Led headlights – de-dazzling the beam

Inspired by comments made on one of the Electronics Weekly blogs, I thought I would have a go at estimating which led-collimators combination would not dazzle on-comers if used in a vehicle headlight.

dazzle.jpgInspired by comments made on one of the Electronics Weekly blogs, I thought I would have a go at estimating which led-collimators combination would not dazzle on-comers if used in a vehicle headlight.

A dazzle-free beam must be possible, because the IQ Fly pictured here can do it.

Now, I have to admit that I am winging the calculations here – so any corrections or refutations to the following are welcome.

As far as I know, the diameter of an optical system is related to two things: the size of the led die, and the maximum rate-of-change of intensity with respect to angle within the required beam.

I am guessing that there is probably proper optical-speak for this last thing, but what I am trying to get at is: the more focussed the beam must be, the bigger the lenses or mirrors have to be.
And if the beam is not circularly-symmetrical, it is the most focussed part of the beam that determines the size of the optics.

I only have one set of headlight lighting regulations on file, and that is an ancient copy of the German bicycle light regulations.

The beam they demand is quite complex, but the most demanding part is that the output must be greater than 10 lux horizontally straight ahead, but less than 2 lux 3.4degrees above the horizontal – this is the anti-dazzle ‘cut-off’ that is dreaded by led headlamp designers working on all kinds of vehicles.

In short, the light has to be 80% down at 3.4deg from the beam centre.

If this can be met, all the rest of any arbitrary beam can be made by spreading it out in the required direction.

As far as I know, if it is not met, there is no way of fixing the situation without a bigger lens, mirror or mask further out in the light field.

The beam width of led collimators and reflectors tends to be quoted at 50% intensity – so the output of a +/-5deg reflector will be 50% down at 5deg from its main axis.

Running a ruler over some optical spectral radiation pattern graphs of collimators and mirrors designed to work with 1x1mm leds, I have deduced that outputs tend to be 20% down about 1.5 times the beam width.

This means an optic quoted as +/-2.27deg would (if my luck is holding out) just meet the 10 and 2 lux figures separated by 3.4deg.

I have had a huge hunt through manufacturers data sheets and the only family of collimators or reflectors that I have found that can do this are sophisticated reflector-refractors designed for Luxeon I and III leds by Carclo.

The 50mm diameter 10025 (scroll to page 31), for example, produces a +/-1.5deg beam.
A bit more rough calculation suggests this could be wound up to 40 lux in the centre before breaking the 2 lux limit at 3.4deg.

The next one that comes close is another innovative reflector-refractor – this time the 30mm +/-2.5deg 202 for Cree XR-Es from Polymer Optics.
However, it is only 80% down at 4.3deg.

So this suggests the minimum diameter of optics needed to meet the German StVZO lighting regs with 1mm leds is between 30 and 50mm.

Something that backs this up is that the B&M IQ Fly (pictured), an XR-E-based bike light that does meet StVZO, and produces 40 lux, has 40mm optics.

These are of the sort I have always thought would be best for led headlights of all sorts – a side-fed reflector much like some RF antennas.

This all bodes ill for my own Mark V which with its 20mm collimators does a fine job of lighting the road from 3W – even a spectacular job – but may have to come of the bike because it produces too much light above the horizontal.


Comment below, or to alice@electronicsweekly.com



  1. Pity the poor cyclists and pedestrians.

  2. You are correct, there is a fundamental relationship between the output diameter of the optic and the size of the source relative to how tight you would like to focus the beam.
    If you are interested in the physics of it, you can look up the term etendue. Roughly speaking, the area of the source multiplied by the solid angle of emission of the source has to be equivalent to the area of the optic multiplied by angle of emission from the optic. (in a limitng case). So you know your source area (1 mm) and the emission angle is +/- 90 degrees for an LED usually, and if you try to squeeze that into a 1 degree beam, you can see that the area of the optic has to get to be about 90X larger than the source area for that to work. This is only rough and not a technically accurate calc.
    Custom optics can be developed that create asymmetric beams to meet the bike regs you llist, but if you just want to use standard collimators to achieve the same thing, your really can just fabricate a small metallic shield to slide over the optic to block the uplight. It probably won’t need to be too long (maybe an 1″ or 2″) to get what you are looking for, and the uplight will get reflected down and possibly into a usable area depending on how you configure the part. So you really don’t lose any efficiency and your on road efficiency my actually increase.
    If you just sketch a cone coming from the end of the optic of the light output beam and then create a cylinder on the top half of the optic that extends from the optic to stop uplight, you would have a starting point for a design to play around with.

  3. The best anti dazzle for headlights would be polarised headlights and windscreen at about +40 degrees.
    Then the oncoming headlights would be viewed at -40 degrees and be attenuated but still visible but the light from your own headlights would be maximal. The rear screen would have to be polairised at -40 to minimise dazzle from the vehicle behind. Simple really. Now if the LED manufacturers can produce a polarised LED ……….

Leave a Reply

Your email address will not be published. Required fields are marked *