LED headlight experiment, part 2... and a discussion on headlight optics

sportyaccordy

Newly Enlightened
Joined
Jan 5, 2013
Messages
7
I saw this thread and the gears immediately started to turn

http://www.candlepowerforums.com/vb/showthread.php?211405-LED-projector-headlight-experiment

I bought a bunch of crap (20W Chinese LED chip + driver, Morimoto D2S mini projector, currently waiting for Autopal H4 reflector, supposed 180 degree beam 50mm wide LED chip and some XM-L 16mm chips) to play with. I took some semi-crappy camera phone pics of the LED chip + Morimoto projector, but first I wanted to just talk about headlight beam patterns in general and what I hope to accomplish with this little project.

DOT beam pattern is basically comprised of a long distance hot spot and some peripheral flooding. As technology has advanced, headlight makers have managed to "spread the hot spot", getting a nice even distribution of light over a long distance with a minimum amount of glare.

Here is a modern DOT reflector beam pattern:
See Rule #3 Do not Hot Link images. Please host on an image site, Imageshack or similar and repost – Thanks Norm

You see the hot spot distinctly with the weak flooding

Here is a modern DOT projector beam pattern:

http://i59.photobucket.com/albums/g297/jvxdriver/hidplanet/DSC_2748h.jpg

You see the hot spot is spread out much wider which makes for more even coverage.

What I want to do with this project is try and create an LED retrofit into a halogen/HID reflector. Once I have the optics down, I'm hoping to create a legitimate LED retrofit product.

Here are the pics of the cheap chip + projector. Chip is the standard 20W "cool white" with a width of about 1.25", projector is 2.5" wide.

http://i.imgur.com/HR3ds.jpg

This is with everything aligned properly. If you compare this to the DOT projector, three things are apparent: 1 there is a lot more foreground lighting, 2 the beam cast is a lot narrower, and 3 the cutoff gets wonky towards the sides. This is because the projector is designed for a reflected point source of light, while the chip has about 25-30 bins. There's two ways to potentially solve this, both of which I will experiment with.

http://i.imgur.com/8FMbH.jpg

This is with the projector shifted to the left (from this perspective). This shows that potentially, if the chip were wider, it would be able to cover a wider spread. Wonky cutoff problems could be addressed by utilizing a flat shield.

http://i.imgur.com/6TPUg.jpg

Same shift to the right.

I have a 50mm LED chip coming so I'm hoping having light cover the span of the horizontal axis of the projector will amount to a wider beam similar to the DOT projector.

Next up, a technical analysis on why LED dies cannot replace point source bulbs, and how I plan to exploit these differences.


Your images are too large and have been replaced with links Please resize and repost.
See Rule #3 If you post an image in your post, please downsize the image to no larger than 800 x 800 pixels. - Thanks Norm
 
Last edited by a moderator:

sportyaccordy

Newly Enlightened
Joined
Jan 5, 2013
Messages
7
Initially when I took this on I ran to Ebay to see what was out there. There's two LED retrofit options on the market but both have optical problems. I am sure you guys have seen the shitty "pineapple" conversions with cheap dies just strewn about. Obviously problematic due to the reflector's need for a point source at a specific location, as well as the low lumen output, lack of cooling etc so I won't even discuss that.

I saw this though and was impressed for a little bit, but then figured out why this wouldn't work:

swypF.jpg


As I mentioned, DOT low beam reflectors are built around point source reflectors, meaning the distribution/intensity of light is the same from any angle. This means a couple of things, but for the sake of brevity I will just say it means a lot of light gets wasted (I saw somewhere that ~400lm of a 1200lm halogen bulb actually gets put to the ground from a low beam projector- I know for certain at least half the light of any low beam projector gets lost behind the shield). LEDs generally have some distribution of light that ranges across 0-180 degrees that goes from 0 to 100% intensity. This is good and bad. Bad because you can't just stick an LED die at a reflector's focal point and get the same beam pattern. Good because you can construct an array to send light only where you want it, and even potentially boost the weak spots on a shitty reflector.

I thought about how to construct an array and how to arrange individual dies to either simulate a point source or even potentially enhance the light source to better distribute the light. I picked a Cree XM-L because it has high efficiency and really tiny size. I decided I would try and see what the distribution of light would be if I put two XM-Ls at an angle from each other to simulate a point source. I looked at the distribution data and found that the distribution was pretty much the top half of a sinusoid which made things 100x easier. Here's the plot of the best angles:

Bg1LK.png


The numbers on the side denote the angle of the LED from the horizontal; 0 corresponds to two LEDs on a flat plane, 90 corresponds to two LEDs faced back to back. This is only looking in one plane (the horizontal), which I figure is the most important as that will correspond to the horizontal plane of lighting output. Here we see 60 degrees comes closest to a "point source" of light. Still a little lumpy though, so I figured let me run the same calc with 5 XM-Ls at an even angular shift from each other + equidistant from the old filament focal point. I normalized it to the # of bulbs (i.e. 1 on the Y axis = max output of 5 LEDs) and found the best below:

x74uZ.png


So this means 5 LEDs at an angle of 158.75 degrees from each other will put out a relatively even distribution of light along their respective horizontal planes. Added bonus, you could really underdrive the hell out of the XMLs and still be delivering a ton more light exactly where the reflector wants it.

I am thinking of adding another LED to the top to add a little foreground lighting if this proves to make too narrow a band of light; but I don't think that will be necessary. I have a machinist buddy who can carve me the mounting bases (and necessary cooling fins as well); the challenge for me now is actually soldering the damn things. I ordered 10 XMLs raw not realizing how tiny they were or that they had to be reflow soldered.

Anyways what do you guys think?
 

SemiMan

Banned
Joined
Jan 13, 2005
Messages
3,899
I think you will have a lot of fun trying, but unless you have some expertise in optics (or a lot), you will not be able to achieve anything street legal. You will never legally be able to put an LED into an existing housing and have it DOT compliant. Not possible. So your only option is complete replacement which at this point is pretty much custom optics.
 

-Virgil-

Flashaholic
Joined
Mar 26, 2004
Messages
7,802
Anyways what do you guys think?

I sincerely admire your passion, drive, and curiosity. I don't want to seem as though I'm trying to rip you down. But what you appear to be proposing (make a viable, legally and technically acceptable automobile headlamp by combining various LEDs and driver electronics with various optical components originally intended for use with halogen or HID light sources) is not possible. You can come up with something that produces what resembles a headlight beam, but that's as close as you'll get. Homemade headlamps are illegal and almost always very unsafe, no matter what kind of components are put together to make them. If you were to devise an item actually capable of producing a headlight beam, extensive tests according to specified protocols would have to be passed -- not just instantaneous photometry but also photometric degradation with prolonged operation, corrosion, resistance to water and dirt ingress and to vibration, and others. If you were to pass all these tests, then you'd have a headlamp. If you were to install it behind any kind of a lens (as in a typical headlamp housing) then all those tests would have to be re-run and passed.

Some specific points:

There's two LED retrofit options on the market but both have optical problems

Any "retrofit" of one type of light source into an optic designed for a different light source is a non-starter. That goes for "LED bulbs" and "HID kits" and any other such "retrofits".

There are beginning to be LED headlamp retrofits, in the (only acceptable) form: complete headlamp assemblies incorporating housed light source(s) and optics -- designed, built, tested, and certified or approved as headlamp assemblies. They are presently available to replace the four main US sealed beam sizes and shapes as well as the 90mm projectors popular throughout the world. But installing LED light sources in an optic designed for something else dooms the project from the start.

I bought a bunch of crap (20W Chinese LED chip + driver, Morimoto D2S mini projector, currently waiting for Autopal H4 reflector

I agree with your assessment; everything on that list is indeed crap. Also, there are no "H4 DOT headlamps"; H4 is not an approved light source for vehicle headlamps (except motorcycle headlamps) in the United States.

DOT beam pattern is basically comprised of a long distance hot spot and some peripheral flooding.

That's a reasonable general one-line observational description, but the actuality is a great deal more complex. There are various minimum and maximum intensities specified at a large number of points and zones within the beam pattern. These must be met or you haven't got a beam pattern, you've got a spray of light that resembles a headlight beam.

Here is a modern DOT reflector beam pattern:

Maybe. What vehicle and what headlamps are depicted?

What I want to do with this project is try and create an LED retrofit into a halogen/HID reflector. Once I have the optics down, I'm hoping to create a legitimate LED retrofit product.

Not technically or legally possible.

Wonky cutoff problems could be addressed by utilizing a flat shield.

Several different cutoff shapes are permitted by US and international headlight regulations, but these, too, are specified in detail. Just saying "I'll just change it to a flat shield!" doesn't cut it.

Do please keep us up to date with your experiments, but please be careful how and where you steer this thread; Rule 11 of this board prohibits advocating illegal or dangerous activity.
 

sportyaccordy

Newly Enlightened
Joined
Jan 5, 2013
Messages
7
I am really just looking to see how close I can get from an eyeball observation point of view. I don't have the capabilities to make something that would withstand the rigors of my current wheels (Kawasaki EX650)

But I am hoping maybe to parlay this into a masters thesis or maybe even a job. And it's just the kind of tinkering I enjoy.

Sent from handheld Minority Report console
 

Alexandrus

Newly Enlightened
Joined
Jan 12, 2013
Messages
1
This sounds quite interesting, I was thinking about the very same thing, multiple XM-L LEDs mounted on a heatsink of some kind inside a HID projector, but only used for flashing, not as an actual driving beam or high beam, actually disabled when the headlights are on (bi-xenon).
 

argleargle

Enlightened
Joined
Jan 7, 2013
Messages
321
...at the very least your initial post to start the thread looked like you're thinking pretty hard. Why not look up the DOT regulations concerning criteria and testing for headlights and sift through it?

I wouldn't want to discourage sportyaccordy from doing it legally in the slightest. Just because you need to get testing and certification done doesn't mean it's illegal to do lab work on this subject... just don't road test it.

DOT headlight stuff doesn't apply if the vehicle is never taken on the road. Sportyaccordy's home made headlights could be 100% tested and used in off road applications. He wouldn't even need a driver's license, insurance, or current license plate to do so... just as long as the vehicle never hits public roadway.

...or am I wrong? I'll apologize and if what I said is illegal, I'll edit my post if I am wrong. I try not to post junk. PM me if you disagree or post it here and we'll talk about it. I'd rather you post why I'm wrong and I'll add "Edit: I was proven wrong" in the offending post.

Update: So off road doesn't apply. I stand corrected and I do what I say I'm going to do.
 
Last edited:

-Virgil-

Flashaholic
Joined
Mar 26, 2004
Messages
7,802
DOT headlight stuff doesn't apply if the vehicle is never taken on the road

That's a common misunderstanding. In fact, the much-used phrase "off-road use only" has no legal meaning. Items of regulated vehicle equipment, including headlamps, must comply with the applicable regulations if they are physically capable of being installed on a vehicle certified as complying with Federal Motor Vehicle Safety Standards. Since all vehicles sold in the US since 1968 are by law so certified, that effectively means all vehicles. Moreover, replacing a regulated item with one that does not meet the requirements ruins the compliance of the vehicle.
 

argleargle

Enlightened
Joined
Jan 7, 2013
Messages
321
Thanks for the reply. I was unaware there were no regs to back up "off-road use only."

...he could still do lab work, right? The only reason I mentioned that is he used the phrase "master's thesis," thought he had an educational background or something.
 

AnAppleSnail

Flashlight Enthusiast
Joined
Aug 21, 2009
Messages
4,200
Location
South Hill, VA
Thanks for the reply. I was unaware there were no regs to back up "off-road use only."

...he could still do lab work, right? The only reason I mentioned that is he used the phrase "master's thesis," thought he had an educational background or something.

The beef here is that putting these lamps into a car makes it a trap for the unsuspecting. Making stock-looking illegal headlights puts a real damper on your day if you bought such a car unknowing and got in a wreck. Even if the beam pattern was quite good, the insurance liability often goes to the driver with faulty safety equipment.

TruckLite and other LED-car-light manufacturers who are doing things right make stuff that clearly isn't a headlamp. They're cans that sit on the bumper or wherever you put them. Sporty, for your current means it might be best to work on a DIY super offroad light. Without a LOT of backing, it's tough to get everything approved for your own headlight. In the US, the lamp assembly has to literally sit in the desert on a shelf (Carefully facing the sky in a particular way) and still be xx% transmissive of light. There are other expensive tests and other long-term tests. The design/build/test loop for safe-and-legal headlamps gets expensive and time-consuming. This is why OEMs do it, and why headlamps are often old technology. They are that way to ensure that they'll meet some minimum of performance after years of use, and be reliable in nearly all use conditions.
 

-Virgil-

Flashaholic
Joined
Mar 26, 2004
Messages
7,802
Without a LOT of backing, it's tough to get everything approved for your own headlight.

There's a lot of work required, and quite a bit of cost, but there is no approval involved.

In the US, the lamp assembly has to literally sit in the desert on a shelf

No, only sample chips of the lens material have to pass the 3-year Florida and 3-year Arizona tests. These tests are not done on actual headlamp lenses or on headlamp assemblies. AMECA maintains a list of acceptable lens materials and coatings that have passed the tests; as long as a headlamp manufacturer uses materials on that list, there is no further materials-selection obligation. Note that these materials are not available in sheets or other forms readily usable by do-it-yourself/home hobbyist situations. They are for injection molding, and the coatings (which must be used -- this is not optional) require costly and complicated processing equipment.
 

sportyaccordy

Newly Enlightened
Joined
Jan 5, 2013
Messages
7
Some updates. I got a paper from a manufacturer of LED headlights explaining the challenges and requirements of creating an LED headlight, and how they went out and actually did it for manufacturers

www.lpi-llc.com/Papers/SPIE05_5942-13.pdf

Here is the ray trace for their end product

kN0XiOg.png


And the design of the headlight

bK8QFer.png


To me it seems there are 3 pieces- a central hot spot, a right side flare and side spill, which all peak at intensity at the top horizontal line. Fair enough

I played with the cheap 20W LED for a couple of hours last night and actually found out some cool stuff. I came to the conclusion, like them, that one LED die wouldn't be enough- not even because of manufacturing tolerances or whatever, but simply because within the tools I have right now nothing really creates a convincing beam. So for my design now I would need at least two elements.

Here is a pic of my "wide beam element"- basically the LED chip shining through a rectangular slit:

rpHGgpM.jpg


Here is a pic of the central hot spot- same assembly shining through a typical projector convex lens (never mind the flare on top- that will be eliminated with a shroud)

3UsEAtU.jpg


Both of those pics had the same exposure/ISO/aperture/position

And then here are the two superimposed over each other with the brightness of each tuned to emulate the LPI end product

BG6gMoV.jpg


Theres a couple of problems. The biggest being that my flood doesn't properly vary in intensity on the y axis. Also stemming from that, the y axis spread is too big. X axis spread is actually great and would translate to really wide coverage on the road. I think to address the wide angle y axis problem, I will get a bar magnifier (basically a cheal cylindrical convex lens) and put a shield above the central horizontal axis to collimate that beam and hopefully better control the intensity. This will collimate the beam in the y axis while *hopefully* not at all interfering with that awesome horizontal spread + distribution. Another idea would be to put a polarizer in front of the assembly, which would both collimate the beam and properly distribute the intensity, but also prob blunt output to the point that it would kill any efficiency the LED assembly was supposed to create.

The central hotspot actually seems alright but could use a little work. I would like it wider. I found that with a wider LED chip, it doesn't come close to the width of a typical point source/ellipsoid beam, but it is wider. I have a wider chip on the way. Also I used a flat cutoff shield. I will make one with a flare. The only problem is with these multi die chips, cutoff lines orthogonal to the axis of the array get blurred, so instead of a nice clean step you get a fuzzy one. Still, not really a big deal.
 

Alaric Darconville

Flashlight Enthusiast
Joined
Sep 2, 2001
Messages
5,377
Location
Stillwater, America
No, only sample chips of the lens material have to pass the 3-year Florida and 3-year Arizona tests. These tests are not done on actual headlamp lenses or on headlamp assemblies.

Which may introduce a flaw into that test. On actual headlamps, the light (including any UV) would hit the reflector and potentially bounce back through the lens for a double dose. (Depending on the orientation of those plastic chips and any backing, etc.)
 

-Virgil-

Flashaholic
Joined
Mar 26, 2004
Messages
7,802
There are many flaws in the 3-year Florida/3-year Arizona test protocol, including the one you mentioned.
 

argleargle

Enlightened
Joined
Jan 7, 2013
Messages
321
Another concern, aside from the million others, is that LED output has been proven to change over the lifetime of the emitter. It can increase or fade. Thoughts?

This is actually one of my favorite sub-forums. It's incredibly hardcore and I'm treading as softly as I can.
 
Last edited:

AnAppleSnail

Flashlight Enthusiast
Joined
Aug 21, 2009
Messages
4,200
Location
South Hill, VA
Another concern, aside from the million others, is that LED output has been proven to change over the lifetime of the emitter. It can increase or fade. Thoughts?

This is actually one of my favorite sub-forums. It's incredibly hardcore and I'm treading as softly as I can.

Most are rated to 70% output at XXthousand hours with YY temperature at ZZ current. You bring up the interesting point that most LED headlamps would not be able to be physically de-rated to have sufficient output at EOL. They'd have to have a loooooong timer stepping up current.
 

-Virgil-

Flashaholic
Joined
Mar 26, 2004
Messages
7,802
Yep, emitter output characteristics (including change over time) have to be taken into account when designing LED headlamps. Another issue that's new to LED headlamps specifically is aging characteristics of optical materials (plastics, polycarbonates, coatings, etc.) used in close proximity to LEDs. The relatively high blue output of most white LED emitters can cause yellowing and opacification of many materials that start out optically clear. Once it begins, it's a runaway exponential degradation process.
 

argleargle

Enlightened
Joined
Jan 7, 2013
Messages
321
...and some sort of optical feedback mechanism would only add another layer of complexity and yet another (or more) point(s) of failure and probably about 5,000 pages of goverment documents as well.

So. Can't regulate based on current versus light output, right? I've read of emitters changing impedance not necessarily related to light output. I'm scratching my head wondering about a good solution aside from underdriving the emitters, hoping that the output would hopefully hold stable for a longer time. It's a false hope. I know that chemical reactions and degradation in the presence of an electromagnetic field isn't that simple.

I suppose a valid question is: "What is a reasonable mean time before failure" value for a LED headlight? Not this 100,000 hours stuff. We know better, that the emitter won't be the same. Should it have a digital-kill-switch after a factory break in period? That raises issues of being safe in of itself. At least a "warning light on the dash?"

Thoughts?
____________
Sent directly from my brain using synapses plus the OSI layer model.
 
Last edited:

-Virgil-

Flashaholic
Joined
Mar 26, 2004
Messages
7,802
Yeah, a feedback mechanism would be a non-starter from the cost and complexity standpoint. It's more along the lines of designing the headlamp so that its output will still comply with the regulated requirements for some reasonable timeframe, with emitter and optics aging effects taken into account. The auto industry doesn't care a lot about headlamp durability after the new vehicle warranty and any available service contracts have expired -- if they did, we wouldn't see all the clouded headlamp lenses as we walk through every parking lot! The regulations permit a wide range of performance, so designing in a margin for degradation while remaining within the regs for seven years (or ten or whatever any given automaker's decided target will be) isn't all that hard.

I suppose a valid question is: "What is a reasonable mean time before failure" value for a LED headlight?

Very good question. I was talking some time ago with an engineer at a Tier-1 supplier (i.e., supplying headlamps to vehicle makers). The subject of lens durability came up, and I remarked that it seemed odd for them to have chosen a particular lens material for their LED headlamp, because it's a material known to cloud up and degrade with time -- not a huge deal if it's a lens on an inexpensive halogen headlamp, but a different matter on an expensive LED headlamp with "to infinity and beyond" light source lifespan. Shouldn't the lens durability come closer to matching the light source durability? The engineer's answer sort of came out of left field for me. He said they'd discussed using hardened glass or a more durable polycarbonate, but decided against it because despite the practically infinite rated lifespan of the LEDs they were using, there would certainly be some failures before the end of time. They decided that if an LED failed, it would give their company (and LED headlamps in general) a "black eye". But if the lens clouded up and became useless in seven years or so, well, all headlamps do that; it's just how things work, sigh, guess it's time to replace it.

I don't agree with that kind of cynical philosophy, but the US market often demands that things be done in ways that are decidedly less than optimal.

digital-kill-switch after a factory break in period? That raises issues of being safe in of itself. At least a "warning light on the dash?"

HID headlamps dim substantially with usage. This could be detected either directly or by proxy via ballast power characteristics, but that is not done and we have no dashboard warning light or kill switch. All polycarbonate headlamp lenses develop "cataracts" which ruin the safety performance of the headlamp. This could be detected in several ways, but that isn't done either and we have no dashboard warning light or kill switch. I don't foresee any such eventuality with LED headlamps, either.
 
Top