However much a lit ultrabright LED (Light Emitting Diode) light may look like a light bulb, it is at heart an electronic
assembly. As such, there are many specifications that are thrown about, most having little meaning
to the lay person. As there are many sites which give explanations to experts, consider this as
information for the 'less than expert'.
How 'Ultra Bright' L.E.D.s workUltra-bright LED lights consume little power, but produce far less light per watt than a fluorescent lamp. Where LED's excel is in applications where you want a lot of light in a given direction, such as a reading or chart light, or a specific sector, such as a running light on an RV (recreational vehicle) or a vessel (boat or yacht).
The physical arrangement of an ultra bright LED is really no different than Fresnel used in his lighthouse designs using a kerosine wick as the basic light source. The LED has a light source (a light emitting PN junction, think of it as the filament in a 'regular' bulb), a focusing lens (call it a director), and a reflector. I find it easier to think of the light emitted as if it was a radio antenna:
The rated light cone angle that we show on our specifications is what in antenna terms would be called the 3db point. The term db, or decibel, was named after Mr. Alexander Graham Bell, who (besides the telephone) did a tremendous amount of research on sound and sound perception by people. One decibel (1/100 of a bel) is the smallest difference in sound that a person can perceive. 3 db (which loosely corresponds to an 'S' unit on your HF radio if you have one), is the smallest difference that you will actually notice. The scale is logrithmic and to increase (or decrease) 3db in sound, requires DOUBLING (or halving) the input power.
A person's perception of light is similar. A rated light cone angle doesn't mean that there is no light after 15 degrees off center, it means that there is LESS light.
We don't recommend this (you could damage your eyes), but IF you stared straight into a single Bebi marine LED, you will see these properties illustrated. For a total of 30 degrees, you will notice little change in light intensity. After that, you can start to see a large difference, until you get to the point that you see what are called 'side-lobes'. The shape of the large forward lobe indicates that when the change occurs, it is very abrupt. Using a large number of LED's as we do is harder to manufacture, but, from a distance, the over-lapping cones of light yields the same apparent brightness (intensity), regardless of viewing angle.
Comparing LED Output to Incandescent and Flouresent LightsComparing light output between incandescent, flouresent, and LED products is like trying to measure apples and oranges because their methods of generating light are so much different.
Imagine yourself in the exact centre of a clear plastic ball. If you shine a flashlight outwards, you will only have a single dot of light on the surface. Now, pretend that you can ignite a ball shaped fire in the middle of the ball, and the entire surface of the ball will then be lit up. (In engineering terms, the surface sections are called stearadians, with 4π stearadians per ball, and the light measured is lumens/steradian).
The output of a single LED is greater than that of a 25 watt incandesent over the area of the ball's surface that is projected by the LED. The balance of the ball has almost no light from the LED, and a large amount of light from the incandesent.
If you continue with all of the arithmetic, and you add in (in the case of the Masina Malosi®) 29 more LED's in a circle for a total of 30 LED's, you will get a band of light with 'bumps' (from the curve between the blue '*'s on the graph above) on the inside of the ball. Within this band, the LED's are again much brighter than the incandescent bulb, but again, the area outside the band is much dimmer than the light from the filaement bulb.
Another graphic representation would be the petals of a daisy. Each petal would be equivalent to an individual LED, the maximum light intensity the furthest length of each petal, the minimum the overlap of the petals, and the average light intensity being halfway between the two.
Just to make things even more confusing, incandesent bulbs are measured in mean spherical candlepower. This implies that the light source is a pin-point, (the 'ball of fire' above) but it can't be in real life. Especially in the case of vertical filament bulbs, the light source is a long line, so the light output isn't a sphere either, but a doughnut shape (torus).
And, of course, component age and the environmental temperatures all effect the output of both devices, but I'm trying to keep this simple! (A sudden wave hitting the light on a boat, will instantly knock out the filiment bulb, but the LED will keep shining).
Another factor is battery voltages. An incandesent will produce light from a volt or two up until it blows up from overcurrent, but the light output will also constantly vary. With our current regulator, the light produced is the same between a system voltage of 11-16 volts. Below around 10.5 volts (this varies a little unit by unit), the LED light will begin to get dimmer, until it cuts completely off, somewhere in the region of 7.5 volts (again, unit variences, but this is a 'safe' number to work with).
Now, on a running light, you aren't all that concerned about light shining upwards or downwards, so a lot of the light generated by an incandesent bulb is 'wasted', but it is where an LED shines (pardon the pun).
A Graphic Comparison Between LED Lights, Halogen, and Incandescent Bulbs
In order to give the consumer a rough visual feeling of comparison between our lights, and also between different common low voltage DC lights, we created the following plots. We regularised the comparison by assigning a white (255/255) colour as the most intense, that emitted directly under our Pure White Kalokalo (0˚ axis), 38 lumens, and black (0/255) being the least intense, emitted 25 degrees off the 0˚ axis.
The only problem with this approach is that the light is far from 'black' where the outer ring is, as you can see by the accompanying numbers, there is light there, it is just a reduced volume of light.
The radius of each plot is based on a lamp height of 1 meter (39.6 inches), with each ring representing 5˚, so that the outermost ring is a 25˚ radius. The numbers have also been 'regularised' to how your eyes will perceive the differences by logarithmic scaling.
You will note that the center of the light pattern is considerably brighter than either the 10 watt halogen or the 25 watt incandescent bulb, but the brightness drops off sharply as you move off the center axis. The numerical average of all of our LED spot and reading lights are greater than the 10 watt halogen, but less than a 25 watt incandescent.
In addition, to be fair, these pictures are roughly what you would see if the halogen or incandescent lights had no light output out of the backside of the bulb. Many have reflectors which will increase the amount of light seen with the 25˚ cone, but by an unknown amount.
You'd think it would be simple, life isn't so kind.
In CIE 127:2007, 'Measurement of LEDS', there is one measurement of light intensity (candella), and two fundemental measurements of light volume commonly used with LEDs, lumens the total amount of light integrated over a spherical surface, symbol θv, and partial flux angle.
Candella is the intensity of light striking a single point. Beamwidth, or cone angle, is used to describe the angular distance from the 0˚ axis where the intensity has decreased 50% (maximum candella/2).
As above, total luminous flux is a spherical volume of light
Partial flux is the luminous intensity of the LED measured solely across the emitted angle of the LED.
Although partial flux can give a more accurate representation of the amount of light you would expect to see across a surface, it is easy to 'play the specs game', quoting a partial flux number as if it was a measurement of total luminous flux. For example, if we were to quote the partial flux of a 30˚ NSPW500DS, it would be 12 times higher than that of the total luminous flux; 42 lumens (partial flux) instead of 3.5!
All of the LEDs manufactured by Nichia are rated in compliance to CIE 127:2007, eliminating any sort of 'games' being played with the specifications. Where we have converted from candella to lumens, we have used the typical midpoint between maximum candella and the 50% down point (75% of maximum), then converted as below. This method has the advantage of disallowing the 'wasted' light emitted from the side and back lobes of the T1-3/4 package for a 'more fair' comparison of our products to others.
Another point of misunderstanding is confusing the physical LED package design with efficiency. At this writing, the maximum commercial (instead of laboratory) limit for LEDs is in the 100 lm/watt range. Manufacturer's of LEDs produce LED devices in a variety of case types. Some surface mount LEDs have multiple LEDs within them, thus, from a single device they achieve a higher output of light, however, the lumens/watt does not change a whit.
Yet another 'game' revolves around what is termed 'binning' within the industry. Production of high output LEDs is still in its infancy, and the actual behaviour of the assembled LED is not known until after it is tested (although it's getting better). As a consequence, after a lot of LEDs are produced, the LED manufacturer will 'bin' the lot based on the test results. For example, on the Nichia NSPW500DS, we can choose from two of three intensity bins (we always buy from the highest), although we could save money buying only from the lowest. Again only for illustration, the Nichia NS6W183, an SMT device incorporating six LED 'dice', has six luminous bins, varying from 245 lumens/package to 170 lumens/package, three forward voltage drops, and six different 'shades' of white. One of the best known of the competetion, Phillips-Lumiled, has ten luminous flux bins ranging from 50 to 180 lumens, six forward voltage drop bins, and eight 'shades of white' bins.
Thus, if you read an advertisment about an OEM using a brand of LED, or even a given model number, unless the advertisment also supplies the bin information (like we do), it really doesn't give you a lot of information as to the performance that YOU can expect to see. This is especially true if you try to 'roll your own'; if you purchase in lots of less than 10,000 devices, what you get will be 'pot luck' as far as which bin the LED actually comes from.
For the OEM (such as Bebi), there is a very large advantage to using SMT devices compared to a T1-3/4 (a 'regular LED'); it lowers the assembly cost of a final unit as it can be robotically assembled faster. In addition, more devices can be placed into the same footprint.
The downside to this, which may effect the consumer, is that thermal management is now very critical, and heatsinking becomes a requirement, or else the product will prematurely fail, esp. if located in a fixture with inadequate air circulation.
If you take the time to download the product data sheets of the reputatable LED manufacturers, you will observe that all of them do not recommend their SMT devices for exposed locations, as they cannot be weatherproofed and meet the necessary thermal management requirements.
Thus, we continue to use T1-3/4 package devices in our products. We will also continue to use lumens or candella to describe the output of our lights, as the terms are more universal, and are less subject to abuse.
Tabulated Data About Our Lights
Bebi Electronics uses exclusively Nichia® LEDs. Effective October 2007, we are using their NSPW500DS and NSPW510DS LEDs. These devices have a 29% greater output than the NSPW500CS LEDs which we had been using. Due to their greater output and our manufacturing techniques (read about this here), we are now mixing the 30˚ and 50˚ in our product. This allows for a wider spread of light, without any loss in the center of the light pattern.
During November 2008, we also began adding to our selection the NSPE500DS AND NSPE510DS, true 'warm white' LEDs. These devices use the same 'chip' (PN junction) chemistry as the 'pure white' NSPW devices, so they are equal in light volume and intensity, but have virtually the same poly-chromaticity (colour) as a 10 watt halogen bulb. The exception is the Kalokalo Katakata, where we've changed the mix of 500 and 510 devices to produce a more even light gradient, which many people prefer for area type lighting.
Based on the amount of drive current and the different outputs for each intensity ranking, as follows is the raw data from which the above plots were created, using the 'typical' values shown.
Our lumens (luminous flux) specifications were integrated from the output in candella, integrated using the formulae of Total Luminous Flux Fv=Iv*(1-cos θ/2)*2*π.
COLREGS applied to LED lightsThe short answer is that our Masina® and Owl® series of lights meets the COLREG requirements for vessels less than 20 meters.
For the long answer...
The distance a light can be seen is estimated in Annex I Section 7 paragraph (A) of COLREGS. It says:
As mentioned in other pages, luminous intensity cannot be summed. An example would be if you had two
cars with a maximum speed of 100 kph each, if you bought both cars you would not be able to travel
at 200 kph. Therefore, any claims of intensity based on multiplying the number of LED's by the individual
intensity of each LED is, to use one polite word instead of two rude ones, bogus.
Calculating the Bebi Electronics Owl® and Masina® series of lights using the new 'DS' LED, takes a bit of arithmetic. The complexity is due to electronic devices being sold by specification ranges, the distribution not being a 'z' (normal, or 'bell shaped curve) and hence the average being substantially different than the median, or typical value, and the rest because of the 'flower pedal' shape of the light cone as above.
Thus, for a single row of LEDs with a 30˚ cone angle, each individual light produces in excess of 9.3 cd (candella, 1/1000 of which is the millicandella or mcd), an average of 14.0 cd, a 'typical value' of 16.2 cd, and an expected maximum of 18.6 at the 0˚ axis. As you move off the 0˚ axis, the intensity of the light will decline. With 15 LED devices, the absolute minimum of this row would then be 2.8 cd with an average of 4.2 cd and a 'typical' minimum of 4.8 cd. Taking this ring as a whole over 360 degrees, the expected average minimum would then be 6.0 cd, the average 9.1 cd, 'typical' 10.5 cd and the expected maximum would be 12.1 cd.
As can be seen from the quoted COLREG paragraph above, this exceeds the required minimum of 4.3cd for a 2 mile white light
Because of the reduced quanity of LED's in the circumference of the Beka® series of lights (12 opposed to 15), there is a larger drop in the average light intensity about the perimeter of the light as the absolute minimum at has an absolute worse case in the order of 1.9 cd, an average minimum of 2.8 cd, and the upper minimum range of 3.7 cd. Integrating the parts into a whole circle, we then see a 5.6 cd minimum, 8.9 average, and a maximum of 11.2 cd, once again exceeding the 4.3 candella as specified by COLREG as a 2 mile light.
With these numbers, it verifies what some of our customers have reported in 'real life'. If you would like to read what users have found in practical applications, S/V 'Amulet' comments here, and also S/V 'Capensis' here.
In the case of a steaming light, which COLREGs calls a 'Masthead' light, the requirement for vessels under 20 meters is 3 nm or 12 cd. In this case, our Volasiga® light appears to meet the requirements, as each row of 30˚ cone angle lights overlap so the greatest off axis distance is 6 degrees. Using the above data about the 0˚ axis, but derated to the 6˚ point, yields minimums of 5.8 cd, average 8.4 cd, typical 9.7 cd and the maximum over the 6˚ range of 11.2 cd.
Again integrating into a arc, yields an absolute minimum of 7.55, avg 11.2, typ 13.0, max 15.0. As the typical (NOT maximum!) exceeds 12 cd, it again meets the requirements.
In a replacement light, which goes into a fixture of unknown brand of an unknown age, Bebi cannot predict the transmission losses through the lens, merely that losses will occur.
We do find of interest the last sentence in the paragraph which begins 'NOTE'. Based on how you (or an insurance company) could choose to read it, you could be held to task for having lights which are too bright for your vessel length! That the means of limiting light intensity is specified is an indication that somebody has been held to task for this violation.
COLREGs and Vertical Angles
In the case of vertical angles, our lights do meet paragraphs 10(a) and 10(b)(i) of required minimums of (2 nm) for 'stand alone' running ('side') or masthead (NOT tri-colour, which is an 'all round' light, and is NOT to be used under power!) lights of power and sail vessels under power of ±5˚ about the horizon with no less than a 50% reduction in output. At the ±5˚ mark, the degradation of our lights is no greater than 10%.
As a tricolor light for vessels of less than 20 meters under sail, our Masina Malosi® of light has a second row of LEDs to meet the requirement of 10(b)(ii) of no less than a 50% reduction within a 25˚ about the horizon axis.
On the row of LEDs with the 50˚ LEDs, each individual light produces in excess of 2.6 cd, an average of 3.8 cd, an expected typical of 4.2 cd and an expected maximum of 5.1 at 12 degrees off the 0˚ axis. As you move towards the 0˚ axis, the intensity of the light will decline about 10%! However, at ±25˚, the the degradation will have a minimum of 1.0 cd, an average of 1.5 cd, a typical of 1.68 and the expected maximum range would be 2.0 cd. Integrating the ranges into a circle as above, the absolute minimum is 1.8 cd, average of 2.7 cd, typical 4.2 cd and a maximum of 3.6 cd. As the typical value is again in excess of 2.2 cd, we believe that our LED tricolor lights meet the vertical angles required by, and thus comply with COLREG.
As there are no vertical angular requirements specified for anchor lights, our assumption is that they are the same as paragraph 10(a) and 10(b)(i) of rated minimum intensity ±5˚ about the horizon.
In the case of our Masina Kuaka® and Masina 'Afa®, if they are being used exclusively (without a tri-color) as side lights on a sailboat under sail, they will not meet the vertical angle requirements. However, if they are installed inside an existing lens as they are designed to be, we believe that they are likely to meet the requirement due to the vertical angle disbursement of the lens itself.
COLREGs and Color Requirements
As can be seen below, the white colour co-ordinates required by Annex I paragraph 9 of COLREGs are very skewed to the 'red' portion of the color spectrum instead of the 'equal energy' region of C0 (likely so that low voltage dimmed incandescent bulbs can meet the requirement!) are also met by the NICHIA® LED's at ambient temperatures between -30˚C and +30˚C. Above +30˚C, the light may become too 'blue' to meet the COLREG definition of white light, in spite of it still being within the C.I.E. definition of white.
The color skew towards the blue spectrum seen from the 'white' LED's is a result of the light from the PN junction actually being blue, instead of white, which is combination of all frequencies of light. The emitted white color is from the phosphor coating on the reflector which 'bends' the blue into white. If there is a concern of a 'blue-ish' cast thru a green lens, you can see that the allowable definition of 'green' extends significantly into the blue portion of the spectrum.
The C.I.E. color chart shown was developed by Dr. Curie as an attempt to define how humans see color: Not just wavelengths of mono-chromatic light (indicated on the chart by the values expressed in nanometers around the periphery of the chart), but also in reaction to apparent light intensity (sometimes called 'brightness'). Color perception is very, very individual. For example, the author of this paper has great difficulty seeing most shades of green as anything besides grey or black. The chart only attempts to predict what MOST people see, it will not predict what YOU will see.
IMPORTANT-PLEASE NOTE: After we were reliably informed (27 Sep 08) that our Pure White Masina Malosi lights, when fitted into an AquaSignal Tri-colour combo manufactured (in likely) 2007 or 2008, did not render the green to be within COLREG spec (by extention, we assume that this affected their individual green running light fixtures). To counter this, in November 2008, we introduced our Masina Malosi Vevela with the NSPE devices.After testing, we are content that this corrects the colour shift, and the remaining colors remain in specification. Compare the colours below with that of the 1931 Chromaticity diagram above).
On other brands of fixtures, and earlier Aqua-Signal fixtures, of course, remain unaffected, and can use either varient of the light.
As stated elsewhere on this site, our lights are not certified by any agent of any government, and, are thus, used at the customer's own risk. With the above mentioned exception, we nevertheless believe our lights to be in compliance with COLREG.
The Expected Lifetime of LEDs'Regular' (incandescent) or halogen bulbs have one failure mode - open.
Fluoro's, in whatever shape or form, have essentially three failure modes: Bad ballast (lamp won't ignite), 'bad' bulb', or marginal bulb (flickers or dim).
On LED lighting products, they are closer to fluoro's in their available failure modes. LED's are not new devices, the technology to produce a 'white' LED is new.
First, the CURRENT REGULATOR (not voltage!) can fail, similar to a fluoro ballast.
Second, the PN junction in the LED itself (think of it kind of like a filiment in an incandescant bulb), can fail open like a 'regular' light bulb, or it can fail short (not like a regular light bulb). A quality LED, IF it is assembled and regulated correctly, should have a life in excess of 100,000 hours.
Third, and really of greatest importance, is that the phosphor reflector that turns the blue LED into a white light degrades over time until you can only tell that the light is on or off. Consider the reflector phosphor as having the same principle as a screen saver on a CRT monitor: The longer and harder (brighter) that you energize the phosphor, the quicker that it becomes useless. The monitor will still turn on and work fine, you just can't see the picture on the screen!
The period of time until this occurs, which is a function of initial LED quality, and heat, is the real determinant of the useful life of the product.
Some of the lesser quality LED's have a 50% phosphor life of around 1k hours. Not surprisingly, these are the cheapest. Likely not a shock, the longest lived reflectors are the most dear.
Effective May 2006, Bebi Electronics, Ltd. has switched to using exclusively Nichia® brand LED's. This frankly costs a packet, but we have found that their products have the longest lived reflector phosphors on the market. In November of 2007, we have switched from the Nichia CS series LED to the DS series, which has an even greater lifetime and produces 29% more light than the CS LEDs.
However, even the finest LEDs can still prematurely fail if they see too much heat or are assembled incorrectly.
First, and easiest for the engineer, is to only sell units in Antartica. Dropping the temperature to -50˚C will extend the life of the LED phosphor for a century. Barring that, reducing the current (which is measured in amps) to the LED also reduces the heat and extends the phosphor, but produces less light than running the light at the rated 'safe' maximum. By using an active current regulator, we maintain a drive level around 50% less than the rated maximum.
By using 'lamp-type' (roundish) LED's we avoid the need for external heat sinks which are required with SMT devices. By allowing air to circulate around the lens and by keeping the internal lead lengths fairly long, we wick more of the heat away from the PN junction into the surrounding air.
A common failure mode, especially prevelent on 'lamp-type' (round-ish) LEDs is the manufacturers failure to abide by the LED manufacturer's recommendations for the soldering distance between the base of the lens and the solder joint (3mm is recommended minimum, yet many are flush mounted on boards and soldered ~1.5mm from the lens). Improper soldering like this allows the epoxy to soften, and the lead will stress from the heating and cooling, 'breaking the filiment'.
Yet another failure mode due to poor manufacturing techniques is the failure to match series strings. Since each (white) LED has a forward voltage drop of around 3.5 volts, you can get sexy and put 3 in a string for a 12 volt nominal system, and 'waste' less energy in the regulator.
The fly in the ointment on this is the wide spread of forward voltage drops within a batch of LEDs.
If the 'weak sister' in a batch of LEDs is mated with a 'strong brother' (sounds not only sexist, but incestuous!), the weak sister will pull more current than planned in the string and may fail prematurely as a consequence.
One of our (repeat) customers wrote of us "Your lights are the best. They don't fail." This isn't an accident and it isn't rocket science. It's simply a matter of good manufacturing techniques combined with sound engineering. Scout around on the internet and find out just how poorly our competitors fare in this.
The net effect of these changes are that our Masina® series of lights have an expected mean 'useful' life of 5 years of continuous use before they fall below the COLREG two mile intensity minimum. Assuming that running lights are only carried during night, this amounts to 10 years of daily use.
Disclaimer on USCG CertificationIn spite of our Masina® series of lights apparently meeting or exceeding COLREG requirements for a 2 nm light, you will note that we do not offer certification by Good Housekeeping, the USCG or any other agency. The USCG does not certify anything. Independent laboratories test light fixtures, with a light source supplied by the submitting party. The results are then submitted to the USCG who may issue a certification. Only light fixtures, with a light source included by the manufacturer, can be submitted. If you go to any catalogue selling incandescent, halogen, fluorescent, or dilythium crystal light bulbs, you will not see a single USCG certified light bulb, regardless of manufacturer.
What the USCG has to say about certification is found in 33CFR Sub-Part M, paragraph 183.801-803:
As you can see, the burden of certification is solely on the manufacturer/distributor of boats, not the consumer,
although you still MUST comply with COLREG. We will not attempt to address requirements of the 50 individual states
and the territories!|
If you have insurance and you are involved in a collision at night, your claim may be dis-allowed if you have a non-OEM light bulb, whether it is an LED, incandescent, halogen, or fluorescent, in the fixture, regardless of the real reason for the collision.
Having said that, Lexis/Nexis searches have not shown any judgements based on type of light source, nor have there been any in Canada, as per http://www.admiraltylaw.com/collisions.htm. There have been collisions, where lighting has been cited in a judgment, but only because they were either not carried, or not lit.
Are you more likely to be involved in a collision because you need to save power at night and either don't turn on your running lights, or just use an anchor light or an illegal strobe*? Of course. If your incandescent lamp burns out at night, or, especially on powerboats, a cold wave hits the hot fixture, how likely are you to actually change the bulb in the middle of the night? These are the types of incidents which have led to judgements against a boat owner. The redundant circuit design, which is on all of our lights recommended for being fitted into vessel navigation lights, greatly lowers the chances of such an event from happening.
And yes, if you have bought a Bebi light prior to our change over, we will still stand behind our Limited Lifetime warantee (it's good for as long as we live!).
Have fun and stay safe with Bebi Electronics.