounted on my helmet. My intent was to use the EL-HL100 as my primary light and only turn on the HL-MC200 when I felt I needed additional light. I anticipated that I would feel the need for more light in situations such as high speed descents or particularly rainy nights.

Many people approach night riding from an absolute perspective and seem intent on reproducing the visibility they have in broad daylight. I've often heard people say things like "I need a 35 Watt system because I ride a lot on unlit country roads." In my experience, the unlit country road scenario is actually one in which a minimal light can be quite adequate since the human eye can adapt to the dim conditions. Urban riding, on the other hand, tends to require more powerful lights, since oncoming traffic and other light sources interfere with the eye's night-adaption mechanisms. In the urban environment I've found that a cycling cap is one of my best pieces of night-riding equipment. With the cap tucked under my helmet and the brim shielding my eyes, I can avoid the worst of the glare from oncoming vehicles.

Both the Cateye EL-HL100 and the translucent-cased HL-MC200 allow some of their light to spill out through their cases. In the case of light spilling out the sides, there may be some marginal added safety offered by this light, but the light spilling out the top of the light does nothing but interfere with night vision. On my lights, I made the top of the lights opaque with black electrical tape to eliminate this top glare. Also the mechanism holding the case of the lights together is a bit fragile, as is the single screw holding the light body to the mounting clamp. I made some thick rubber bands by cutting up an old bicycle inner tube and I use these bands to keep the lights and clamps secure. Finally, Zach Kaplan has reported that he's had problems with water accumulating in his EL-HL100. I ride a lot in the rain and haven't had this problem, although the above mentioned modifications make the lights better sealed than unmodified lights. Zach also had his EL-HL100 mounted upside-down and this may contribute to water accumulation problems.

Throughout December of 2001, I rode with my dual Cateye plus Matrix lighting system. Every work day I'd commute for 1/2 hour in the dark going to work and 1/2 hour in the dark coming home from work along suburban and urban roads. In addition, several nights a week I'd add in a night loop around Mercer Island, Washington or around Lake Sammamish. On the weekends, I'd do longer rides that would begin at my suburban home at 3:00 AM and head out into the wooded country roads heading north and south of Issaquah. In the month of December, I logged 30 hours of night riding. I used the halogen light for a total of 20 minutes in that month, while I used the LED lights for the full 30 hours. I never had to replace the batteries. I was not run over by cars. I had no close calls.

I did descend a bit more slowly in unfamiliar terrain and I did find that the beam cast by the Cateye HL-EL100 to be pretty diffuse. The beam cast by the white LEDs is quite reminiscent of moonlight and the LEDs cast a 20-degree beam that doesn't carry very far. While I found the HL-EL100 to be adequate, it was far from ideal.

In late December of 2001, Dr. Marty Goodman posted a note to the Bike Current newsgroup in which he described modifications he'd made to a Cateye HL-EL100. Marty had replaced the three 6,000 mcd white LEDs with three 10,000 mcd green LEDs. While these green LEDs draw the same power as the white ones, they put out more light. I quickly wrote to Marty to see if I could get him to make me up a light. He generously agreed in exchange for a very reasonable fee and the promise that I'd write up a report of my findings. This web page is that report.

Since the HL-EL100 and the HL-MC200 use basically the identical case and mount (it differs only in the color), when I got the Green-EL100 I mounted it on my bike in place of the seldom used HL-MC200. I loaded fresh lithium AA batteries into both the White-EL100 and the Green-EL100 and since January 1, 2002, I've been using these lights in tandem as my bicycle lights.

Several factors combine to make the Green-EL100 superior to the stock White-EL100. First off the LEDs really do seem to be putting out more light. 10,000 mcd is more than 6,000 mcd and this is a noticeable difference. Second, the white LEDs have a 20 degree viewing angle while the green LEDs have a 15 degree viewing angle. This means that the green LED projects a nicer, more focused beam. Finally, the frequency of the green light, 525 nanometers, is right between the scotopic and photopic peaks of human eye perception. The scotopic peak (555 nm) is wavelength of light where the cones in the human eye are the most sensitive. The cones are used primarily in daylight for distinguishing fine detail in an image. The rods of a human eye, on the other hand are most sensitive to light at photopic peak of 505 nm. Rods are more numerous than cones (you have about 120 million rods in your eye, but only 6 to 7 million cones) and while rods don't distinguish color, they do provide you with your night and peripheral vision.

This raises the question, doesn't it seem weird to ride around with everything lit up green? The answer is no, because night vision is colorblind. As the Moody Blues observed:

It's actually not the moon that removes the colors from our sight, it's the physiology of the human eye. Studying the physiology of the eye explains another phenomena many night riders have noted: sweeping your gaze and not looking directly at the item lets you see things better at night. Jack Eason explained to a friend of mine how he'd learned that trick back in the war and I'd always wondered why this was so. It turns out that while the cones have their highest density at the center of the retina, the density of the rods peaks a few degrees off center. So in low light conditions, your best vision is not at your center of focus.

At the time of this writing, I've been riding with the Green-EL100 and White-EL100 for three weeks worth of night riding (about 22 hours of riding in darkness). The first night I was out a man walking his dog on Mercer Island commented "Wow, that's some light!". I'd been concerned that perhaps just having a green light up front would confuse drivers, but I always ride with the green and the white light on simultaneously and I also have the white Matrix helmet lamp on as well. As was the case in December, I've had no close calls with any drivers, potholes or nocturnal animals. With the Green-EL100, I can see further and I've descended hills at speeds up to 50 kph.

I'd pretty much decided that the dual Cateye system (one white and one green) would be my head lights of choice but LED technology keeps moving forward and companies continue to produce new and interesting products. One product that consistently gets good reviews is the Princeton Tec Impact flashlight.

The Princeton Tec Impact uses four AA cells to drive a single white LED. I haven't opened up the LED assembly, but I don't think the light has any sophisticated driver circuitry. To get the best performance from this light, I've been running it with lithium batteries. Princeton Tec claims the Impact will run for 150 hours on 4 alkaline AA cells, with lithium cells the light will probably run for twice that time. Like the other Princeton Tec products I've seen, the Impact is very well made and weatherproof.

The Impact does have some very sophisticated optics. The Impact uses a reflector and a curved lens to focus the light from the single LED into a very tight, well-defined beam. The beam cast by the Impact is far superior to that cast by the stock white 3-LED Cateye EL-100 and is even better than the beam cast by my modified green 3-LED EL-100.

By wrapping a bit of inner tube rubber and electrical tape around the body of the Impact, I made the light fit into a bike light holder designed for a Pelican Versabright II flashlight. In subsequent test rides I have found the Impact to work quite well as a bike light. The beam does a good job of lighting the road.

One disadvantage of the Impact is that because the beam is so well-focused, there really isn't any light directed out the sides of the light so a rider is probably a bit less visible to cars approaching from the side. However the use of other lights (helmet lights, flashers, etc) can address this problem. Also, I think the optics of the Impact are designed around the beam spread of the White LED, so it might loose it's sharp focus if one replaced the white LED with a green LED with a different beam spread. But performance of the stock Impact light is good enough that I don't think such modifications are needed.

Based on the few hours of night riding I've done with the Impact, I think it's a very good light. I have heard that Princeton Tec is working on an actual bike light. While this may be something as simple as a single LED light with Impact-style optics packaged in in a bike mount, they may in fact be working on something even better. A light having multiple LEDs together with good optics and voltage regulator circuitry would certainly be interesting. But for now, the Impact is the nicest single LED light I've seen. It's what I'm using now as my main headlight.

I'd like to conclude that while I've found that LED lighting solutions work for me, every person is different in what they find adequate for night lighting. If you try LED lights and don't find them adequate for your needs, please don't try to tough it out. Find a lighting system that works for you and be safe out there.

Here are a few web pages with more detail on night vision and LEDs.