I am an electrical engineering student currently undertaking a third year design course. The project I am designing is an autonomous caving/mining headlight, detecting the environment’s ambient light to accordingly adjust the output light intensity and frequency. This will allow for hands-free operation, with the goals of saving power, increasing safety and convenience, as well as a more accurate colour reproduction of the dark environment. Some of the other features considered are the control characteristics of the light, type of light, weight distribution of the circuitry and a disable function to allow manual control of the light
I found a similar product ‘Petzl Nao’, unfortunately only after my design proposal was accepted. Their product measures the reflective light of its own beam which gives a small sensing angle and disregards light from the surroundings (e.g. from other cavers). Also there product comes with a $200 price tag (my market research on this product has everyone saying the price tag is its major flaw) for a max 355 Lumens. If the use of LDR’s proves to be sensitive enough, I believe mine would be much more cost effective.
Also talking to a caving instructor in New Zealand, he said that he took a flame into a cave once and gave a realistic 3D feel to the environment. That’s why I have a manual colour adjustment for the user.
Has anyone ever used a colour other than white light that gave a much more realistic visual of the environment?
At this stage I am trying to gather as much information as I can for the design. So far I am planning to use ~3 CREE XM-L/2 LEDs (possibly with multicolour LEDs), and simple optics with basic LDRs for light sensing. For the batteries, I will probably use 18650s or AAs, and a microcontroller to handle the logic of the dimming and colour variation. I am wondering if these are good choices; any feedback or ideas would be of great assistance.
Sounds like an interesting project
Have a look at Tint, Binning, and CRI Explained (For XM-L LEDs) for info on tints and CRI (Colour rendition index).
There is quite a difference between LEDs and how they display colours.
A while back, a member here named ToyKeeper built a 3 xm-l2 light using a cool white led, a neutral white led, and a warm white led. This gave it far superior color rendering as compared to using any one of the individual tints. Just an idea………
Also, I would highly recommend 18650 batteries for their much higher energy density.
I would suspect that using one simple high CRI led would do the job fine. I guess that the 'bad 3D perception' that you mention is caused by the widely used cool white led (6000K-7000K), using a neutral (3500K-5000K, depending on taste) 80CRI Cree XM-L led will already make all the difference in the world (80CRI is the light quality that is the current standard for home lighting). If that is not sufficient, the Nichia 219 (A or B) 92 CRI led is as good as it gets (at the cost of less efficiency than the Cree XM-L led).
RGB-leds will only be best when colour lighting is needed (as in: when real colour is needed, not for tints of white), for 'white' illumination the leds mentioned above are far superior because they emit a whole spectrum, not just three discrete wavelengths.
A 'microcontroller to handle the logic of the dimming and colour variation' is a widely used way to handle that, but because leds are more efficient at low currents, there are led-drivers out there that dim by actually reducing current instead of fast switching (at the moment a new promising driver by BLF-member led4power is tested that does that, it might be an idea to look into his driver).
When using a switching microcontroller, the switching frequency often seen for dimming is a few hundreds of Hz, at this forum (we're snobs) we want the switching be at least 1000Hz, or even better 10KHz, to avoid nasty strobe effects when illuminating fast moving objects. Look into the NANJG105C driver for a good standard constant current li-ion led-driver with already lots of handy features. There are drivers derived from that one that have even better options.
While these ideas seem simple enough, proper implementation of these ideas is nontrivial. I believe the issue with 3D with the candle is that the flame is dynamic (it is moving), which allows you to notice the different reflections and depth better.
I’ll post more on the subject later but for now here is a video of a RGB only (no white emitter) light in an underground environment. I love RGB lights but RGB only is not the right choice for this project, you NEED white light, maybe one high driven XM-L2 HCRI supplemented with RGB emitters would do well.
Toykeeper’s wide spectrum BST comes to mind.
There are aeveral other videos of this light, in both dark and with some ambient lighting on my YT channel. I also have some 100m outdoor beamshots I’ll dig up, the colors do combine to make “white”, white objects show up white but the light coming out doesn’t really look white. I’d venture to guess its only around 20cri (comparing it to LPS/HPS lighting in terms of CRI).
If you are looking for primary/rechargable battery compatibility, the spark SX5 battery container comes to mind.
If you are seriously considering RBG (W) in your light, the XM-L color has almost everything you are asking for.
If you are going to push high currents through the LED, I recommend a noctigon direct thermal copper board, or a sinkpad. The cooler the led are, the more efficient they are.
a smaller led die will “throw” light farther than a larger led die with the same reflector. eg. cree xp-e vs cree xm-l.
mtn electronics has really good prices, is US based, and is a user on this forum (username: RMM)
"I would suspect that using one simple high CRI led would do the job fine."
I would suspect that too. Also I think that High CRI is it far behind it's potential and more High CRI need to be had on the neutral range than on the 2800K-3200K where CREE insists on their High CRI.
Mixed triple of bad tints can offer "far superior color rendition"? Just a question based on a reply in the thread. Maybe the word "far" is too far?
If I am to take that literary I cannot believe it is is true compared to a single Neutral High CRI Nichia LED for example. I mean mixing a random bad CRI with greenish tint like 1C with warm white 3000K (random also) will give me a neutral light which cannot have far superior CRI over something that is supposed to have high CRI (unless it is the 3000K High CRI from CREE which is not what literary people get when thinking of CRI). Or what I am trying to do with the bluish or yellowish LEDs in the first place? If it is for my personal entertainment sure I like it, but I am not sure I can be that undecided to choose from 3000K or 7500K both with unreal rendition and in the end settle at neutral white by mixing the two because that is so much better than those 2 separate and that is why neutrals are sought-after in the first place.
A mistake is done if light is considered from blue to yellow (Kelvin) and the green to pink axis is ignored, which is a larger spectrum, must give Credit to CREE that in the 5XX tints the precision of the tint is pretty good, It would have been nice if that was the case for 4X tints, but that is why they are not 4XX tints.
1, the beam in a reflector will be not-round since each die is off center and there arnt any optic’s for it. IMO using individual emitters in reflectors/optics that can each be canted slightly towards each other (or a triple/quad optic where this is done already) is the preferred method to get the best “white” as well as the best overall beam pattern, to get the colors to mix you have to get the beams to mix, the best way is to have them pointed at exactly the same spot in exactly the same optic setup.
2. Direct thermal MCPCB’a are just about impossible to get, they exist from sinkPAD but you can’t buy them unless you want 45. I know, I tried, I managed to get 5 of them (gave one to DJozz) but it was not an easy task and I almost guarantee the way I got them won’t happen again.
3. While each die can be driven to ~2-2.25A from DJ’s test I doubt they’ll be happy being driven to that level at the same time. With XP-E2’s you can run them each at 2.6A at the same time no problem. In the XML color the die’s are sharing a single thermal path, even with the individual XP-E2’s in a triple or quad board each die get’s its own thermal pad’s / pathway away from the die.
Hey guys, thanks for all your replies, they have all been extremely helpful in formulating a final design concept, which I have included in the following link.
If you can spare 3 minutes to fill out a quick survey about the product for the quantitative part of my market research that would be greatly appreciated and any additional feedback can be posted here or in the survey if you see any major flaws.
Interesting concept. As I said in my survey, good luck keeping the price down where you want it. If you haven’t already, check out the Kavelight by user kevinm on candlepowerforums. In particular, the battery box is awesome — you could drive a truck over it.
