# Designing brake light with specific luminous intensity

I am currently designing a brake light for a Formula Student car and I am looking into existing solutions on the market for inspiration.

The two specific solutions are this, and this.
The LED I found to be suitable is this one.

Based on my research I was thinking about having a regulated buck converter converting from 12 volts to the forward voltage of one LED (2.3 V in this case), and putting them parallel to each other. The converter being regulated, it would maintain that voltage so it would not cause thermal runaway, meanwhile maintaining low power consumption. I would not like to have multiple LEDs connected in series, because if one fails, multiple would fail.

I have troubles finding a straightforward way to calculate the total light output for the brake light from the luminous intensity of a single LED, and the way these correlate to each other.

How does having multiple LEDs next to each other affect the total light output, what spacing and viewing angle would be the optimal, do I need a lens on top of these LEDs and are there any better options?

How should I approach this design question which goal is to achieve similar light output as those I listed previously?

Also, the bulb size for the LEDs in the listed examples seems to be 9.5 and 6.8 mm respectively.

• Do you need the two brightness levels - rain and brake? Is this possible with the led you have chosen? – Solar Mike Oct 13 '18 at 12:27
• While this isn't a bad question, it's really a lot of questions all in one, and some of them really aren't electrical engineering (viewing angle, lenses, etc) - making the whole thing overly broad. I suggest you narrow it down a bit and/or split into multiple specific questions. – brhans Oct 13 '18 at 13:01
• I made an LED brakelight for a motorcycle. All I can say, is expect to make several engineering changes. Test and re-test. Every LED is different. What might seem fine on a bench, may be blindingly bright at night. – rdtsc Oct 13 '18 at 14:27
• @brhans where do you suggest I should post a question regarding the viewing angles and lenses, maybe Physics? – domenix Oct 13 '18 at 15:45
• @rdtsc the car is only going to be used during the day, so blinding at night is not really problem, although I want to limit it to max 300 cd, which is the maximum value in North America, but the UN regulation is even less 185 cd. The regulation in the competition rules only state that it has to be clearly visible during daylight from behind, although I would like to make it similar to existing products in terms of capability. – domenix Oct 13 '18 at 15:50

You simply add the output of each LED to get the total output.
The distance between the LEDs will have negligible effect.

One confusing aspect of the angles used in optical measurements is the candela "view angle" (luminous intensity) and the angle at which the light travels (illuminance).

For example let's use the LED you selected.
The Kingsbright LED has a view angle of 70°.

At a distance of 20 feet the position of the LED is inconsequential. If the LED were positioned 1 inch over, within the break light fixture, point C would also move 1 inch.

Only directly behind at 0° will the LED been seen at 100% illuminance .
As the angle increases the illuminance decreases.

I created a web app that shows the relative illuminance/irradiance between various LEDs. Angle Irradiance Web App
This compares the angle irradiance with one LED facing sown and the other facing up.
Compare SSL 80, SSL 120, and SSL 150. These are OSRAM Oslon SSL LEDs.

In the datasheet there is a graph that shows the intensity (candela) with respect to view angle. The highlighted area is the 70° (±35°) which is included in the specified 3.2 cd (3200 mcd).

This graph show for each angle what the relative illuminance is.
For example at 20° the illuminance is reduced to 70% compared to 0°.

The angle here shows the illuminance decreases as the angle increases.

What this example of 0° vs. 20° is saying if you were 20' directly behind the brake light (at 0°, point B) and a light meter read 100 lux if you moved 7.3' to the left or right (20°, point C) the lux meter should read 70 lux according to the above chart.

This angle has nothing to do with the view angle specified for the luminous intensity.

## Comparison of the two products you linked

It is not simple to compare two LED when you have the proper specifications. The problem here is the authors of the specifications do not understand photonics.

One is 250,000 mcd @ 0° and 200,000 mcd @ 10° for all 15 LEDs

The problem here is candela is a measurement of lumens (lm) per steradian (sr). Where a steradian is the amount of light within a cone of a specified angle.

This candela view angle of the cone has nothing to do with the "off axis angle" chart shown above.

A steradian of 0° does not exist. It has zero light. If you were to convert candela, with a view angle of 0°, to lumens the result would be zero lumens.

This product gives the spec for each LED of 5,700 mcd @ 30° for a total of 102,600 mcd for all 18 LEDs @ 30°.

What is troublesome is where this one says:

Highest light output available, exceeds 900,000 mcd in light meter tests.

The problem is a light meter measures in lux (illuminance) which is distance dependent so this spec is meaningless. The distance of the illuminance measurement is very important as the illuminance (lux) is subject to the Inverse Square Law.
I have an app that converts lux based on distance:
Inverse Square Law App

I can compare this 5700 mcd @ 30° to your Kingsbright 3200 mcd @ 70°

5700 mcd @ 30° = 1.2 lm
3200 mcd @ 70° = 3.6 lm
Source: Rapid Tables online cd to lm converter
NOTE: lumens is a measurement of the total light being emitted from an LED in every direction. Candela measures only the light within the cone.

I have a very detailed explanation regarding comparing LEDs in this answer I wrote earlier today: Cheap way to power blue LED on coin cell

The 5700 mcd @ 30° does not specify the wavelength so I cannot estimate the difference based on the Photopic Luminous Efficacy from the Relative Sensitivity Curve for the C.I.E. Standard Observer

Looking at the photos of these products it appears they are more red-orange than red. When the color red is specified the wavelength can vary for 620 nm to 660 nm.

Here I converted wavelength to RGB using CIE standards.

• Both specification sheets are pretty clear to me. Specifying Cd at various angles is the only sensible way to specify the pattern of the LED. You might be confusing at 30 deg with over 30 deg... Intensity is measured over an infinitesimal solid angle, at some position. – tomnexus Oct 14 '18 at 5:41
• @tomnexus Oh really? Is cd @ 0° sensible? I'm not confusing anything, 30° is 30°, not "over 30°". What is an "infinitesimal" solid angle? Zero degrees? 0° is nonsensical. How does one measure 900,000 mcd with a lux meter? Both product datasheets are erroneous. If they are clear to you it is because you do not understand photometric measurements well enough to to see the errors. For further info see my answer at electronics.stackexchange.com/questions/400835/… – Misunderstood Oct 14 '18 at 6:37
• If you chose. Lens design for 10 Cd at 1m with a 15 deg beamwidth, why would you want to measure lumens over a 1 radian cone, when the light does not fill a cone. > @tomnexus – Tony Stewart Sunnyskyguy EE75 Oct 14 '18 at 17:01
• Consider how you would specify the luminous intensity at the centre of the beam. You'd say x Cd at position angle 0 degrees. If you don't like the number zero for some reason, then give it a name like "the middle" or "boresight"; it doesn't change the intensity. Then, at the edge of the beam? This would be y Cd at angle z. It would make a graph, and you included one. So over every tiny part of its beam, the LED has a different luminous intensity. The first datasheet you quote clearly gives it at two points: the middle, and the edge, 10 degrees away. That is all the numbers mean. – tomnexus Oct 14 '18 at 18:23
• @tomnexus You do not use the term luminous intensity to measure the center of the beam. Luminous intensity is a measure of the amount of photons (flux) within the specified cone. Candela is always specified with an angle and the angle must be greater than zero. What you are referring to is illuminance which is measured in lux and not cd. You can measure the lux at 0°. But you cannot measure cd at 0°. You cannot compute lux from cd. Easiest to measure lux with a meter. You can estimate lux from cd. You can learn how here: intl-lighttech.com/ilt-light-measurement-tutorial – Misunderstood Oct 15 '18 at 3:05

Firstly, power LEDs are bound to be driven by a constant current source. Direct parallation of LEDs is also a bad idea. Unless you fit each LED with a CC driver or resistor (when using a resistor you should limit the current well under maxim) It's better to put some in series. The chance of led failure is extremely low if you drive them correctly. In a series connection every member of the circuit gets the same current and it eases the voltage difference as seen by a switchmode driver. You can use a hysterisis mode current driver designed specifically for LEDs.

Modulate the LED brightness by applying PWM >4KHz through chip control or adjust current feedback in some way. Use powerful LEDs more powerful than you need so that you can tune the brightness.

LED chips are usually very broad by default. Some types are pre fitted with lens so they are narrower. If you have specific regulations requireing specific brightness or specific angle, please let us know.

EDIT: Having viewed the 2 links provided by you I think they are 10mm LEDs. https://cdn.solarbotics.com/products/photos/e8bdbcc82c6ba23a4476b929e30dd0c4/ubled10-img_0116.JPG

Oh another thing. If you decide to use a resistor approach your drive voltage should be by some order higher than the LED Vf. Say if your LED Vf is 2.3V, do not drive from 2.5, 3V, 3.3V source. Drive from 5V or higher through a resistor instead. This is because modern LEDs are sensitive to minute changes as a result your light may flicker if your drive volt is too close.

• When you wrote "It's better to put some in series," did you mean that it is better to have parallel strings of series LEDs, so that if one fails then only that string fails and the rest still work? – Andrew Morton Oct 13 '18 at 14:34
• Hi Well, first of all you shouldn't really worry at all about LED failure. It is extremely rare as long as you drive them correctly (this includes heatsinking if they get hot) and especially if you have a reputable part source like digikey. I understand that your goal is to make the design reliable but spontaneous LED failure is typically not factored into the design. Since you will probably be driving all leds from 12V you can't make them all go in series so it has to be partly in series and partly in parallel. (derate them as well if you don't feel utterly confident.) – user199643 Oct 13 '18 at 14:41
• Why you drive LEDs with constant current: electronics.stackexchange.com/questions/298564/…, and maximintegrated.com/en/app-notes/index.mvp/id/3256. Note that if you parallel two LEDs, whichever one is warmer will want to "steal" current from the other -- which makes it warmer yet. Any time you have forward-biased diodes (such as LEDs) you need to do special things to make them behave in parallel; connecting them in series (if you can) solves the problems automatically. – TimWescott Oct 13 '18 at 16:13

1). You need to get familiar with LED specs.
2)Then finally after you learn a bit, you may prefer the benefits of man-years of R&D and choose the best tail lights that fit your surface design or modify to fit. Your choices are amateurish. Pick from Buick, Lexus, Mercedes, BMW etc. Audi was the first to use LEDs over 10 years ago. Go to a pick-a-part auto wrecker. Do some research into the lens, fixture and electronics. You have so much to learn.

# Indicators:

= direct light; Generally millicandella or [mcd] or candela [cd] is the observed luminous intensity.

This attenuation from a point source by the distance squared relative to the standard measurement.

= indirect reflected light; Generally measured in total lumens of light output in all directions where beamwidth affects the mcd or apparent brightness.

• The unit of lumen is not useful for tail lights.

## Illuminance: [Lux or candle power]

• The preferred method to verify a Luminaire (lamp fixture).
• measure with a calibrated light sensor in Lux at a specific distance such as 1m with no ambient light.
• You can also use an old 35mm camera to get relative measurements or an inexpensive Panasonic CIE corrected 5mm Light detector with an ‘L05 reg and 9V battery, for a few bucks then calibrate with a Lux meter using a DVM in Vdc with a log scale spanning 4 decades.
• you can also simply compare car tail/ brake light intensity in 5 deg increments at an exact distance of 1 m or more for relative good Illuminance results
• if you use the examples in your question, these are rookie designs.

Unit system SI derived unit Unit of Luminous flux Symbol lm. (No cap.)

## Unit conversions

The lumen is defined in relation to the candela as

## 1 lm = 1 cd ⋅ sr.

• A steradian is also equal to the spherical area of a polygon having an angle excess of 1 radian, to 1/4π of a complete sphere, or to (180°/π)^2 ≈ 3282.80635 square degrees.. Ref Wiki

There are some caveats to acuity and apparent brightness or emittance which needs to be large enough to exceed the retina resolution. So the larger area may be dimmer for the same optical power but better acuity from the longest range viewable. But you don’t need runway lights,head lights , just brake and tail lights.

The geometry of modern cars tends to spread the apparent area with a stepped fresnel lens, yet still have narrow beamwidth with less loss. Look for these, as they are ideal.

The colour you want is probably high efficiency RED which is 630nm dominant, or nmD which is eye-corrected wavelength vs true deep red 660 nmD which looks nicer but requires more power for the same brightness.

The beam angle is important to consider what you need. If you cut the angle in half, the brightness does not quite double due to lens loss but this is my rule of thumb

• x2 -10% intensity change for each 1/2 reduction in the beam width from lens magnification, measured at half power angle. 30° is common, also 15° whereas SMD with no lens has a wide angle thus no lens gain [in mcd or cd]. (Similar to antenna gain vs diversity angle or 1/2Power beam width)

But your power specs are critical to this requirement of luminous Intensity.

I used to sell only the best 5mm LEDs and were twice as bright as the ones used here with 30° Kingbright , when normalized mcd / mW and still have tons left. They cost a lot more than open bin, but much cheaper than these amateur epoxied assemblies in your links.

For PS , you need to specify Vin range and desired output power and efficiency. I might consider a 98% Boost regulator with constant current over a 92% buck regulator, the choices are many, but do you want to make or buy?

• Intensity at a distance is illuminance measured in lux (lm/m²/s) incident a surface. Candela (intensity) is the flux within a steradian cone per unit time. cd= lm/sr. Good answer overall. I also prefer 660 nm deep red especially for brakes, emergency vehicles, and Christmas lights. And growing plants. – Misunderstood Oct 14 '18 at 2:46