Jul 07, 2023
Electronics Weekly's big guide to LED optics
Specifying an LED optic is easy isn’t it? Just buy one with the right beam
Specifying an LED optic is easy isn't it?
Just buy one with the right beam angle, and you are off. Right?
Well maybe, but probably not.
Because there are a few more things to think about before you can be pretty sure that what you buy will do what you want.
"The first thing a customer will say is: ‘I need a 10° beam angle, which optic do I need?’, and the first question we ask is, ‘which LED?’ said Mike Bean, head of design at LED optics firm Carclo.
Why does he have to ask? Because the LED choice will greatly affect the beam angle as well as the beam quality produced by any given optic.
There is a physical property of optical systems called etendue (French for space, expanse, stretch) that limits the minimum beam angle possible from any efficient optical system with a certain source size and a certain maximum diameter.
You are never going to get a pencil beam from a 1mm square LED with a 10mm diameter lens, collimator or reflector – unless you throw almost all the light away.
Luckily, you don't have to do the maths.
Instead, optics manufacturers such as Carclo, Ledil, Fraen and Polymer Optics publish beam angles for mix-and-match combinations of optics and LEDs, and some offer web-based parametric search facilities as well.
Actually, even if you wanted to do the maths, it would be difficult because LED makers are coy about revealing their die sizes.
For reference, Lumileds’ Rebel has a 1x1mm die, and the Rebel ES has a larger one, probably 1.4×1.4mm, both in the same size package.
In Cree's XP series, the XP-C, -E and -G are probably 0.7×0.7, 1×1 and 1.4×1.4mm respectively, again in identical packages.
To a first order approximation, doubling the die width doubles the beam angle for a given optic.
For example, Carlo's 30mm diameter 10755 optic gives 3.7° with the 0.7mm XP-C, and 7.1° with the 1.4mm XP-E.
And the figures will be similar for good quality 30mm optics from other manufacturers.
In this case, the angles are quoted ‘full-width, half maximum’ (FWHM) which is the included angle between points in the beam that are 50% of the peak intensity.
It is legitimate to describe a 3.7° FWHM beam as +/-1.85°, and also to describe it as 1.85°. In each case, the convention has to be clear in any discussion. All the angles in this article are FWHM values.
Occasionally ‘field angle’ is referred to, which is the angle between the 10% points.
20mm is a common diameter for optics and, loosely, an LED with a 1mm die plus a good 20mm optic from any reputable manufacturer will deliver a minimum beam angle around 11°.
10mm optics are going to be delivering something larger than 22°.
Spreading out the beam
All the figures above are the minimum beam angles possible with high efficiency – above 85%.
Once the light has been collected efficiently, it can be spread out in a smooth beam by moulding features into the front face of the collimator.
So collimators with plain fronts are the narrow beam versions, and similar ones with bumps (called micro lenses) or ridges on the front are wide beam types – with taller bumps loosely equating to wider angles.
‘Micro-lenses’ moulded into the front face of a collimator spread its beam wider. Here, Ledil also spreads with doughnut-like curvature on its Eva optic.
Torch, aeroplane, or painting?
Plain collimators produce beams that are bright in the middle and gradually dim towards the edges.
This kind of beam is great for a hand torch because you don't want deep blackness around the beam. Instead you want light to spill out so you can see things near the object you are pointing at.
But what about illuminating a painting in a gallery?
Ideally, you want a beam that evenly illuminates the painting with little variation across the canvas, that cuts off sharply at the edges so that the surrounding wall is not lit up.
This kind of beam is called a ‘top hat’ beam – from the graph shape when illuminance is plotted against distance (see graph).
For the same reason, the standard beam produced by a simple collimator is sometimes called a ‘bell’.
"To illuminate the picture, the customer's best illumination option is a top-hat function, not a bell, because they want constant illumination right across picture, then nothing outside.
"To get a circular top hat, we use ‘ripple’ optics to give as near as possible a top-hat."
Ripple in this case refers to rings moulded into the collimator front, that look like ripples in cross-section, and push more light to the outside of the beam to achieve even illumination.
They are so shaped to push very little light beyond the nominal cut-off angle.
And can the beam be square like the picture?
"You can do that using a non-rotationally symmetric optic," said Bean.
Rectangular beam collimators are produced by most optics makers by adding straight ridges to collimator fronts.
Their beams tend to be more letter-box than square, with rounded ends – sometimes called elliptical beams.
If the picture illumination task is that critical, either a custom optic or projector can be designed – the Mona Lisa is illuminated by Lumileds LEDs combined with Fraen optics – or a square mask can be suspended in space some distance from the front of the optic.
In the centre, Carclo uses a plain front to get the narrowest beam from its 26mm optic, then frosting on the left for a smoother beam with slightly wider spread, and striped micro-prisms on the right for a broad elliptical beam. In fron t is a 20mm Carclo elliptical, with two 30mm Ledil wide beams at the rear.
Wide beams
While etendue limits the narrowness of a beam, the physics of total internal reflection restrict maximum beam width when using collimators.
"TIR collimators are limited to angles of about +/-25° as efficiency starts to drop when you try to bend light too far," said Bean.
Ledil pushes this a little further with its Iris series of collimators that have a doughnut shape moulded into the front as well as micro-lenses.
Carclo's answer to very wide beam angles is its ‘bubble’ lens series which move completely away from the standard collimator design.
Bubble optics can produce a wide even patch that allows more efficient lighting of large areas rather than using an LED alone.
With no optics you get a light ‘hot spot’ that follows a COS(theta)^4 dependency. But with optics they push most of the LED's output to the edge to compensate for this dependency.
"The bubble option gives a very even beam because it is designed to throw a lot of light to the side," said Bean. "Looking at it, it is not bright straight on, but brighter edge on, then nothing."
And this is where aeroplanes come in – in the difference between lights to see by, and lights designed to be seen.
All of the example above are lighting tasks where illuminance, measured in lux, is important – the number of lux across the picture has to be constant.
The bubble optic produces even illuminance on a flat surface, but its intensity differs dramatically from different directions – this intensity (measured in candelas (cd)) is far from constant.
However, the lights on an aeroplane's wing tips, or on the tops of high masts, have to have constant intensity, at least in the directions that they are designed to be viewed from.
Essentially: evenly illuminating a flat surface requires a lot more intensity in the edge of the beam, whereas being evenly visible from any direction is the same as evenly illuminating the inside of a sphere.
On a tall transmission mast, it would be realistic to mount eight 45×10° collimators around the top at 45° intervals to deliver all-around visibility with sufficient vertical spread to be useful.
The beams would be profiled to deliver even intensity across their 45°, so the mast light will appear to be the same brightness from the same distance in any direction.
The ‘top-hat’ illuminance plot of the +/-65° bubble lens mentioned in the text, showing how evenly it would illuminate a wall – the pool of light is 10m across even though the LED is only 2.5m from the wall.
The twin-spiked intensity plot shows what had to be done to intensity to get that illuminance, and how the same optic would not be suitable for a beacon that had to be viewed from anywhere within that +/-65°. (Cree XP-E LED with Carclo 10406 optic in both cases).
A glimpse into a secret world
The exact profile of any particular LED optic combination is generally down-loadable in machine-readable ‘photometric’ files for optical modelling programmes, either in Elumdat (.ldt) or .ies form.
These formats are extremley useful if you want to predict the combined effect of putting more than one optic together. You don't need to build any optics and you can simply look at many different combinations.
The .ies form is also human-readable with experience. "You can open a .ies file in Notepad," said Carclo's Bean. "There is a header that tells you about the LED and optic types then a series of values."
The top few rows of values provide details about the lamps (or LED's) and the angular ranges covered (both vertical and horizontal).
Then beneath the angle values are rows of all the intensity values.
Two formats are available for ies data. Type A where the intensity values are read in a raster angular slice. Type C where the values are read by taking radial angular slices.
So, for Type C ies format, with a rotationally symmetrical (circular) beam, all the values in any particular column will be the same or similar.
How much light do I need?
Once a beam width and beam profile have been chosen, the amount of light needed to drive the system is estimated.
The efficiency of a good optic, which is typically 85-95% (or lower in smaller optics), has little to do with how much light ends up in the nominal beam because it is a measure of how much light passes out from the front face of the optic compared with light leaving the LED.
Efficiency takes no account of how much light makes it into the nominal beam, and how much spills out between the edges of the beam and +/-90° – it can be a significant fraction.
To allow for this spillage, the figure to look for in the manufacturers data is the cd/lm or lux/lm figure which described how lumens from the LED will be turned into intensity or illuminance at the centre of the beam.
Intensity (cd) is constant with distance, illuminance is quoted at various distances.
Both are the same at 1m range, so apply the inverse square law to convert from cd to lux.
Say 10lux is need on a picture 4m from the LED-optic.
The beam angle is calculated by simple trigonometry and a suitable top-hat beam optic is chosen.
The data sheet says it delivers 4cd/lm
Cd and lux are the same at 1m, so at 1m it will give 4lux/lm, and through the inverse square law it will give:
(1/16)x4lux/lm at 4m = 0.25lux/lm.
40lm from the LED will deliver the required illumination
One last thing
Collimators are designed to be mounted a certain distance above the LED.
Optics makers supply holders to fix this height for each LED-optic combination.
If no holder is used, great care must be taken to mount the LED at the right height, and concentrically with the die.
Manufacturers do not publish this information, but will provide it.
And so
Understand your beam requirements – the necessary beam profile will depend on what you are illuminating, and will be completely different if you are making something to be seen, like a beacon or bicycle back light.
Use the geometry of your application to determine the required beam angle.
Then use cd/lm or lux/lm figures to work out how hard to drive your LED.
If the LED can't do it, try finding an optic that spills less outside the beam.
For special applications, take a peek at the .ies data.
If you get stuck, have a conversation with an optics company.
Collimators, also known as optics, for LEDs are actually two optical components merged into one.
Around the edge is the equivalent of a parabolic mirror – directing sideways light from the LED forwards by total internal reflection (TIR).
Light that would miss the reflector is caught by the second element, a more traditional lens structure in the middle of the optic. Sometimes there is a hole in the front of the optic above the lens. Its convex surface gives the optical designer yet another chance to hone the beam.
Mirror reflectors do not have the equivalent of this central lens and spill more light sideways, with one exception – The 225 from Polymer Optics is a reflector with a Fresnel lens suspended in space over the LED.
While the reflector in an optic produces a smooth beam, the lens projects a square image of the die which the optical designer smoothes out using various tricks including frosting the front of the optic.
Spreading out the beam ‘Micro-lenses’ moulded into the front face of a collimator spread its beam wider. Here, Ledil also spreads with doughnut-like curvature on its Eva optic. Torch, aeroplane, or painting? In the centre, Carclo uses a plain front to get the narrowest beam from its 26mm optic, then frosting on the left for a smoother beam with slightly wider spread, and striped micro-prisms on the right for a broad elliptical beam. In fron t is a 20mm Carclo elliptical, with two 30mm Ledil wide beams at the rear. Wide beams The ‘top-hat’ illuminance plot of the +/-65° bubble lens mentioned in the text, showing how evenly it would illuminate a wall – the pool of light is 10m across even though the LED is only 2.5m from the wall. The twin-spiked intensity plot shows what had to be done to intensity to get that illuminance, and how the same optic would not be suitable for a beacon that had to be viewed from anywhere within that +/-65°. A glimpse into a secret world How much light do I need? One last thing And so Collimators, also known as optics, for LEDs are actually two optical components merged into one. Around the edge is the equivalent of a parabolic mirror – directing sideways light from the LED forwards by total internal reflection (TIR). Light that would miss the reflector is caught by the second element, a more traditional lens structure in the middle of the optic. Sometimes there is a hole in the front of the optic above the lens. Its convex surface gives the optical designer yet another chance to hone the beam. Mirror reflectors do not have the equivalent of this central lens and spill more light sideways, with one exception – The 225 from Polymer Optics is a reflector with a Fresnel lens suspended in space over the LED. While the reflector in an optic produces a smooth beam, the lens projects a square image of the die which the optical designer smoothes out using various tricks including frosting the front of the optic. Steve Bush