LED Lighting Comes of Age - Part 3

Main Index

Lamps & Energy Index

I have stated before that I firmly believe that the idea of LED 'replacement' lamps is stupid.  There are many dedicated fixtures now available where the fitting itself is designed specifically for LEDs, and there is no provision for changing the 'lamp' as such.  In many cases, we now see fittings such as streetlights where the LEDs are in modules that can be replaced if needs be, and the most vulnerable part of the fitting (the power supply) can be changed easily.  LED modules can usually be replaced as well, so the housing can be used for a very long time - provided spare LED modules and power supplies are available of course.

I recently saw a suggestion that incandescent lamps should not have been banned, just the sockets (Edison-screw and bayonet styles) they are used with.  This would prevent the sale of new light fittings that used interchangeable lamps of any kind, and also eliminates two rather dangerous designs that would never obtain approval if anyone were to suggest them today.  Edison-screw and bayonet lamp holders will easily allow the standard 'test finger' used for safety tests to make contact with live mains connections, and they are only permitted under 'grandfather' provisions in most safety standards.

The LED revolution is really starting to show what happens when people start throwing money at problems.  LEDs are now available that can provide the same light as a 75W incandescent lamp - from a single LED! There are also LEDs that can produce as much as 160 lm/W at 1W (350mA), which makes them the highest efficacy white light source currently available.  At the time of writing, the highest claimed luminous efficacy is 180lm/W.  Philips introduced a 'light engine' some time ago that uses high efficiency 'royal' blue LEDs, and has the colour shift phosphor in a separate sheet of plastic that also acts as a diffuser.  These are also known as 'remote phosphor' LED lamps/ light engines, and have made some inroads into the market.  The range of LED lighting products increases daily, and the various industry newsletters that I receive have something new on the topic at least weekly, and sometimes more often.

I really like the Philips approach ...  one reason for separating the LED and the colour shift phosphor is simple - the phosphor is a known cause of white LED failure due to molecular migration within the LED housing.  This effect is worse at elevated temperatures, so this is a good reason for keeping LEDs as cool as possible.  Of all the issues with LEDs (both real or imagined), temperature remains the biggest single problem.  High temperatures reduce both light output and longevity, and there is something of a scramble as manufacturers strive for the most thermally efficient packages and mounting methods.

There are two issues now facing LED lamp manufacturers ... keeping temperatures low and developing power supplies that are as rugged as the simple lighting systems that are being replaced.  Considerable effort is being put into both areas, and new solutions are announced almost weekly.  Unfortunately, heat can only be removed successfully with a heatsink, but this is being integrated into the light fitting itself for many new luminaires.

It has become very obvious of late that many people (including some of those selling lighting for a living) have grave misconceptions about the common MR16 lamp.  It seems to be that most people associate MR16 with the 12V bi-pin lapms used in downlights.  However, the term 'MR16' only refers to the type of reflector and diameter - it has nothing to do with the input connections.

The description 'MR16' only indicates that the lamp has a Multi-faceted Reflector (MR) and is 16 x 1/8" diameter - i.e. 2" or 51mm.  There are actually several different sizes of MR lamps, with the most common being MR16 and MR11 (35mm diameter).  Of these, the GU5.3 base is the one that is usually associated with these lamps, and that's the standard bi-pin arrangement that is so common.  There are two pins, each 1.45 - 1.6mm diameter and 7mm long, spaced 5.33mm apart.  These lamps are always low voltage, with 12V AC being the vast majority.

Likewise, GU10 has nothing to do with the lamp itself, but refers to the base.  A GU10 base is a two pin parallel bayonet fitting - the pins are 10mm apart and project from the rear of the base.  The pins have a rectangular 'mushroom' head (5mm diameter x 3mm high) that interlocks into the socket with a twist.  GU10 lamps (including MR16) are always mains voltage, either 120V or 230V AC.

Figure 1 - MR16 Lamp, LED Replacement, With GU5.3 Base (Left) and GU10 Base (Right)

It is very important to understand the difference, and to ensure that the correct terminology is used.  Although there are LED lamps that are classed as MR16, they are no such thing, as there is no reflector of any kind, and the one that isn't there obviously can't be multi-faceted.  The term is generally used to indicate that the lamp will fit into the same socket as a true MR16 GU5.3 12V lamp, but it's actually wrong and misleading.

The LED lamp shown above is the same diameter as the MR16, but it's quite obviously not an MR16 - even though it will fit into any MR16 gimbal (commonly & incorrectly spelled gimble).  Naturally, it won't fit the GU5.3 socket either, and it runs from 230V, not 12V.  While you will see similar lamps referred to as MR16, this is an error - it's no such thing strictly speaking.  Unfortunately, the term 'MR16' will continue to be used, both for LED and CFL downlights, despite the lack of a multifaceted reflector.  It would be more appropriate to refer to them as 51mm lamps, and specify the base type - at least it would make sense that way.

I'm certain that I'm not alone in seeing LED lighting systems that have failed in service.  Assuming the LEDs are kept cool enough to prevent failures, the next most vulnerable part of the system is the power supply, and this is going to cause a few users some grief until the many manufacturers come to grips with the reality of electronic power supplies.  Of course, LEDs are also vulnerable, especially if pushed to their limits and when used in high temperature environments.  I've seen 100W LED modules pushed to ~130W, but even with a very good heatsink they only lasted for about 3 years - admittedly with fairly heavy usage.  Had they been run at perhaps 90W instead, I'd expect a very, very long life.

There are two things that will kill any SMPS (Switch Mode Power Supply) - high peak (spike) voltages and heat.  For transient peak voltages, many manufacturers think that all they need to do is include a MOV (Metal Oxide Varistor) and nothing bad can happen.  This is a rather naive approach, and doesn't consider a 'normal' failure mode of MOV devices.  It's very common that when a MOV fails it becomes close to a short circuit.  I've seen all sorts of products where MOV protection devices have literally blown themselves to pieces.  In many cases, all that's left is vestigial component leads and a burnt spot on the PCB.  In some cases, a MOV can fail explosively and just remove itself from the circuit.

The following circuitry (usually a power supply) then continues to operate, but with no over voltage protection at all.  The first mains disturbance that exceeds the failure threshold of any part of the circuit causes it to cease operating (best case) or a total meltdown (not uncommon).  Once the MOV has failed, the next time there are high voltage spikes on the mains, the power supply is unprotected and will probably fail.  It is vitally important that if (or when) the supply fails, it doesn't destroy everything else.  It's very easy for a 10 cent part failure in a power supply to destroy $50 (or more) worth of LEDs if there is no protection against supply failures.

Any electronic power supply can also fail due to heat.  Electrolytic capacitors are commonly rated for a maximum of 105°C, and manufacturer data typically claims 2,000 hours expected life if operated at maximum current, voltage and temperature.  Like all electronic parts, electrolytic capacitors prefer low temperatures, and if the ambient temperature can be kept below 40°C the entire power supply can be expected to give a long trouble-free life.  It is accepted [1] that the life of electrolytic capacitors doubles for each 10°C temperature reduction.  Using a higher voltage part than really needed also helps, but temperature is the #1 killer of electrolytic caps.

Naturally, anything can fail at any time and nothing can be expected to last forever, but tempting fate is, historically, a very bad idea.  Some supplies I've seen use low-value high-voltage electrolytic caps (e.g. 1µF/ 400V), and locate the cap next to parts that get hot.  This type of capacitor is notoriously unreliable at the best of times, and using them when other options are available is asking for trouble.  The expected life will rarely be better than 20,000 hours, which is very poor compared to the expected life of the LEDs.  I've analysed several failed fittings that use 1µF/400V electros, and in every case the LEDs were as good as new, and only the cap had failed.

It doesn't help that people are rather perverse.  Many will cheerfully (?) pay vast sums for a light fitting, but then expect the actual light sources to be cheap.  Light fittings that are a complete disaster in terms of efficient use of their light sources are common, presumably because they make a fashion statement.  I don't profess to understand this - to me, any product is first and foremost about function, and I won't compromise function for fashion.  To many others the reverse is true, and if it doesn't have 'the look' (whatever that may be) then it doesn't stand a chance.  There is no reason to make LED lights as ugly as sin, but some makers seem to strive to achieve just that.

Heat remains the biggest obstacle to be overcome with LED lighting products.  Heatsinks are usually not very attractive and are difficult to hide, but there's no point having the greatest looking lamp in the world if it kills expensive LEDs in a matter of a few thousand hours or requires the most power-hungry and inefficient light source available, then throws most of the light away anyway! It might look 'nice', but at what cost to the end-user and the environment?

Killing LEDs is easily done if the design doesn't allow sufficient cooling.  At present, there are really only two main options for LED lighting.  One (and it's quite common) is to use a large number of low power LEDs in a tube or other fitting.  Anything up to 300 LEDs is not uncommon, with each operating at a current of about 20mA.  This keeps the power to each LED down to about 60mW (typically), but if there are 300 LEDs that's a total of 18W.  With high brightness LEDs, you get a lot of light.  Because it's distributed over a large area (perhaps 48cm² for a typical tube light), this arrangement is relatively easy to keep cool, and no special steps need to be taken provided there is some airflow around the tube itself.

The other option is to use a small number of high power LEDs - for the same power as the previous example, we might see 6 x 3W LEDs used, again for a total of 18W.  The space occupied by the LEDs themselves is quite small, but we still have to dispose of close to 18W of heat.  To make matters worse, the heat sources are highly concentrated - just 6 small points, each no more than about 4mm².  This has taken us from a nice easy option to one where some serious feats of engineering are needed to keep each LED cool enough to ensure longevity.  This is not a trivial problem, and there are few solutions presently available.  One approach is to allow the LED chips to operate at higher temperatures without premature failure, and the Philips technique of separating the LEDs from the colour change phosphor is likely to make this a viable option.

Only a few years ago, it was considered quite an achievement to create a 1W LED, and one of the first was the Luxeon Star™.  Now there are many high power LEDs, ranging from 1W (still popular) up to arrays rated at up to 150W or so.  IMO these are not the most sensible idea, because it's too hard to get the heat out of the LED junctions.  3W and 10W LEDs are common, and the original star pattern created for the Luxeon Star has been adopted by many manufacturers.  The heat problem remains though, and is only partially mitigated by the higher efficacy available now compared to even 5 years ago.

Figure 1 shows a 10W LED, in this case it already uses LEDs in series/parallel internally.  There are 9 LED chips inside the case, and these are arranged in three parallel strings of 3 LEDs.  The quoted voltage is between 11 and 12V, and the current can range from 830mA to 900mA.  The biggest issue with this LED can be seen through the phosphor.  The 9 small squares are the LED emitters, and each will have to dissipate about 1W.  This may not sound like much, but this is from a tiny piece of silicon, and the heat has to be removed with the minimum thermal resistance possible.

Figure 2 - Typical 10W LED

To make matters worse, there are nine of these tiny pieces of silicon, all within less than 1mm of each other, and all dissipating ~1.1W each.  To be able to maintain a respectable temperature for each die means that the thermal resistance between the LED array backing and heatsink needs to be as low as possible.  The external heatsink has to be extremely efficient, and unless it is cooled using a fan (or is water cooled - something I've not seen thus far), it needs to be rather large.  There is a complete article on my site that looks at heatsinks and maximising thermal transfer (see Heatsinks for the full story).  This is not a trivial subject by any stretch of the imagination, and it's made worse when the product is subject to the whim of interior designers, as is the case with lighting.  A lamp that has the desired look will invariably be chosen over another that is designed for function and longevity, and it can be extremely hard to make a product that can combine the all parameters successfully.

When I tested the 10W LED shown above, it was immediately apparent that running it at maximum power is not worthwhile.  The graph of light output vs current is never linear with LEDs, and there is a noticeable reduction of luminous efficacy as the maximum is approached.  At 500mA, the 10 W LED is running at about 5W, and gave a reading of ~3,000 lux in the quick and dirty test that I ran.  Increasing the current to 1A didn't produce the expected 6,000 lux - I managed about 4,500 lux.  By operating at half power, it's easier to keep the LED cool, and the reduction of light is not as great as expected.

If you've never tried it, disposing of a measly 10W sounds as though it would be fairly easy.  It is easy under two conditions - use a nice big heatsink and provide plenty of air circulation, or allow the device to get hot.  As a rough guide, the thermal resistance of a heatsink is (very roughly) 50 / √ Area in cm².  On this basis, a piece of aluminium 50mm square will have a thermal resistance of about 7°C/ W, so a 10W LED will cause the temperature to rise by 70°C.  Needless to say, this is much too hot.  (Remember that both sides of the heatsink are usually exposed to the air.)

To maintain the temperature of the LED at 10°C above ambient, you'd need a heatsink of 1°C/ W, which means a radiating area of 2500cm² - a 350mm square of aluminium for example, which also need to be thick enough to distribute the heat evenly.  There can be no doubt that this will be difficult to hide in a designer luminaire.  As you attempt to keep the LEDs cooler, the heatsink size grows out of all proportion, so high temperature operation is the only likelihood for usable light fittings.

As noted above, the PSU (power supply unit) is often the weak link.  The range of SMPS that can be used with LED lights is increasing all the time, but the vast majority are used in manufactured products and are not available separately.  Many commercial supplies look just like the ones you can buy, but they are different.  Almost without exception, when you buy a power supply from your normal supplier, you get constant voltage; 5, 12, 15 or perhaps 24V output.  These supplies are rated for a maximum power, and that determines the output current.  For example, a 24V 5A power supply is 120W.

LEDs are best driven from a constant current power supply, and a constant-current supply may be rated for 120W and 6A output, so it's immediately apparent that you can't have maximum current at the maximum voltage the supply can provide (typically it might be around 24V for this type of PSU).  The voltage will change depending on the load, and this makes the power supply design and specification slightly more irksome than you might expect.  A string of 4 high current LEDs might have a voltage of 14V, and if these are 3A LEDs there will be two strings in parallel.  Eight LEDs in all, in two parallel banks of four LEDs.  We'll assume 3.5V across each LED.

Because the voltage across each individual LED is never perfectly matched to any other, it is was common to include low value resistors in series with each string of LEDs to balance the current.  We'll assume 0.2 ohm resistors for the time being.  At 3A through each, we lose 1.8W so the total loss in the resistors is 3.2W.  This is heat that we have to dispose of too, and preferably without it raising the temperature of the LEDs, the heatsink or the power supply.

Each LED + resistor string now has a voltage of 14.6V at 3A, or 42W of LEDs and 1.8W in each resistor (87.6W total).  As the LEDs get hot their voltage falls, so the power is also reduced.  It can be a fine balancing act to ensure that the power supply and LEDs are well matched so there is minimum over-specification of the power supply.  It is also wise to incorporate some form of protection when high power LEDs are driven from a constant current power supply.  If one of the strings described above becomes open circuit, the power supply will do everything it can to force its rated 6A through the one remaining string.  Needless to say, LEDs won't last very long being run at double the rated current.

A more recent trend is to use matched strings of LEDs so that the 'ballast' resistor is not needed.  The majority of new LED lamps do not include any resistance in series with paralleled series strings of LEDs.  While it might seem unlikely that the series/ parallel strings can be matched closely enough to ensure proper current sharing, those I've looked at are almost perfectly matched at any current.  Elimination of any series resistance improves overall efficiency because there is no external power loss.

To a significant extent, it's fair to say that very few LED luminaire manufacturers have included sufficient protection to ensure that the LEDs are protected against failures in the power supply, or even failure of one LED in a series string.  There is no guarantee that a failed LED will be open or short circuit, but either way the remaining LEDs in the fitting will (in most cases at least) be stressed to the point of failure.  It's unrealistic to expect that anyone would try to monitor the health of 300 small LEDs in a fitting, but it becomes worthwhile if there are only perhaps 6 or 8 expensive high power LEDs - 10W LEDs are not cheap!

As the technology improves, we can expect to see power supplies that are simpler, smaller and cheaper than many of those used at present.  It is already possible to build remarkably simple supplies, but the limitation is that they are not isolated.  This means that they cannot be used with small high power lighting systems, because the heatsink needs to be electrically isolated from the LEDs.  This adds either considerable thermal resistance or cost, because the isolation barrier has to be 100% reliable for safety, yet still be capable of removing the heat from the LED junctions.  Lamps that have an accessible heatsink will almost always use an isolated power supply, which complicates the power supply but simplifies LED mounting.

With electronic devices that are presently available, there are still some real challenges to building a very compact power supply - especially where the output is isolated.  However, things that were simply impossible even a few years ago are commonplace now, and we can expect to see more and more dedicated integrated circuits to facilitate compact, high performance power supplies.  Constant current output is the most desirable way to drive LEDs of all sizes.

The greatest hurdle remains the heatsink.  The sooner manufacturers decide on a common form factor for LED 'light engines', the sooner luminaire designers and manufacturers can standardise their designs to provide the appearance and light distribution users want, but with housings (the luminaire itself) that provides the necessary heatsinking to keep temperatures as low as possible.  A couple of standards exist, and Zhaga is one that appears promising.

I remain a staunch supporter of LED lighting, and continue to think that it really does represent the future of lighting for most domestic and office environments.  However some people in the industry are becoming cowboys, and will try to convince the buying public that LEDs are more efficient than any other light source, will last forever, improve your sex life and cure baldness*.  This is obviously not the case.

*   I'm actually not kidding - Many people will have seen 'massagers' that utilise the "healing powers of light" - a few perfectly ordinary red LEDs that do nothing beneficial whatsoever, but the most miraculous claims are made in some advertisements.

If we assume that reasonably average LEDs (compared to the best and brightest) are generally in the range of 70-80 lm/W, this needs to be compared against other light sources.  To be the true devil's advocate, we'll look at the upper limits for existing incandescent, halogen and fluorescent lamps compared to the median of 75 lm/W for LED lamps.  This is deliberately pessimistic.

It has to be considered that for most traditional light sources, a large amount of the light is wasted.  Fluorescent tubes emit light evenly around their circumference, and in many cases the light that is not in the direction you need is either lost completely or reduced by a significant margin.  A fair and reasonable estimate (although it's often rather optimistic) is that about 50% of the emitted light goes where it's needed, so for the typical case (T5, at 105 lm/W) you can expect about 53 lm/W where you need it.  Naturally, there are luminaires that will be much better (I've seen some), and others that will be a lot worse - I've seen some of those too.  LEDs direct the light where they are pointed, and there is little or no loss because no reflector is needed in the fluorescent fitting (known in the industry as a 'troffer').

In order to equal a T5 with 50% light efficiency (about 53 lm/W x 27W = 1431 lumens), at 75 lm/W for LEDs, we will get the same amount of light where we want it with around 19W.  Both examples have ignored the losses in the electronic ballast, but their efficiencies will be similar, so the 27W fluorescent ballast will waste a little more power than the 19W LED PSU.  This is generally pretty close to the claims made by reputable LED tube suppliers, and the comparison is slightly better for traditional T8 tubes with a magnetic ballast and a troffer that offers reflectance of close to zero because they are often old and dirty.

Troffers that match this description perfectly are found in underground railway stations, many factories and other workplaces, home workshops (using "free" troffers salvaged or scavenged from goodness knows where - I have many of these).  When we consider incandescent lamps in dodgy reflectors (or with none at all), a LED replacement is in a class of its own - even the cheapest and least efficient will murder an incandescent lamp, and if it has an integral reflector will probably beat a CFL easily as well.

Making a comparison between LEDs and 12V quartz halogen lamps is also worth looking at.  A 10W LED is extremely bright - so much so that looking into the LED array even from some distance is painful.  Despite this, they are still well behind a 50W Halogen downlight as one would expect.  Since I was using the LED at 5W, it's simply not possible for it to be 10 times as efficient as a halogen lamp to be able to match the latter in light output.  That would require a light output of well over 300 lm/W (assuming 30 lm/W for the halogen).  For this reason, LED replacements for halogen downlights would seem impractical.

For a LED array to match a halogen downlight at 24 lm/W (a total of 1,200 lumens for a 50W lamp), we'd need around 17W of LEDs to give the same light.  Fortunately, in all but a few cases, no-one actually needs the full output from their downlights, so many either use zone switching or dimmers.  Add to this the extraordinary amount of heat these lamps give off - they have caused a great many house fires in Australia, and no doubt elsewhere as well.  As a result, even though LED alternatives will not achieve the same light level, they are a much better alternative.  Anyone who has been seated directly below a 50W halogen downlight knows just how unpleasant this can be, and the sooner they are replaced the happier I'll be.

When we look at very high power lights, LEDs prove to be less useful.  A 1,000W metal halide lamp can generate perhaps 100,000 lumens, and even if enough LEDs were assembled to generate the same amount of light, the space occupied would be huge, and cooling would be a nightmare.  Disposing of almost 1kW of heat is no easy task - especially if the temperature has to be maintained at no more than perhaps 70°C.  By comparison, the metal halide lamp is quite small (200mm max length), and is fully expected to run hot ... very hot in fact.  The housing, reflector, protective glass and ballast make the fitting somewhat larger, but it's still very small compared to an array of LEDs trying to produce the same amount of light.  Just imagine 100 of the 10W LEDs shown above with all the heatsinking and weather protection that would be needed.  Given that a metal halide replacement lamp is so cheap for the light output, it will be a very long time (if ever) before LEDs are suitable.

Streetlights are an odd conundrum.  On one hand, we assume that lots of light is needed, but this really isn't the case at all.  Many streetlights are much too powerful, and a lot of their light output is often wasted dazzling drivers and/or causing general light pollution.  LEDs are perfect here, because it's easy to target the light so that it illuminates the street, but causes the minimum glare and almost zero light pollution.

The requirements differ widely, but there are common elements.  Illumination of adjoining properties (especially residential) is usually to be avoided, and the light levels are generally quite low.  For residential areas, as little as 1-3 lux is required, and even major arterial roads may only need 10 lux or so.  It's usually cooler at night and there's no sun to heat the LED heatsink, so keeping LEDs cool is fairly straightforward.  LED streetlights of 20 - 60W are quite common, and even a 30W unit provides a surprising amount of light that's well controlled and has minimal spill beyond the designed coverage angle.

Philips Lighting has done a lot of work on this topic, and concluded that LEDs are even a better alternative than high pressure sodium vapour (HPS) - one of the most efficient light sources we have available [3].  The primary advantages of LEDs were very even distribution, minimal wasted light (thus reducing light pollution) and a much lower total cost of ownership due to the long life of LEDs which do not require regular replacement.  Add to this a much better colour rendering index than HPS and a choice of colour temperatures, and zero mercury - it is obvious that LEDs will ultimately become the streetlight of choice.

When you include the low operating voltage this becomes even more attractive.  Even 120V and 230V streetlights generally have a switchmode power supply that's integrated with the constant current driver for the LEDs.  LED streetlights can operate from 24-48V DC internally, so they are ideal for solar chargers and battery operation.  For remote areas where mains power may not be available and a relatively small number of lights is needed, this is a combination that cannot be met conveniently (if at all) with traditional light sources.  Incandescent lamps have been used, but can only be very low power and provide minimal light or the solar charger and battery bank becomes far too expensive to consider.

The maximum possible luminous efficacy of any light source is 251 lm/W for white light.  There is no LED that comes close to the theoretical maximum, although LEDs now exist that can manage 160 lm/W in normal operation.  It has to be assumed that's at a junction temperature of no more than perhaps 50°C since this was claimed to be an attainable light level.  In general, you should assume 25°C if the temperature isn't specifically stated.  At present, most well cooled (or highly distributed) LEDs are between 80 and 120 lm/W for real life applications.  Higher efficiencies are available, but at greatly increased cost.

In the foreseeable future, LEDs will become much more common for residential and decorative lighting, where extremely high light output isn't necessary or desirable.  In this respect, nothing else can compete ... except on initial purchase price.  This too is changing, and it's nice to see that more options are becoming available all the time.  There is no reason that household lights need to run from 230V or 120V - it would be nice if all lighting were 12-24V to keep high voltages out of ceiling spaces altogether.  This naturally also means that lighting is easily backed up with batteries that can be charged via compact solar panels.

This article was prompted (in part at least) by the acquisition of a pair of the 10W LEDs shown above at a pretty reasonable price.  Since I wanted them primarily for experimentation, this was ideal from my perspective.  One of the first things I did was to tap threads into the baseplate (I enlarged and tapped the existing mounting holes to 3mm), and I also took some trouble to ensure that the base was dead flat (the bases were anything but flat as received).  Unfortunately, I can measure light output in lux at a given distance, but not lumens/ Watt - this requires specialised equipment.

As with many other areas in life, manufacturers have found new and exciting ways to lie about the performance of their LED products, and some very strange and highly suspect claims are being made.  Some suppliers are even claiming a scotopic advantage (when the light level is so low that our eyes rely on the rods for monochrome low-light vision).  Normal vision for the work we do from day-to-day requires enough light so we can use the cones (colour sensitive detectors), so this 'advantage' is pure bollocks.  You will also see claims that LEDs outperform all other light sources, but none of the limitations will ever be mentioned.

At present, LEDs are at a disadvantage when you need very high light levels to cover large areas.  To properly illuminate a stadium, metal halide is still the lighting of choice, because each lamp can provide far more light than any known LED of even many times the physical size.  The metal halide lamps are also expected to run hot, so there is no need to try to keep their temperature down with heatsinks and fans.  A 1kW metal halide lamp isn't physically much larger than a 400W version - very different from the needs of LEDs - even if it were possible to get anything like the same light levels.

Despite this, high power LED floodlights are available (I've seen as high as 450W - they are blindingly bright, even in daylight).  These offer the advantage of much longer life than metal halide lamps, which is important when they might be 30m or more from the ground! Even though most LEDs at present are only marginally more efficient than (a new) metal halide lamp, we can expect this to change.  With up to 160lm/W becoming commonplace, and the ability to distribute the light evenly where it's needed with little 'spill' [3], even the metal halide lamp's days may be numbered.  This is especially true when one considers the lumen depreciation of metal halide lamps, something that can be avoided (or at least minimised) with LED fittings provided the manufacturer pays close attention to thermal management.

Page 1   Page 2

These articles are a work in progress, as there are more LED ideas yet to be covered.  Pages 1 and 2 cover some of the products I've tested.

1. Capacitor Life Calculators - Illinois Capacitor, Inc.
2. Wikipedia - Luminous Efficacy Table.
   
3. Philips White Paper: Street Lighting
   

Main Index

Lamps & Energy Index