I’m about to start work on a slight overhaul of my lighting equipment, firstly because I hardly have any continuous lighting at the moment and secondly because my studio flashes are a bit big and unwieldy for lighting small subjects. I do have a bunch of portable shoe-mount strobes, which are quite versatile, especially when they’re used with the flexible strobe trigger I built a while ago, but I’m feeling the need for some revamped lighting options, and to do something creative and useful with LEDs. Before I start actually working out the details of a ‘suite’ of new lighting equipment I thought I’d better do some research into how to choose the right LEDs, as previously I’ve mostly used them as indicators in circuits, so their power and colour (other than choosing red or green) was pretty irrelevant. So what’s the story with building your own LED lighting? Lots of figures are quoted for LEDs, but what do they all mean, and which ones are important?
Depending on the manufacturer or supplier, the specification or datasheet for the LED will variously contain figures for some of (but probably not all of) colour temperature, power, intensity and CRI, amongst others.
First up colour temperature, which should be a pretty familiar concept for anyone who’s ever fiddled about with the white balance in their camera and had to fix it in processing. Colour temperature for a light is quoted in Kelvins and is basically the temperature (in degrees Kelvin) at which a heated piece of steel will emit a light of that colour. Generally speaking I’d always be aiming to have my light sources producing colour temperature of somewhere round 5500 Kelvins, as this matches the old ‘daylight balanced’ standard for film. In LED specification terms this falls into the colour band described variously from ‘cool white’ to ‘daylight’. The other common white for LEDs is ‘warm white’ which is closer to what you’d get with a tungsten lamp (at about 3000 Kelvins).
With continuous lighting, if they’re not quoted in the specification sheet, power and intensity should be fairly easy to measure the consumption and output are continuous. Most traditional continuous lighting was just quoted by the power of its bulb, so a fairly average 1000 watt photo flood would be designated as just that – 1000 watts. That however is just the electrical power that the lamp consumes, and what doesn’t necessarily give a great idea of the amount of light the lamp produces. That will depend on the efficiency of the bulb, which for incandescent bulbs is low, as in addition to visible light they produce a lot of infra red radiation (otherwise known as heat). Anyone who’s used (or been lit by) kilowatts of photo floods will know that they run incredibly hot – all that heat is power that isn’t being turned into light. In fact, due to the amount of infra red they produce, the efficiency of incandescent lamps is just about the lowest for any commercial electrical light sources. So the big question is how much of that power is turned into usable light.
To find this out a sensible route seems to be to measure the quantity of light, or luminous power, being produced rather than the amount of electrical power being consumed. Luminous power, also referred to as luminous flux, is measured in Lumens which are defined as “the SI derived unit of luminous flux, a measure of the total quantity of visible light emitted by a source per unit of time.” Bear in mind here that not all the radiation being produced is necessarily light, so the measurement needs to be restricted to the visible spectrum.
Given that things are rarely simple, directly measuring the Lumens produced by a light source is quite difficult without a lab setup. The work around is to measure the amount of light falling on a subject and back calculate the number of Lumens from that. Enter a new measure: Lux. This is the SI unit of illuminance, and is a measure of luminous flux per unit area. With this being an SI unit, the area is naturally a square metre, so 1 Lux is defined as 1 Lumen per square metre. (other units are available – in the US and, historically, the UK the unit is/was the foot candle)
The good news here is that you can buy a Lux meter pretty cheaply, and the calculation part isn’t too difficult either. If the light source is nominally omni-directional, light is emitted equally in all directions and a surrounding area of one square metre would represent the inside of a sphere.
Given that the formula for the surface area of a sphere is:
A = 4πr2
Where r is the radius. Rearranging this gives:
r = √(A/4π)
Which for a 1m2 surface area works out as a radius of 28.21cm.
So if we place a Lux meter’s sensor 28.2cm away from the light source we will effectively be measuring the output of the source in Lumens. This is good news, as it should give the ability to verify the quoted figures, or measure unquoted figures, for various light sources. This should start to give an idea of how many LEDs or LED strips are going to be needed to make a continuous light source that generates the amount of light we want.
On that basis I bought myself a reasonably priced Lux meter. When it arrived I took some readings in daylight conditions with the Lux meter alongside my old trusty photographic incident light meter, to get a handle on the relationship between them and see if the Lux meter was making sense.:
Daylight conditions – even but bright overcast; Lux reading around 19,000; EV reading about 13; Exposure given 1/125 @f8, ISO 100
Early evening – overcast: Lux reading 4700, EV reading 11: exposure 1/125 @f4, 100 ISO
The formula for converting EV to Lux is:
Lux = (2EV * 2.5)
So for an EV of 13, we should get 20,480 Lux; and for an EV of 11 the value should be 5120 Lux. Given that I don’t have an accurate reference I’m reasonably happy that with less than 10% between the two measurement methods the results from the Lux meter readings are likely to be reliable enough for what I’m trying to achieve.
Getting back on track, I set up simple a 28.2cm jig to measure the output of my existing lights, and take readings from the range of lighting and LED products I had to hand:
- 100W tungsten bulb: 1207 lm
- 20W CFL bulb: 1130 lm measured (claimed 1200 lm)
- 10W LED bulb: 1294 lm
- 3W, 3.2 volt SMT COB LED: 339 lm
- 100W COB LED chip running at 30V (rated to 36V): 15830 lm
- LED panel built from 5050 strip with 288 LEDs: 5370 lm
- Neewer CN-160 (160 LED) video light – no diffuser: 15950 lm
A couple of notes on these results:
- I was only able to run the 100W COB LED chip at 30V, as my lab power supply can only output a current of 2A when it’s configured to give a 36V output – this isn’t enough to drive the LED at full power.
- For the home-built LED panel, each of the 5050 LED’s has a power rating of 0.2 watts, so for the panel the power is about 57 watts.
I don’t have an old fashioned 1000W photo flood to measure, but the quoted output for these in Lumens appears to be about 16,000 lm. That’s broadly similar to the output of the 100W COB LED chip (at 30V).
Quoted figures I’ve seen seem to indicate that tungsten bulbs generally give 12-17 lm/W and fluorescent lights 45-75 lm/W: these are in line with the measurements I took with these types of light sources. LED figures vary, as LEDs seem to be getting more efficient: my results ranged from just under 100 lm/W for the (old) 5050 LED panel, through 113 lm/W for the 3W LED to 150+ lm/W for the 100W COB LED chip – that’s about 10x the light intensity per watt for tungsten.
So the answer to the question of how much LED power I’m going to need to build into these lights will depend on the application. Using a target figure of about 10,000 lm (requiring about 60W of LEDs) for small scale studio work is likely to be fine. For more general studio work, once you start backing the light away from the subject the inverse square law kicks in and with only 10,000 lm available shutter speeds are going to start getting very slow: in this case I think I’d be aiming for 75,000 lm or more – now that should be fun to build!
With the question of power/intensity answered, the other key factor in LED selection is colour rendering, which I’ll be looking at in my next post.