I just posted the second in a series of articles on LEDs on the subject of color rendering in LEDs and how it can be improved.
We’ve taken a vacation from the blog to put some key products out over the last 18 months. In the meanwhile, Light Emitting Diodes (LEDs) were becoming a new lighting source for film and television and also have been a bit misunderstood in what they can and can’t do. Much hype is out there about LEDs and new products coming out every day. People are understandably interested in their capabilities, but some are skeptical. Overall, very little information out there about what they are and aren’t.
We’re coming back now as we bring out our first LED film lighting product with some companion articles. Here, you’ll come to understand LEDs and how useful they can be in certain kinds of film and video lighting and also what their limitations are. You’ll also come to understand that they are far more efficient than lumens or lumen per watt specs can tell you. Finally, you’ll see real world examples that illustrate the value of lux or footcandle measurements vs. lumens and what the difference is.
KEEPING AN OPEN MIND
At one point I was fairly convinced that if you look at lumen output and lumen per watt specs as your sole points of comparison then LEDs don’t add up to being usable as real lighting. As a lighting designer I’ve come to use lumen output of bulbs quite a bit and many people use them as a point of comparison in evaluating the efficiency of different light fixtures that use these bulbs. This conviction kept me from working much with LEDs or trying to make a product.
On the other side, there are photographers, filmographers and even some videographers that are interested in photometric specifications of fixtures as taken with a light meter. This is becoming less prevalent in the digital video age where scopes and zebras have taken the place of a light meter for many as an indicator of proper exposure.
In any case, I saw some photometric specs (lux or footcandle measurements) of real world LED fixtures and felt I needed to know more though about how they could make what seemed to be outlandish claims in the face of anemic looking lumen specs. How could, for instance, an LED fixture be the equivalent of a 500w tungsten fresnel when the LEDs add up to about 1500 lumens (using lumens per LED multiplied by the number of LEDs in the fixture) and a 500w tungsten bulb would have about 10,000 lumens output? All the same, the footcandle or lux output was claimed to be about comparable between the LED fixture and the fresnel, so something wasn’t adding up and more study and an open mind were needed.
All this was a useful exercise though because it made me question more the relationship between lux, footcandles and lumens which is not well understood by many. It wasn’t until I started experimenting with LEDs that many of the realizations soaked in, the paradoxes were explained and I learned what an LED is and is not. Sometimes you just have to get the hands dirty, to actually realize something that seems unfathomable!
During this experimentation period, I also realized that lumens per watt and lumen output are totally irrelevant in trying to compare LEDs to other lighting. Its a paradigm that was invented for bulbs and only works well for bulbs alone. To compound things, LED manufacturers are so clueless at this point in their evolution about measuring or communicating the strength of their product so its not easy to believe any claims made in comparisons to conventional lighting.
For instance, Its appropriate to use an integrating sphere and spectro-radiometric computer to measure the light output of a conventional bulb because of its 360 degree output. The sphere is perfect for this and is the only real method of bulb spec measurement. The bulb dangles at the end of a cord in the middle of a plain white large sealed sphere, so the environment is predictable, dependable and repeatable. All our current Cool Lights bulbs (fluorescent, HMI, tungsten and CDM) use this method to measure their specs. LED manufacturers use this same measurement paradigm too because they think of themselves as a bulb manufacturer.
The truth is, they haven’t really thought deeply about what they are yet, nor do they know all the best ways to measure or portray their products. They are all so new and inexperienced and they are learning. An LED is not a bulb so shouldn’t be compared to a bulb–this is a central theme of this article and a statement likely to surprise some. In truth, if these LED manufacturers wanted to portray their product in the best ‘light’, they would not use the sphere to measure output strength, nor would they report specs in lumens but would rather report in lux or footcandles. If you want to know the real ‘effective’ lumens per watt of an LED, (1). you would need to take the photometrics with a light meter, (2). compare it to a real world fixture with similar photometric output, (3). find the lumens per watt of the bulb used in that fixture and you have the real effective lumens per watt of the LED. Or even better, figure the Lux per watt as a more apt comparison between the two.
Thus, as you will see in my logical reasoning within this article, these LED manufacturers are putting themselves at a competitive disadvantage and contributing to misunderstandings about the real strength of LEDs. They are not building bulbs but rather fixtures. They are fixture manufacturers who are trying to compare their product to bulbs and its not apples to apples as it should be. In effect, when an LED company produces what they call a ‘bulb’ made of LEDs (ones that can swap out in common sockets like Edison type that were designed for other kinds of bulbs), what they really have is a fixture made up of a quantity of smaller fixtures. Then that ‘fixture’ screws into another fixture where you used a conventional bulb before.
You will thus come to understand why a light meter should be used to measure fixtures and/or LED output–not the integrating sphere as is done today.
Use lux or footcandles to measure output for fixtures (beams) and lumens for bulbs (by themselves). That’s a main point to understand here.
So, to view things in proper perspective, you have to agree with me that an LED is a fixture and not a bulb. If you don’t agree, read on and see more of my support for this claim. I’ll logically lay out for you why lumens and lumens per watt don’t apply to make any relevant evaluation of what an LED can really do.
Disclaimer: Despite the great efficiencies that LEDs have and advantages over conventional lighting systems, I will not take the view with what’s available today, that LEDs will replace everything from tungsten to fluorescents to HMI in a short time. To me, that’s hype. Despite the fact they have been around for a while, LEDs (when used as lighting) are still in an early stage of development and should be taken like that to view them in their proper perspective. In my opinion, because of their stage of evolution and cost, LEDs have their own niche uses, particularly where batteries as a power supply are appropriate, but aren’t a one-to-one replacement for any other established lighting tools (yet). Consider LEDs and the fixtures they come in on their own merits and you won’t be disappointed with what you can do with them.
LUMENS VS. LUX/FOOTCANDLES OR BULBS VS. FIXTURES
To understand how LEDs, which on paper with lumen output seem to not come even close to some existing light sources, can actually be more efficient than it would seem possible, we need to understand what these lumen specifications are and what they’re really meant to be used for and why they are irrelevant for use in LED light output measurement or comparing to bulbs. If you follow along, you will see why lumen output of bulbs is really only interesting when comparing bulbs and not fixtures. To understand all this, just keep repeating to yourself:
A single LED is just a tiny fixture and not a bulb so keep apples with apples and oranges with oranges. And lumens are used to measure conventional bulb output but lux or footcandles should be used to measure fixture output. Since an LED is a fixture, don’t use lumens to measure its output, use lux or footcandles…
Light output from bulb manufacturers is measured in lumens simply because they have no way of knowing all the different fixture configurations the bulb may be used in, nor do they care. It’s also a great way for customers to compare the output of one bulb to another. Lumens are conveniently measured in the scientific isolation of the aforementioned integrating sphere that all bulb makers use. A light meter wouldn’t work as well for this use and wouldn’t be adapted to the repeatability or dependability needed for the factory and assembly line environment. The sphere has been the output measurement paradigm for a long time and makes sense from a lot of points of view–particularly the fact that these bulbs have a 360 degree output and the sphere is made to measure that kind of output.
However, lumens won’t tell the story adequately once a bulb is actually integrated into a fixture. Then we no longer have a 360 degree output but a beam. A fixture would be defined as a system with bulb and other components to make the bulb emanate a beam as strong and efficient as possible. There are exceptions of course, like the China Ball fixture which emphasizes the 360 degree output but that’s not relevant for our topic here. Putting a beam fixture into an integrating sphere to measure lumen output would not work because the sphere is not made for the measurement of a beam. The more appropriate measuring tool for a fixture/beam is a light meter. Photographers agree that a light meter and its lux or footcandle measurement is the best way to evaluate fixtures and exposure, but for different reasons. They’re also more interested in fixtures than raw bulbs as well for obvious reasons.
LED manufacturers are following this same measurement methodology of the sphere to their own detriment. The reason for this is that LED makers think of themselves as bulb makers but nothing could be further from the truth. An LED simply doesn’t emanate light the same way as a bulb so comes off weak in the 360 degree measurement environment of the integrating sphere.
LEDs are made to put out a beam and are therefore a fixture.
The problem with this measurement quality control and sales paradigm for many types of LEDs is that the LED IS a self-contained micro fixture already with a beam (which by itself isn’t much use for any real lighting but we’ll come back to that later).
The tiny chip that goes inside the LED is the “bulb” and the acrylic (or other type of) housing is the fixture which focuses the output into a highly efficient beam. So, doesn’t it seem silly to try to measure this like you would a bulb in the sphere–especially after i explained that the sphere isn’t efficient for something producing a beam and not a 360 degree pattern?
What other bulb can you think of that has a spot lens built right in like our 5mm LED? This would never be economical for large quartz bulbs with their relatively short life and it works in the LED world precisely because of the LEDs long life. They do have an efficient lens already built in (and no reflector needed as that doesn’t work anyway for 5mm types). However, we can’t do much real lighting with a micro fixture though!
LEDS ARE LIKE ‘DIGITAL LIGHTING’
We need a quantity of these micro fixtures to make a “macro” fixture. A “digital” light (so-to-speak), just like we make displays out of pixels (one pixel won’t do us much good), we’re making a usable light out a bunch of 5mm LEDs.
Note: Because of this, one shortcoming of our macro LED fixture will never be a perfect point light source like a tungsten or HMI fresnel or par spotlight. The perfect point source has a great “shadow rendering index” (SRI) or ‘project-ability’ because it is a hard light source and produces super well defined shadows with single edges. LEDs and fluorescent sources have terrible SRI because they don’t cast well-defined shadow edges. They give multiple, diffused or “pixelated” kinds of shadows. Because of this low SRI, an LED array or fluorescent fixture won’t render a cucaloris (cookie) pattern correctly on a wall. So we’ll have to go back to our conventional tungsten or HMI fresnel or ellipsoidal spotlights with their single point light source, when we want to do that.
On top of all this, we haven’t even treated the subjects of color temperature or color rendering capabilities in relation to LEDs which could be an entire article on their own.
COLOR TEMPERATURE AND LEDS
Color temperature and CRI of LEDs is not specifically what we want to talk about in this article but it does bear some discussion about the issues involved.
You never see color temperature meters at a bulb (or LED manufacturer either). They know for instance that correlated color temperature (CCT) can’t accurately be measured on many types of new technology sources by anything that can fit in the back pocket of your jeans—particularly if that source is not full spectrum. Plus its just not repeatable or adapted to the accuracy and dependability needed for a factory environment.
LEDs are not full spectrum sources at this time. This is a reason that they don’t mix well with other fixtures like fluorescents or HMIs. Some LED fixtures make up for this shortcoming by using RGB LEDs to mix custom color temperatures and simulate full spectrum. A very quick fix for this issue is to use a 1/4 minus green filter in front of the LEDs, then the fixture will mix well with other lighting.
Again, LED makers only use integrating spheres with attached spectro-radiometric computer peripherals to not only obtain the lumen output but also to get a spectral analysis which includes CCT, color rendering index (CRI) and other relevant specs. This is the only really accurate, isolated, predictable and repeatable way to read the simulated color temperature and CRI of sources that are measured with a CCT, like LEDs. While the lumen part of the spec is weak in telling the story of an LED, the other parts of the spec are very necessary and hard to obtain accurately from other test equipment. So, having the sphere does make sense for CCT and CRI specs.
All the same, only real full spectrum light sources such as real tungsten or real daylight can dependably be read by most color temperature meters because they are not measured with CCT but simply CT. This will most likely continue to be the case until we get more color temperature meters that reliably read CCT and also non-full spectrum sources.
So, we have talked about what’s behind all the confusing specs that mislead people about the real strength of LEDs. Now, lets really prove the point by taking a look at a real life example bulb and how its light output efficiency can change based on what fixture its in and how its lumen spec is the same regardless of what fixture its in.
SEVERAL FIXTURES / ONE BULB
Once you get to fixtures in combination with bulbs, you’re talking about shaping and beams being possible, different efficiencies and uses—we’re coming out of the laboratory into the real world. In the real world, we need lux or footcandles to tell us how strong the beam is and how far it can throw. Lumens are fairly irrelevant in the real world…
Its time to prove some of these claims with real world examples now. To really illustrate how lumens mean nothing once you actually get into a using a bulb in different kinds of fixtures, let’s take a look at some examples that all use the Osram 575W HMI bulb:
Lumen output is 49,000 lumens. Nice, but its too abstract. To say that all fixtures that use that bulb put out 49,000 lumens would be misleading in terms of communicating true strength. Let’s start with a basic fixture to use it in, like the Arri-X5 HMI 575w floodlight for instance:
Once you see the manifestation in this first fixture, you believe you know what this abstract 49,000 lumens is by shining it on a wall or a scene/subject. You mark this all in your mind as 49,000 lumens: a very bright, daylight white output at 6000K CCT. It’s a simple fixture too, just a reflector, clear safety lens and the bulb.
Next, take a 575w HMI fresnel like the Arri Compact 575w HMI for example:
Seems a bit stronger and perhaps you account for that by the lens addition.
Then you learn of a fixture that’s even more powerful but uses the same bulb: an Arri 575w HMI Par.
It’s visibly brighter than the other two and puts out more light all by itself, with just the safety glass lens on the front. Looking inside, perhaps you realize the efficient parabolic reflector, with bulb in the middle, helps harness the 360 degree output better than the fresnel or floodlight. You also find out that there are other lenses you can apply to the front of this par from a super wide beam lens (which scatters all the light energy over a wider beam and is thus not concentrated enough to send so much light out a far distance) to a super spot lens (which does concentrate and allows more light energy to throw farther—harnessing more of the light into one direction). Your vision of 49,000 lumens and what it means has been totally remade.
If we carry this one step further and find a really well made ellipsoidal or follow spot made for the 575w HMI bulb, you’ll see the specs are even more impressive thanks to the convex spot lens. The point has been made though.
This is when you realize that it’s not just about the bulb but about the entire system. All these are using the same bulb, the same 49,000 lumens, but they’re doing different things with it and getting different results. Lumens per watt calculations do you no good either, other than for comparing bulb efficiencies at a very high level—all these use the same 86 lumen per watt solution but some are clearly getting more for their ‘86’ than others.
That’s why photometrics are a more viable way to measure output from fixtures and ‘throw’ for photographers because we’re out of the 360 degree world of the bulb by itself and into the manufactured “beam” of a fixture system and what it can do for us in lighting an actual subject.
You should now understand why lumen output is only a way to compare bulbs (and in a weak manner for LEDs too) but tells you nothing of the actual system (housing, reflector, focusing, lens, power supply, etc.) that the bulb will be used in—any of which may be more efficient than another. Nor does it tell you anything about throw which is important as well.
Changing the beam angle clearly changes the throw and therefore more intensity concentrated in one spot. That’s relevant because sometimes we need that spot concentrated as much as possible to “fight daylight.” If you put a light meter in that spot, it tells the story that an integrating sphere isn’t made to tell. Lux or footcandles as measured by that meter are the preferred methods used by photographers and what they most care about. They’re also the most appropriate way to measure beams too.
Lux is simply the amount of lumens found in one square meter produced by whatever source we’re measuring. Foot candles are Lux divided by 10.76 (or for simplicity sake just by 10 as much of the lighting industry does). Lumens by themselves don’t take into account any distance or area so aren’t useful for actual real world uses with fixtures and especially the beams they create.
Look at the photometrics of these three previous example fixtures, all using the same type bulb / lumen spec to illustrate the point.
MEASURING THE BEAMS TELLS THE REAL STORY
Before when we looked at the fixtures, we didn’t get very scientific about the output. We were just going by feeling about the perception of the intensity. Let’s really measure each one (using a light meter) to finally prove the point.
575w HMI Arri-X5 Floodlight (just a flood fixture with no focusing ability or lens):
28 foot candles at 20 feet
575w Compact HMI Fresnel (focusing ability, mediocre mirror/bulb setup and fixed Fresnel lens—can’t change easily):
75 foot candles at 20 feet in flood mode
625 foot candles at 20 feet in spot mode
575w HMI Par (super efficient par reflector plus interchangeable lenses to focus the beam):
113 foot candles at 20 feet with super wide lens (50 degree)
250 foot candles at 20 feet with wide lens (20 x 45 degree)
750 foot candles at 20 feet with medium lens (10 x 20 degree)
1875 foot candles at 20 feet with spot lens (9 degree beam)
5000 foot candles at 20 feet with super spot lens (5 degree beam)
Remember at the beginning when you thought you knew what 49,000 lumens were with the 575 floodlight?
We went from 28 foot candles to 5000 foot candles and never changed the bulb! Can you start to see how lumens are irrelevant when considering anything other than a bulb by itself?
So, what lens and mirror are being used is super relevant in comparing how the same bulb will be used to get different results. How does all this relate to LEDs being too weak to use or not?
Take an example where you choose one of the weaker configurations for the 575w HMI above like the flood or the fresnel. Now, choose a weaker bulb too, but put it in a more efficient housing, mirror and lens and try to get the same output as the higher wattage bulb in the fresnel or flood fixture.
For instance, take the CDM 150 type metal halide bulb with lumen output of about 12,000 lumens. Lumens per watt is about the same as the 575w bulb at 86 LPW. However, it’s a fourth of the lumen output of the 575w HMI bulb and also almost a fourth of the wattage draw too—how could the 150w CDM ever be a match for the 575w HMI bulb? It’s clearly a story of the underdog triumphing.
DAVID VS GOLIATH
How to get that unfair advantage? Put the 150 in an ellipsoidal fixture. Effectively a small spotlight with super efficient reflector, mirror and convex lens.
189 foot candles at 20 feet in wide 35 degree spot mode
527 foot candles at 20 feet in narrow 15 degree spot mode
Almost as good as the HMI 575w Fresnel in spot mode but only drawing 150w to do that. Pretty amazing too, when you compare it to the expensive 575w par with wide and super wide lenses. Remember how we were scattering the 575w HMI par light energy in a wider area with those lenses and not getting so much throw? You clearly gave up something to do that.
If all you knew how to do was compare lumens, and lumen per watt figures, it wouldn’t tell the whole story about how a smaller wattage bulb (or LED for that matter) with weaker specs on paper can actually do better than the other higher wattage bulb with better specs by using it to its best advantage. You have to get off comparing things on paper and do some real world comparisons and research using a light meter to see any of this though. This is why the light meter must be our tool of choice for measuring beams and fixture output–not the lumen specs of the bulb! In a nutshell, this is how the weak appearing LEDs in an array can be as strong as a 650w fresnel.
THE MICRO-MINIATURE SPOTLIGHT
Back to LEDs: just think of a 5mm LED as a miniature spotlight with a very efficient lens that trains the light forward with as little waste to the sides as possible. Primarily though, the degree of beam angle in the LED (or for that matter the lens of a regular fixture) is a main factor in what you’ll get out of the fixture in the way of strength and throw too.
You can buy 5mm LEDs in practically any beam angle you want. You may find differences in efficiency, CRI or CCT issues when trying to find the ultimate LED for photographic use—as I did. Ideally, you’d get something like a 20 to 30 degree beam LED and you’d really have an efficient setup with most of the light energy going forward, strong throw and not being wasted. Then to get a bigger beam, just keep adding LEDs to your array until the beam is the size you want—creating our “digital” beam out of LED “pixels”. That works until you consider CRI which is terrible in such a sharp angle LED (most of the time). It’s not until you get to the more “flood” class of LED (50 degree and higher) that you will start to see a better CRI.
Another way to keep things efficient and bright is to choose an LED of 8000K or higher color temperature with 30 degree beam angle. 8000K to 9000K LEDs have better output as a general rule (than those in the lower ranges from daylight down to tungsten), and seem to more easily obtain a better CRI (if it’s a relatively good quality LED). Photographers won’t appreciate it because its not 5600K “daylight”, but you could gel it down when you need something lower. 5mm LED daylight exceptions exist but they’re more expensive and the fixture doesn’t add up in terms of cost/output.
1W to 5W LEDs operate under different rules as well and other things are possible when using those—but that’s another story for another day and another product.
COMPARISONS BETWEEN LEDS AND REAL WORLD FIXTURES
Now that we’ve seen how its possible for an LED fixture to be far stronger than we think, lets prove it by looking at some photometrics comparing LED products with an Arri 650w tungsten fresnel. All readings below are in lux.
All this illustrates why it’s totally possible for an array of micro-spotlight LEDs, which on paper have weaker specs, and have them roughly equal to a 650w Fresnel when you create a macro fixture out of them. The lumen output of our 40 degree panel is actually about 2000 lumens when considering the measurements of output of each individual LED (as measured in a small integrating sphere made just for LEDs), then taking that figure and multiplying by 600 for the number of LEDs in the array. Considered by itself and only knowing that spec, you’d think the panel was indeed about as strong as a 100w tungsten bulb (tungsten with an efficiency of about 20 lumens per watt) which would also be about 2000 lumens.
But knowing the rest of the story, that you got here:
About how bulbs work, fixtures or beams vs. raw bulbs, how bulbs or LEDs are integrated into fixtures, when to use a light meter or an integrating sphere to measure output, and finally when to use lumens or lux / footcandles–then and only then, you can start to understand how such claims about such a panel really being equivalent to a relatively high output fresnel spotlight are possible.
In conclusion, if you look at lumen figures of bulbs and lumens per watt alone, it doesn’t tell the whole story of LED efficiency vs. other kinds of lighting. First, you’ll have to acknowledge the difference between fixtures and bulbs. You’ll also need to start considering photometrics for real comparisons of what photographic / film lighting bulbs or LEDs can really do once they are harnessed into a fixture and beam output. Plus, you’ll have to approach a new paradigm like LEDs with an open mind, realize they are in an early development stage and that they are not a one-to-one replacement for anything, but rather have their own applications that they are strong in and should be considered as just another tool in the lighting arsenal. Not at all to replace the other tools but complement them, at least for now.
Next time, we’ll take an in-depth look at the very misunderstood subject of LEDs and their poor color rendering index (CRI) performance along with ways to improve it.
(c) Copyright 2009 CoolVideoLights.com. All rights reserved. May not be used in part or total without the express permission of the author in writing. Waiver of damages: This information and all included material are provided as is, and Cool Lights USA can not under any circumstances be made responsible for any damage, injury or losses caused directly or indirectly by implementation of the information in this article.
I posted the last and final part of the Metal Halide Fresnel series in the Articles section of the blog. In Part IV, I discuss the Cool Lights Bulb choices for our coming Hardlight series.
I just posted the part III of the Build Your Own Metal Halide (Low Cost HMI) Fresnel in the articles section of the blog. In this part, I talk about what we’ve been up to behind the scenes to leverage all this information to create what we call the Hard Light Series which will be the perfect complement to our already existing and soon to expand Soft Light Series. Part IV will be posted shortly where I talk about the standard Osram bulb choices in the Hard Light series and the equivalent replacements that can be had for lower cost.
I just posted the long awaited and eagerly-anticipated part II of the Build Your Own Metal Halide (Low Cost HMI) Fresnel in the articles section of the blog. I also posted a new article on Electrical Safety Guidelines.
I just published Part 1 of a multi-part article on building your own low-cost HMI type fresnel in the Articles section of the Blog. In part 1 we cover a bit of the theory behind HMI, what it is and how we can find “generic”, less expensive versions of it to serve in our low cost fresnel.
I just put up the test results for our bulbs. I’ve had them for a long time and had been meaning to put it up but finally got around to it. We also included a test report of an Osram Dulux L 3200K bulb for comparison purposes. You can find the test results here.
I just posted two new articles relevant to fluorescent lighting for media production in my blog. They are:
You can post comments if you have some relevant information or corrections to make!
Richard (from Houston, Texas)
My name is Richard Andrewski and this is my company website and blog. Cool Lights is dedicated to producing DIY and other low cost solutions to video lighting. In this blog I’ll be posting a lot of thoughts about DIY video light building, low cost fluorescent studio lighting and the issues surrounding that, fluorescent bulbs, ballasts, and in general, everything about building your own video lighting. (more…)