The Future is not Orange,... but White

David Gibson and Stuart France describe the use of white LEDs for caving lamps and attempt to demolish a few myths. This article is the first of a series that will include practical circuits for LED caving lamps.

Contents

  1. Introduction
  2. The Myths
  3. Why use LEDs?
  4. Comparing LEDs with filament bulbs
  5. Drawbacks of using LEDs
  6. Controlling LEDs
  7. Building a LED lamp now

Introduction

There is currently a great deal of interest in the use of white LEDs to make caving lamps. For example, in a recent issue of Descent magazine (Gear Review: LED lights, Descent 155, p33, Aug/Sept 2000 ) two LED lamps were reviewed. However, although a number of advances have been made, we feel that it is important not to get too 'carried away' with the new technology which still has some shortcomings. Current debate on the Internet, in the Cavers Digest and Speleonics mailing lists, shows that there is a fair amount of misunderstanding of white LED technology.

The Myths

Those statements seem to be fairly common 'myths' about LED lighting; and none of them is true. Here is our interpretation of the real situation.

White LEDs do not produce 'proper' white light

A white LED is, in reality, a blue LED (light-emitting diode) surrounded by a phosphorescent dye that glows white when it is struck by blue light. This is a similar process to that in fluorescent lamps, where the coating glows white when it is irradiated by the ultraviolet light that the tube generates internally.

Some people have assumed that a white LED must work by mixing light of the three primary colours - red, green and blue - and that this means that the light must be spectrally deficient and have a poorer quality to daylight. This is an erroneous assumption - the light is not spectrally deficient at all.

Other people have noticed the pronounced blue peak in graphs of the power spectrum, and claimed that this, too, gives an unnatural quality to the light. Again, this is not true - the blue peak is almost outside the visible spectrum, as can be seen if you calculate the eye-weighted spectrum.

A white LED has a continuous spectrum similar to daylight, i.e. slightly 'blue'. This, coupled with the fact that the beam is so pure, is why the light can look slightly 'unusual' underground.

White LEDs are more efficient than halogen lamps

At the present time, the efficiency (measured in lumens per watt, lm/W) of light generation is similar to that of conventional 'flashlight' halogen bulbs. This means that if you want to replace your halogen lamp with an equivalent LED lamp it is likely to get just as hot and you will end up with a similar battery life. But this is rather a sweeping statement - the pure nature of the beam means that you might get away with less light than you would otherwise need.

An LED is similar to a 0.25A krypton bulb

Objectively, a 4V 0.25A krypton bulb uses 1W of electric power and gives out perhaps ten lumens of light power. If we assume that white LEDs have a similar efficiency then fourteen LEDs, running at 20mA and about 3.5V will be equivalent in power to the krypton bulb - so a single LED is nowhere near as bright as a conventional bulb.

Subjectively, on the other hand, it is actually quite difficult to compare LED lighting and tungsten filament lighting. Which seems the more powerful will depend on beam focus and other factors, which we will discuss below. But, in any case, caving on a pilot bulb or 14 LEDs (or fewer) may not make for a good day underground.

Battery life is greatly extended

Battery life can only be extended if the light output is reduced. To get long run-times you may have to put up with a very dingy light, as the above example should indicate.

LEDs draw very little power and do not waste energy as heat

20mA is the maximum safe continuous current for many LEDs so - yes - a single LED consumes little power (70mW) and does not have the opportunity to get hot. But, since you need dozens of LEDs to make a decent 'main beam' caving lamp, the total power is much the same as for a conventional lamp. Powerful 'main beam' LED arrays such as we have constructed do, certainly, get as hot as conventional lamps. This can lead to severe problems, as we will discuss later.

Using a lot of LEDs is just a waste of power.

This converse objection is also flawed. The fundamental point about using a lot of LEDs is that you do not have to run them at maximum power. You can efficiently dim an LED lamp to whatever power level you like. Moreover, the LEDs are more efficient if you under-run them so - cost reasons apart - comparing two lamps with the same brightness, the one with most LEDs will be more power efficient.

Why use LEDs?

Having demolished a few myths, is there any reason to use LEDs at all? Well, despite what we have said above, LED lamps are definitely the way forward...

They are more rugged than tungsten filament lamps

Flashlight bulbs contain a fragile incandescent filament that can easily be broken. In this respect LEDs are much more rugged.

They have a lifetime of perhaps 100,000 hours

A high-power flashlight bulb might have a lifetime of only 20 hours, but even the longer-lasting bulbs have a lifetime of only a few thousand hours; compared with this, LEDs will last 'forever'. However, it is important to realise that this does not mean that your caving lamp is going to last for 100,000 hours. What it means, of course, is that when your caving lamp finally breaks (for example, you smash the housing, the contacts corrode, the cable falls off, or the electronics fails) then the LEDs will still be OK. It seems that the phosphor in the LEDs may start to degenerate well before the 100,000-hour point, changing the colour of the light, although this is probably not significant.

They can be dimmed efficiently

Vacuum-filled tungsten filament lamps (i.e. not krypton, xenon or halogen types) that are designed as low-current pilot bulbs can be very inefficient - as low as 1 lm/W. Under-running your main-beam (perhaps using a commercial 'dimmer' circuit that is available for bicycle lamps) is equally inefficient. LEDs, on the other hand, can be under-run with no loss of efficiency - in fact, they are more efficient at low currents. Thus a dimmable LED lamp would allow you to choose the illumination level appropriate to the conditions and consequently to save battery life.

They do not need a reflector or focussing assembly

LEDs come with their own built-in micro-reflector that produces a pure, even beam without the need for any external optics that can become damaged, dirty or corroded. However, you still need a waterproof box or conventional headpiece to protect the LEDs from damage.

Comparing LEDs with filament bulbs

To slightly contradict what we said earlier, good white LED lamps probably are somewhat more efficient than filament lamps when all factors are taken into account. But the gain is not huge, and we do not really subscribe to the view that a five or seven LED lamp will give you enough light to cave by. Backed by practical tests of our own lamps, our suggestion is that you need at least 24 LEDs for a 'main-beam' lamp. However, there is something wonderful and amazing about the pure, blue-white beam from an LED lamp that - subjectively - may cause you to prefer even a 3-LED lamp to a conventional one.

Unfortunately, a general comparison is very difficult to make, for the reasons now given. We will discuss these problems at greater depth in the future.

Efficiency of filament lamps and LEDs both vary

Tungsten filament bulbs vary in efficiency from under 1 lm/W to around 20 lm/W. White LEDs start at around 5 lm/W and present production may be up to around 15 lm/W or higher. Clearly there is a lot of overlap here, and your impression of LEDs will depend on whether you presently cave on a cheap flashlight, or a high-quality 12V quartz-halogen system.

Beam pattern is different

The uneven, blotchy beam from a badly adjusted caplamp is so very different to the pure smooth beam from an LED array that even a dim LED lamp may appear to be better.

Beam focus is different

To match the performance of the halogen bulb's reflector you need well-focused LEDs. So an LED with a beam angle of 20° will be better at turning lumens (the light 'power') into candelas (the light 'brightness') than a 50° product. The fact that any lamp can be made 'brighter' simply by focussing it to a narrower beam makes comparisons all the more difficult. It even explains why some people are happy with 3 or 7 LED lamps. For example, David's 48-LED lamp uses parts with a beam angle of 50°, which produces a spot 1m in diameter at a distance of 1m. This is a very wide beam. If, instead, the lamp were to use 20° LEDs then the beam width would only be 35cm at 1m but, within this beam, the light would be over six times as bright. Or, to put it another way, for the same brightness as the wide-beam lamp you would need 7 LEDs instead of 48. (Fig 1, below)

figure 1

Figure 1 - beam comparison. 7 LEDs with a 20° beam are almost as bright as 48 LEDs in a 50° beam, but the area illuminated by the narrow-beam lamp is, of course, much smaller. (The LED ratio for equal brightness is 6.25:1, because it goes as the square of the beam angle).


Colour is different

The blue-white light from an LED array looks 'cold' compared to the yellow glow from a filament lamp. Which seems best will depend on the colour of what is being illu-min-ated. For example, the yellow bias of a halogen bulb flatters mud, sand and 'stal' in caves whereas the green/blue bias of a white LED is better for the outdoors. We are experi-menting with introducing some yellow or orange LEDs to a white LED array to shift its colour balance to better suit caves, and this may also reduce the overall cost. Stuart has painstakingly constructed several arrays and loaned them to cavers to test.

Shadow-rendering and depth perception are different

This is a difficult one - some people have reported that an LED array makes it difficult to focus on smooth surfaces (e.g. mudbanks) and that distance is difficult to judge. However, some of these people were using 3, 5 or 7 LED lamps for this test - not our 36 and 48-LED lamps. A dim lamp makes it more difficult to resolve the scene. On the other hand, perhaps the 'blotchiness' of a halogen bulb beam actually helps the brain to distinguish the edges of objects because the 'moving uneven illumination' has more information content? However, there is no problem with LED illumination if you have enough of them.

Drawbacks of using LEDs

Having discussed the advantages of LEDs, there are some disadvantages to be overcome, as we will now demonstrate...

High cost

Although the cost of LEDs is falling, the best still cost £1 to £3 each depending on the supplier and quantity. If you think in terms of driving each white LED at about 70mW then to equal a 2W halogen bulb you need to use around 28 LEDs. To match a 4W Oldham bulb you would need 56 LEDs and also a well-stuffed wallet.

It is worth warning about the futility of comparing LEDs by comparing the candela rating. As explained above - the narrower the beam angle the brighter the light. Bright LEDs (particularly the coloured 'hyperbright' LEDs) achieve a high candela figure only by virtue of having a very narrow beam angle. If you are comparing the brightness rating of LEDs, make sure that the parts have the same beam angle.

Lack of a long-distance focussed spot

LEDs tend to have a wide beam angle and a smooth beam. There is nothing to match the 'spotlight' focussing ability of a conventional lamp, although it is possible to use a mixture of narrow-beam and wide-beam LEDs and it is conceivable that a commercial lamp could use a lens to provide additional focussing and a spot/flood beam. · Overheating can cause damage As we have said above, contrary to popular views, LEDs can get very hot during operation. Local overheating can lead to an imbalance in the current, with the result that the hot LEDs get still hotter. This process is called 'thermal runaway' and, if not checked, will cause the LEDs in the lamp to fail one by one. For the larger lamps, careful thermal design is required, as well as control electronics that can detect the fault condition or, ideally, prevent any hazardous situation arising in the first place.

LEDs need control circuitry

All light-producing devices work by converting electric current (not voltage) to light. Conveniently, conventional incandescent lamps have the property that their voltage is proportional to the current flow, so they can easily be controlled by connecting them to a voltage supply. However, this is not the case for most other light sources - including fluorescent lamps as well as LEDs - both of which require a controlled current supply. If you connect them to a voltage source without a 'ballast' to limit the current they are in danger of burning out. The current can be limited with a resistor, but this is only really suitable for smaller lamps, as it can be inefficient. A more sophisticated control uses a current regulator; and more advanced still is a power converter, known as a 'switch-mode' supply. This takes some expertise to implement well, and there is the attendant problem of designing reliable cave-proof electronics.

LEDs are dangerous to your sight

The light from an LED is almost a point source. This means that there is a very high power density, and when this spot is focussed onto your retina, there is the potential to cause eye damage. The situation is not as bad as with laser diodes, but it is still a cause for concern. You should not stare into the beam from an LED lamp. Perhaps some form of diffuser would be advantageous?

It is difficult to make a lamp cave-proof

Lamp housings suitable for caving already exist, and so we intend to produce circuit boards that will fit inside a miner's caplamp, with its tough shell and glass front. Stuart has fitted a 36-LED array and its switch-mode power supply inside a British CEAG headpiece, as fitted on some Kirby and Speleo Technics lamps. Conveniently, this supports through-the-headpiece battery recharging.

We have also experimented with waterproof polycarbonate boxes with a transparent lid. Unfortunately these scratch too readily in caves, and they also seem prone to cracking around the screw collets due to the heat of the LEDs. (A good demonstration that LED lamps do get hot!). We will return to the subject of housings and the cave-proofing of circuit boards in a future article.

Controlling LEDs

As we described above, LEDs need control circuitry in order to operate safely and efficiently. A sophisticated controller, using a switched-mode power supply will allow you to dim the LEDs to save power, and it will allow you to extract the maximum amount of energy from your battery. At the other end of the scale, you can simply not bother with any sort of current control - you might get away with it.

The stages of sophistication can be split into five broad categories.

1. No current controller

You can, if you are careful, connect LEDs directly to a battery. This is simple, but very restrictive of battery voltage, reliability and useful battery life. You might do it for a single LED key-fob, but you would be foolish to try it with a multi-LED lamp, since it could be an expensive lesson for you.

The voltage across a single white LED, running at 20mA varies between about 3.3V and 4.0V depending on temperature and manufacturing tolerances. If you happen to be using a diode at the lower end of that scale, and you connect a 3.6V battery across it then it will over-run considerably because - unlike a filament lamp - the current in an LED is exponentially proportional to voltage. As the over-run diode gets hot the current will increase further, leading to thermal runaway and damage. On the other hand, if you use a diode at the top end of the scale, it will be rather dim because you will be under-running it.

There is a particular problem if you try to connect multiple diodes in parallel unless they are 'matched'. We will return to this in a future article.

2. Resistor as current limiter

It is better to connect LEDs to battery via a resistor that allows you to regulate the current in a crude way. But you are still limited to a small variation in battery voltage. For example, a Duracell MN1302 flat-pack might be 4.7V new, and 2.7V empty. If you used, say, seven 3.3V (approx) LEDs with an 8.2W series resistor then at 4.7V you would be running at 24mA, which is a bit high for safety. But, on the other hand, as the battery dropped to 3.6V the light would dim to nearly nothing - unless you shorted the resistor. In summary - you are wasting power in the resistor, and you are only able to use a fraction of the available battery power.

3. Linear current regulator

You can see from the above that what is required is a series resistor that can be dynamically adjusted. This is, in fact, the principle behind the standard series-pass transistor voltage regulator. The difference is that here we require current regulation instead of voltage regulation. The block diagram of a typical circuit would be as in Figure 2, below.

figure 2

Figure 2 - block diagram of a linear current regulator. The complete circuit uses only a few more components than shown here.


 

The reference voltage is compared with the voltage across the current-sense resistor, and the resultant error voltage controls a series-pass transistor. It is not necessary to use a MOSFET; in fact a good quality bipolar transistor will work better. By altering the reference voltage you can dim the lamp.

Depending on the relative voltages of your battery and the LEDs, this circuit can be quite efficient. For example, David's 48-LED lamp (containing 16 chains of 3 LEDs) runs at about 10.5V and 320mA. With a 12V SLA battery this is an efficiency of 85%, which is similar to that of a moderate-quality switching regulator. When the battery voltage drops below 10.5V the circuit comes out of regulation, but the LEDs will still be illuminated, and will now perform as if they were directly connected to a battery of a 'safe' and gradually falling voltage. An SLA is nearly empty at 1.5V/cell, corresponding to 3V/LED here, so sufficient light is maintained to almost to the end.

David has been using this simple circuit for a while. Details will be given in a future article. You could use a dedicated power-supply chip (a 'low dropout linear regulator') and a current-sense chip here, but we are not intending to pursue that line of thought because we feel that the cost is not worth it - you might as well spend the money on a switching regulator.

4. A switching regulator

A switching regulator, or switched-mode power supply (SMPS) can be thought of as a 'transformer' that converts one voltage to another with minimal losses. The series-pass transistor of Figure 1 is replaced by an SMPS module (Figure 3, below) and, again, current feedback serves to maintain the voltage across the diodes at exactly the right point for safe operation.

figure 3

Figure 3 - block diagram of a switching current regulator. The controller IC switches its output transistor at a high frequency, and the current pulses are smoothed by the external inductor, diode and capacitor. Current feedback from the sense resistor requires more components than shown here, but the basic principle is the same. The current is compared with a reference and the output of the regulator and the output of the regulator adjusts itself to keep the LED current constant despite falling battery voltage or rising LED temperature.


 

Both of us have experimented with several designs of switching regulator, and Stuart has been using one in his caving lamp for some time. The advantage of a switching regulator is that you are not limited to a small range of battery voltages - you could have a lamp that would run from an FX2 (2 NiCd F cells), FX3 or FX5, or from a 4.5V alkaline battery pack with its widely variable voltage.

The disadvantage of switching regulators is that they are not straight-forward to design, and are more costly than the other options discussed here. Using a published design - such as the one(s) that we will be discussing in future articles - solves the first problem. And the additional cost is not a large fraction of the overall cost of the lamp - especially if you are going to use a lot of LEDs.

5. Computer control

Computer control may seem to be a bit of a gimmick but, if you do not trust a microprocessor-controlled lamp, then our designs will work without it. But if you do want some additional features, a small 'PIC' processor (from Arizona Microchip) make it easy to add them.

You could, for example, arrange for the lamp to switch off after an hour or two, so that the battery will be saved if you accidentally leave it on. (Obviously, it should not just switch off without any warning, but it is not necessary to discuss such sophisti-cation here - it should be self-evident that aspects such as this will be catered for in the software which is, in any case, easy to modify to suit the requirements). Another option is to implement a 'fuel gauge', where the lamp can give you an indication that it only has a few hours' life left.

One attraction of a computer-controlled lamp is that it could be controlled by a simple push-to-make low-current switch. Because the switch does not have to carry a high current, nor provide multiple functions it can easily be made totally waterproof. Functions such as dimming the lamp, blinking SOS and so on, are all achieved by clicking or double-clicking the switch. This was discussed in an earlier article (Gibson, 1998).

Building a LED lamp now!

Several people have asked us what they should do if they want to build or buy an LED lamp immediately. The bottom line, if you are buying a lamp, has to be "try before you buy" - the little 3 or 7 LED modules are quite pricey for what they are.

If you want to build your own lamp, do not buy the LEDs from a component retailer. If you want to get a good price, you must order at least 50, directly from a distributor. There is a possibility that we could put together a bulk order for people who are interested.

Decide how bright you want your lamp, and what beam angle you require, because this will govern the number and type of LEDs you require. Remember, you do not get something for nothing with LEDs - if you want a bright 'Oldham' replacement you will need a lot of diodes.

In future articles in this series we will give some circuits of varying complexity. We are planning to produce PCBs, and kits of parts as well as ready-built lamps.

This page, http://www.caves.org.uk/led/led-myths.html was last modified on Thu, 01 Feb 2001 00:00:00 +0000