How to measure the light output of my LEDs accurately?

Discussion in 'The Projects Forum' started by EnjoyIce, Apr 5, 2016.

  1. EnjoyIce

    Thread Starter New Member

    Jan 13, 2014
    Hi guys,

    I am creating an array of 12 bi-colour PWM LEDs which will be powered by an arduino + LED driver chip.

    I need to find a way to measure/monitor the output of these LEDs to make sure they are behaving as expected (so we know the code/chip are both working correctly).

    Is the best way to do this: create an identical circuit using photodiodes and create some code to read the input to the photodiodes and then compare?

    I was thinking about just using one photodiode to test each LED one after the other. However since these LEDs will all be turning on/off dependant on the timings in the code it would be good to know if they are turning on/off at the correct times relative to each other - which I can't do with just one photodiode. (I could maybe do this with one photodiode if I just synced up the results from each LED, that way I could compare their timings accurately).

    If you know of an easier way then feel free to share it! Thanks.
  2. Dyslexicbloke

    Active Member

    Sep 4, 2010
    It rather depends on what you actually want to measure...
    Lux, level of illumination, is probably what you ate talking about but that would be measured at some specified distance or alternativly at a location where it was importent.
    If you were designing an office lighting scheme for example you wouldnt care much what the luminare were doing close up but you would definatly be expected to provide some specified level of illumination, too bright is as bad s too dim, on every desk.
    You may even want to altr the lighting to compensate for ambiant light from windows.

    You should probably do a little diging and decide what it is you want ro achieve, in the broardest sense, and then go looking for a solution.

    Lux meters are available as handhels instruments as well as calibted sensors and both can be used with simple optics to studt a field but of course you would then be looking at feflected light as opposed to a source at a distanceand that is a whole other ball of wax.

    If tou want something cheep and simple that you could calibrate yourself, we'll mostly, you will want a cadnium sulphid cell, light dependent resistor, as they have a better curve that a photosiode.
    Back in the day when verything was descrete or electromechanical, some SLR cameras used a small solar cell and a moving iron meter.

    You may also want to think about colour, more accuratly called colour temperature, and if you are going thre than you will probably need to considder specrum and its effects on colour rendering.
    Articals about building desing and photographic lighting are a good place to start.

    In all honesty its a huge subject and you may not need to get all that involved, like I said earlier, you need to decide what you actually need.

    Whatever you need will be posible
    Hope that gets you started, the electronics are likely to be the easy bit unless your spec is very demanding, but you will need a spec.

  3. wayneh


    Sep 9, 2010
    Agreed. You need to be very specific about what you want to know before you can devise a test rig.

    I would also recommend not building anything. There are scads of commercial devices for measuring light level - you could even use a smartphone. I doubt very much that what you need is not already available.

    One problem will be the effect of any lens. Perfectly on-axis, an LED might be very bright but 10% off-axis, maybe not so much. That means you'll need tight angular control if you want to measure the luminance-vs-angle profile.
    OBW0549 likes this.
  4. mcgyvr

    AAC Fanatic!

    Oct 15, 2009
    -be specific
    -plenty of devices out there already.. (as little as $15 shipped)
    -cosine correction,etc.. can be a biatch

    and really the information is already in the datasheet curves..
  5. OBW0549

    Well-Known Member

    Mar 2, 2015
    I'll second that advice. A quick check at the App Store shows more than a dozen light meter apps for the iPhone, most of them free; I assume it's much the same for Android devices. Don't bother trying to build anything, it's not going to be worth the headache.
  6. EnjoyIce

    Thread Starter New Member

    Jan 13, 2014
    Thank you for the replies - its been interesting to read up on things since I'm new to this stuff!

    Yes I am looking to measure Lux. I am measuring the light output power incident on a surface a specific distance away.

    This is for a project I have been asked to do and it requires accurate measurements - I need to be able to say that: Yes, these LEDs provide a power of __mW/m2 at 2cm distance away etc.

    Luminance vs angle would also be good if possible. The target surface is a spherical disk with a diameter of about 1.5cm, so being able to say that in the middle it has an intensity of *xx* compared to an intensity of *yy* at the edges would be great.

    I believe these are my requirements/spec then: (I may be using the wrong termonology here .. sorry)
    - Measure the light power incident on the surface a specific distance away from the LEDs
    - Measure the difference in that power over the surface (middle vs edge)

    Is there any hardware out there that you would recommend to me for this? Needs to be accurate in its readings (Not sure how good a phone app would be?)

    I do have the datasheets for the LEDs however since it is a project that I need to create a document for (experimental results etc) then I can't just state them from the datasheet. Sure I can say what it *should* be but I need experimental evidence to prove it!

  7. dannyf

    Well-Known Member

    Sep 13, 2015
    The answer to your question will depend on what you meant by "accurately".

    Most light sensors, even the el cheapo ones, do a great job discriminating different levels of light output - they can rank order a bright situation from a slightly dimmer situation.

    However, very few can tell you exactly what the lux level is. One solution is to buy expensive light sensors. Another is to buy an inexpensive one, and to calibrate it against a known sensor / meter.
    OBW0549 likes this.
  8. wayneh


    Sep 9, 2010
    I can think up several ways to measure the light, but I can't think how to calibrate it, unless you could identify some sort of standard source. I'm picturing something like half a tennis ball with lots of fiber optic cables embedded around the surface at lots of different angles. These cables bring the light to a flat surface, either an array of light sensors or maybe a single CCD or a flat bed scanner that could read each cable's signal independently. (If the phototransistors are small enough you could just mount them directly into the hemisphere.) I haven't seen such a device and it makes me wonder how the pros measure LED radiance versus angle. I'm willing to bet it's something more elegant! Probably worth a little research?
  9. EnjoyIce

    Thread Starter New Member

    Jan 13, 2014
    I'm going for the cheaper ones and then calibrating it - seems the best solution! As much as I'd love to create a tennis ball contraption (It's quite an interesting idea actually!), I just don't have the time anymore :(

    I need to measure the light output incident on a surface ~2cm from the LEDs. Will this be able to read that correctly (won't be over powered or anything?). I will test the lights at 100% brightness all the way down to 0% and then create a table with lux/brightness etc.

    I am looking at this:

    I'll probably run the test 4/5 times to help get me an accurate reading on the lux at each level. Just need to make sure the meter will be ok for what I'm planning on doing! If it is not - do you have any suggestions for ones that are suitable?

  10. MrChips


    Oct 2, 2009
  11. wayneh


    Sep 9, 2010
    That could be quite a high intensity. Make sure your meter can handle that.
  12. MrAl

    Well-Known Member

    Jun 17, 2014

    I think it would be rare for you to actually have to measure the light intensity in some standard units. The only reason for this is if you were manufacturing something and you needed it to conform to some set standard in order to create some appeal to the general public or you have to meet some standard in order to meet the specs for a company that sub contracts out to you. A situation like that would mean you would need to match the specs by some standard in light measurement.

    But if this is just for yourself and you only need to COMPARE different light sources, then you are in a different playing field where you can set your own standard and this means you get away cheap and easy.

    Building a comparative light meter is very easy. Two pieces cut out of a pure white cylindrical candle glued together with aluminum foil in between to block the light from each candle body. Shine the light standard into one end, shine the light to be compared into the other end, observe the device from the side. When the lights are equal to a fairly accurate degree it will be apparent. The slices of candle only have to be about 1/2 inch thick. If you always use the same light standard (perhaps an LED with 20ma flowing through it) you'll get comparative readings that tell you how much better or worse another light source is.
    White LED's are fairly linear up to a point too, so you could vary the current to lower the light level to compare to lower level light sources like at the edge of a light pattern from a typical flashlight.

    If you need to measure total light output though then you need an integrating sphere which is harder to build.
  13. EnjoyIce

    Thread Starter New Member

    Jan 13, 2014
    I'll have a look at those converters thanks!

    Doing a quick calculation: The LEDs are 7200mcd and around 25mm away. Lux = 10.76391*candela/distance^2. Gives me ~11,500lux. So a 100,000 lux meter should be sufficient?

    May pick up a converter and wire it up to an arduino and also try a lux meter, best of both worlds :)
  14. EnjoyIce

    Thread Starter New Member

    Jan 13, 2014
    Sorry, didn't see your reply MrAl!

    I need to measure the lux output so I can then calculate the power incident on the surface. I'm creating an array to stimulate cells (like neurons) so the cells have a minimum power they need to receive in order to be stimulated.

    So I need to measure light incident on the surface (cells) and calculate at what brightness of the LED I achieve this. So a simple lux meter is probably good enough? Its just a case of how strong it needs to be! Although by my above calculation (Which could be wrong) then any one thats 100,000 lux limit etc should be fine?

    I need to produce a report/guide for this in which I can say that at 50% brightness the LEDs produce a power of xxxx at the surface etc. So the user can look at up power they need and set the brightness accordingly.