My current Automotive LED Headlight replacement project!

Yes you heard me right! :rock: It sounds crazy, and it will be expensive, but I am not paying for it. Dont worry about how it will be done. I have plans and it can be done! I know, I know. I will get so much criticism over this idea since it hasnt been done by many.

My current concern is how to to drive them. Mosfets will take the current, but my knowledge of using them in a circuit is limited. The more powerful HexFETs will reduce the parts count, but I dont know if they are built differently than the standard mosfet and need to be controlled/wired differently.

All I need is a circuit that has two inputs, a HI and LO. The circuit should have a fixed voltage (to protect the LEDs from spikes) and it should have two fixed current levels (like 350mA and 700mA - each should be set by a resistor). If I have a cluster of LED's wired in six parallel banks of four that are connected in series, then I can feed different levels of current to each 14V sets. I need to see how R-binned or even S-binned) K2's are available, but I have untill the end of summer. With these, I can send 2.1A per headlight (350mA per bank) and get 1080lm per headlight (45lm per LED). Then I can send 4.2A per headlight (700mA per bank) and get 1560lm per headlight (65lm per LED). I can even use a cluster of 28 or 29 for more output.

I chose the K2 because I can safely power (within specs) the LED at two separate levels and still maintain a great efficiency (though not close to HID). I could use the U-bin and have fewer LED's and optics, but then I get a bit more heat and it will lack efficiency even further. I can always power them over or under specs to fine tune the output. Also, they are cheap (-ish).

My findings is that most aftermarket 55W Xenon headlight bulbs output 1500lm (Hi beam?), so I figure that the Lo beams are about 1000lm. Older cars only put out 1000lm. HID puts out roughly 3000lm per headlight. I understand that what makes the headlights so bright is the throw (the candlepower), but this project is just for fun. Also, the beams from the Hi and Lo on this will pointing in one direction and the difference will be the brightness (so far). I may have two 7-LED clusters that act as fog-lights to shine light lower to the ground. Possibly I can have a simple motor or cable system to tilt the LED and heatsink assembly a few degrees to make up for the difference in Hi/Lo beam. Hmmm... :thinking:

The copper heatsink material I plan on using should work well. It is only 60W per headlight at full power, and many processors in your PC today dissipates that much heat. Plus the Lo beams wont require much power and everything should stay relatively cool. Plus I will have a 12V 120mm fan for each heatsink (or two 100mm fans) to keep it cool.

Well, I plan on talking to my friend this weekend to discuss the pattern of the cluster. I was thinking of a simple hexagon shape, but I can put them in different shapes and have a middle bar light up brighter or have two circles each light like an eye. I can go with different LED types depending on the design, but that depends. My friend has an older s-10 painted and fixed up pretty nice (almost not worth LEDs but that is okay). His headlights are the 200mm sealed beam headlights. I found the euro headlight conversion for it and I can cut and modify that to my liking. The dimentions are 200mmX140mm so there is plenty of room for different designs and shapes. I liked the K2's because of how they can handle the current and heat. Plus they are relatively cheap. I know they are still imperfect, but they will do. When you are oncoming traffic or a person getting their mail by the road, the color bin wont cross your minds.

Plus this was always my dream and my friend thought it would be cool. He will be one of the few (if any) that can say his car is lit ENTIRELY by LED's. I get excited everytime I think about this project.

If anyone can post tips and ideas for a driver circuit, that would be of great help! It can be done, it will just take a bit of time and work. Its like fitting a 440 in a pinto. Its been done.

Some challenges you will encounter:

• Once completed, this concept will not be street legal unless you submit to DOT testing at tremendous expense
• Lumileds spec sheets are rather optimistic in their assessment of performance. Given the heat of the nearby engine, I suggest you halve flux performance values and look at those as optimum
• K2's are not available in quantity much anywhere
• If you're using individual emitters, prepare to go to extraordinary measures to isolate each slug from your common heatsink
• I hope you plan on using individual optics/reflectors for each emitter (and a variety of them to get a combination of close-range flood and a shaped downstream pattern with the right vertical cutoff); I doubt you will get a useful beam pattern if you simply cram them all into a reflector designed for a pseudo point-source halogen bulb. Prepare to spend a great deal of time tweaking the beam for acceptable vertical cutoff unless you intend to use this for off-road.
• There are plenty of optics optimized for use with the classic luxeon star - of interest is the incredible number of eliptical models - these will help with close-range flood
• The typical 12V muffin fan won't last long under automotive conditions. I suggest you size your heatsinks for passive operation so your LEDs don't fry whenever the vehicle is stationary; otherwise you'll get excellent forced-air cooling when the vehicle is moving. If you need forced-air cooling, I suggest you look at some compact scroll-cage fans - they're typically more durable than axial fans and can handle signifigant backpressure (read: clogging)
• Automotive DC is specatcularly dirty. Prepare to waste some power just conditioning it for LEDs. You might find some of the "automotive" voltage regulators to be up to the challenge

Hey, I think this is a great idea! Let us know how this works out.



Here are a few suggestions, please feel free to ignore / ridicule them.



I think that Osram ostars would be a better choice than K2's. The ostar emitter has 3 times the output of the K2 so you can use 4 stars to replace a 55w halogen instead of 12. The only reason for using a K2 is its resistance to heat. One drawback of LED as you mentioned is that the efficiency is only slightly better than halogen. Since cost is of no concern, you could increase efficiency by using twice the number of emitters and driving them less but you still won't come close to HID efficiency.



To achieve HID efficiency you could use 100lm/w Nichia 5mm LED's but you will need about 250 of them per bulb. That would look pretty cool. Not sure how you would collimate this though.

If you can pull this off you will be the first person I know of to have made working LED headlights. I have read of concept cars with concept LED lights, but none of them seem to actually make it outside. I have wondered if a simple line of LEDs running across the front might provide more even lighting than a bulb on either side. If you want active cooling consider running mineral oil through the heat sinks, or perhaps heat pipes. I hope you will post build pics often so we can drool over them. Please keep us informed!

High beams are not just brighter, the beam angle is different. Low beams are angled down so the light up the road but do not shine into oncoming driver's eyes.

High beams are angled higher so they light up the road much farther with the down side that they will shine into the oncoming driver's eyes. In a car headlamp this is done by positioning the two filiments of the bulbs so they are at different angles relative to the reflector.

Part of aligning the headlamps is to set the max angle for the low and high beams. You won't be able to achieve this by running LEDs on full vs. reduced power. That will just give you bright and dim but not high and low.

I suppose one could mechanically angle the LEDs for the high low effect. A selenoid could flip the angle up for the "high" beam when you also increase the brightness.

Sorry I didnt reply so soon. Thanks for everybody's comments. I appreciate them. I know that the beam shapes will be a great challenge. I did think of an idea to tilt the heatsink with the LED's. I can use a system where a cable that can be push and pulled on the dash (like a manual choke or the older headlight switch). A geared motor or solenoid system would work also, but I will have to figure that out. With the cable system, a slide or toggle switch could be positioned so when you pull on the knob, it will throw the switch into bright mode. Then it can be angled and the brightness can be slightly increased. I plan on using some of the elliptical beam Fraen optics if needed for lighting the road, but if the heatsink can be successfully tilted and it will work, I could use just standard 20mm or 27mm reflectors for each LED ($$$) or just the tight Fraen optics (a little cheaper each). I never planned on using the original sealed beam reflector to direct the light. That would be useless! I will have to use individual or group optics and reflectors. Now there are so many combinations of LED's and optics to choose from! Aighh!

I thought of different LED's to use. The 45/75lm K2's will work beautifully because of the efficiency and brightness. The higher current 120/130lm K2's (if I can get them) will reduce parts count. 5W LS's are rather pricey unless I can find some for a great deal. The Osram Ostars are appealing! Unfortunately, I dont know where to buy them. They are about $80.00 each though according to someone who bought one on the CPF somewhere (you can probably search for it). They spew out about 420lm each and consume over 20W. They aren't very efficient as the 5W and dont compare well even to the 5W when it comes to 'lumens per dollar'. It will significantly lower the parts count, though. Used with the McR-45mm reflectors, the size of the 6 dies won't matter as much because of the immense throw and light capturing by such large reflector. This will raise prices though, yet there will be so much extra room left in the headlight assembly. I can fill that space with cheap LEDs and optics just for looks (daytime running lights?).

Hmmm... I was searching and I found a short article by Osram describing how they used the Osrams in a prototype Saab Areo X concept car. You can read about it here:

http://www.osram-os.com/news/news_saabaerox.php?lan=eng&id=127

Here's a quote:
"Only two OSTAR® LEDs for high beam and one for dipped beam were needed. Golden DRAGON® LEDs in red and yellow provide the indicator and tail lights."

According to that article, adequate front end lighting isnt as hard as I thought. It says that two of the 420lm LED's were needed for a bright beam in each headlight! Wha?!? Only 840lm per headlight is needed? That means 1680lm total for the brights! Also, that is 840lm for just the low beams. I seen that the good 55W Xenon lights output 1500lm (though more candelas, of course). Now is that 1500lm per headlight or for both headlights?!? HID outputs 3000lm! That cant be per headlight can it? In this case, only a cluster of 9 or 10 of the 80lm Luxeon III's are needed per headlight, right? The 24 45/75lm K2's are overkill then maybe.

Anyways, I dont know. Please leave feedback on what what all of you know about the brightness of individual headlight fixtures or both headlights combined.

All I can say is hmmmmm....

I could have a huge-a-mungo cluster of underpowered 1W to get great efficiency (easily past 40lm/W). That way I can achieve any desired brightness that may be needed (imagine all of them on at 350mA when they are normally on at 150mA or so to meet the correct brightness requirements). Okay, I am dreaming again. I hope I dont put these headlights together as well as I put that BIG word up there together. It would be a tangled mess of wire and scrap metal, he he.

As for being legal, I believe around here in Ohio that as long as it looks and works like normal headlights, it isnt way too bright or dim, or it isnt the wrong color of lights, you shouldn't be stopped by any police officers. I got away with two 3W fog lights I rigged up on my truck and white LED running lights. That's no big deal, but still. My brother still hasn't been stopped for his red front corner/turn lamps on his Grand Am (which actually looks pretty awesome - its hard to explain). As long as it doesn't flash red and blue or isnt too dim, I believe that nobody will care here. I could be wrong and get by friend in trouble which I hope does not happen. I think it will be okay. If it doesn't work well, I can always buy the lights off of him and use them elsewhere or sell them. Off-roading lights on my truck would be nice. Now I would just need to off-road more.

I do not believe that heat should be a huge issue. I planned on using like a 1/4" or 3/8" copper slab (or aluminum if I have to) as a heatsink. I may be able to join strips of copper to the copper slab and create fins. Solder/brazing or epoxy should hold well maybe. It wont be as nice as a commercially made heatsink, but it should work for the number of LEDs. I can insulate or isolate them from the engine compartment and direct cooler air to it. It can take air from the hollow fenders perhaps like the cool air intake kits. I can also use blower fans. I understand that standard brushless fans aren't very durable, but heat isnt a big problem in his 4 cylinder S-10 engine compartment. I believe that the heavy duty 100+cfm 120mm cooling fans that I found out there will do fine. I need to make a shroud so the air moves throughout the whole 140mmX200mm (or possibly smaller) heatsink.

I dont want to go against what you are saying, idleprocess. I dont think that the LED's will get that hot and only give half the advertised light. I know these wont be premium LEDs, but they should still be at least 75% of their advertised brightness. What do you think? Also, I think I know how to protect the LEDs when I use individual emitters. I will be sure that the heatsink will stay electrically neutral. I think that the slugs of each Luxeon are neutral, right? With careful use of thermal 2-part epoxy, I think that they will be isolated. The lead pins won't touch the heatsink with my plans. I may get a thin (0.5-1.0mm) sheet of black or white plastic with holes punched out for each slug or the whole emitter. The lead wires will then be separated from the heatsink. I also have a couple of small, thin sheets of mica from an old soldering iron. Everything is okay with the heatsink with four epoxied Luxeon III's in my modified maglite so far.

If I can't get a power supply together that is run by Mosfets, I have had success with the el-cheapo 12V regulators that you can find around the internet. Adjustable 12Vmax voltage regulators work also. I will just have to find other means of regulating the current to a fine point. I will have to group up a few of those regulators to provide the right amount of current, but that's alright. Those regulators do iron out the voltage spike real well. Ill figure something out. Please give more advice though. This feedback is great!

Finally, what is the availability like with those K2's? They are my best choice so far because of their prices and flux. How many can you order at a time? If I could get some, how many can I get so far? Will I pay that dang premium price for them because of the demand for them? Will I have to wait countless months for them to be ready to be shipped? One watters are cheap but I am afraid to increase the current past specs for the needed brightness (I really dont want any more that 24 or maybe 29 LED's per cluster). Three watters may work and they are the same price as K2's about. The 5W LEDs are simply out of question now. My friend is mortal like us. He has a little extra cash, but I dont think he will want to spend any more than $400 or even close to it. This is just a novelty item (that is hard for light and LED lover like me to say). He understands how this is kind of innovative or daring (or something). I will talk to him later about his budget and I will show him my findings.

Alright, This is long enough so far! Please leave feedback about the circuit design, which LED's and optics to use, and finally the availability of the LED's. I just worked a midnight shift and it is past noon. Time to get some rest. Ooops. Thanks again!!!


-Tony

Gryloc said:

Wha?!? Only 840lm per headlight is needed? That means 1680lm total for the brights!

Click to expand...

You said it yourself. It's a concept car. It only have to look good in the exhibitions, not work in real life. I much rather take an old Volvo 240 estate than the Aero X out for a drive at night.

Gryloc said:

...good 55W Xenon lights output 1500lm (though more candelas, of course).

Click to expand...

Although I am glad you used candela and not candlepower, how many candela your 1500 lumen would translate into is up to your optics. You can certainly get 100 GCd out of that if your reflector is large enough and it's in vacuum, but I assume you already know that.

Here is an idea. its based off of the hids. this way you can use the 100/lm watt Nichia's. Use a large projector. the reflector will collomate the light into a wide beam, and the optic focuses it. very efficient what to aim light.

hizzo3,

What do you want to achieve with that? Regardless of the reflector and optics you use, you would need hundreds or thousands of them to be useful on a car.

Hey! Thanks for the giving your ideas. I thought of something like a replacement LED bulb for the headlight. It should work, it is just a matter of heat management and getting the right people together that can machine it.
Check this out (go to post #8 if you don't want to read it all - I drew it pixel by pixel):
https://www.candlepowerforums.com/threads/107743

A projector isnt quite for me this time around. It's an idea, but I would have to research then experiment with the right optics and lenses. This is a sort of easy job (maybe more costly) with just a number of Luxeons with individual optics. I found I might just have to stick with Luxeon III's because of the lack of the K2's availability. I may go with the Osram Dragon or Ostar or something. Can anybody give me a list of high flux LEDs made by different companies? I cant think of them all. Depending if they are cheap enough or much brighter, I might go a different route from Luxeons (sadly).

It isnt that bad of an idea to use a crap-load of high brightness 5mm LEDs. I could fit a little over 1000 (25 by 40) of the LEDs in my 140mm by 200mm headlight. Given each one would emit roughly 3-5 lumens each (maybe the 40000mcd ones) and they have a 10-20 degree viewing area, they would rival HID. The only problem is the massive wiring mess with 2000 pins and many dozens of separate banks and grids for powing them semi-evenly. Color-eveness would be difficult unless you go with premium LEDs. It would be nice to produce 2000-5000lm at only 560mA@13V and 78W of power though. I checked ebay and they cost about $160 per thousand. Its an idea, but again, not for me.

Anyways, has anybody thought of a circuit idea yet? I searched around and thought that the voltage regulators would work. I found some made to handle 3A and 5A! I can have two to provide an even 13V or so, then have another set wired as current regulators (the nice thing about the LM317, LM150 and LM138 series). It is just a little hard to find the 5A and 3A ones for sale and in stock. The nice thing about those regulators is that they are variable, also. I can have two pots to adjust both the HI and LO beam for each headlight so I can fine tune the brightness. I drew up a circuit using the LM### regulators, but I am hoping someone can find a cheaper and more efficient driver that uses Mosfets or the like. I just need two circuits that can handle about 70W of power each. Hmm. Keep the ideas flowing! I appreciate it! Thanks...

-Tony

Gryloc said:

Can anybody give me a list of high flux LEDs made by different companies? I cant think of them all. Depending if they are cheap enough or much brighter, I might go a different route from Luxeons (sadly).

Click to expand...

If you look at the "LED DB" link in my isgnature, it's a (dated) collection of all the white LEDs I could find spec sheets on about 12 months ago. Just look for power consumption >=1000mW and those would be your high-power models.

You can check out the new Lexus which uses LED headlights and is an upcoming production model. Maybe you can get some ideas.

https://www.candlepowerforums.com/threads/114368&highlight=Lexus

skipping the points covered regarding 'street legal' in the usa...

One of my current hand-held projects it to take my megasonic and upgrade it to 14x K2 emitters.. driven as hard as the driver i have arranged can push them.. over an amp.. perhaps 1.5A..


A computer model.. the circle is 3.5"..just about the size of a headlight.

according to the data collected here the K2 will output between 106 and 131 lumen between 1000 and 1500 mA..

That is 4.2 to 6.6 watts per emitter.. or in my case.. 14x = 59 to 92W.

I run UX1Ls at about 1100mA and extrapolating from the curve.. that will get me about 120lumen @4W... I get 2500 lux per emitter.. or roughly 21 lux per lumen with the reflectors i'm using.. with cheaper reflectors i get 4/5 the lux.. so.. say 16.8 lux/lumen.

now.. take 14 K2s.. which arrange ok as shown by the picture above..

at 1A you are talking about 59W and 1484L. Using IMS20 reflectors ($1.20 each in quantity).. you can expect 16.8lux*1484 lum = 25k lux!

bump up to 1.5A.. now it's 92W.. but 1834L.. that will net you 31 klux.

If i designed this system i would put a zener diode of about 4.5V across each emitter in case of an 'open' failure.. the driver would probably cost $50 to $100 to build but K2s will be about $3.50 in quantity.. so under $70 per assembly for emitters and reflectors.

with the car in motion the airflow will be able to cool hundreds of watts.. cooling will be a non-issue.. have a dimmer to dim the lights when not moving even just to 1/2 power and just mounting all of them on a big alum. finned heatsink like in the photo will be enough.

K2s are high Vf.. not the best lum/watt for those types.. 4.2 to 4.4V.. so about 30V is required if you run them series-parallel 7x2.. 2-3A at 30V.. lemme see if that can be split in a 'car power' way and make a really easy driver:

well.. 13.8v car power.. that gets you 3.13 emitters.. so maybe a slightly looser array so you can fit in 15.. and have 5 sets of 3 emitters..

that would mean 5A at 'low beam' and 7.5A at 'high beam'.

the hotdriver would handle these loads with no problem. it could be set 'current limited'.. and putting a balancing resistor on each leg would mean you would not have to drop that much voltage on the FET.

Hotdriver.. $45. It would be exceedingly easy to get a test case set up... the beauty.. when the car's not running and you get 12V... the output will drop dramatically causing far less heating on the emitters!

the power loss on the FET.. current x vDrop..

13.8V-(4.4x3) = 0.6V x 7.5A = 4.56W on high beam.. not a problemo! efficiency of about 95%.

13.8V-(4.2x3) = 1.2V x 5.0A = 6W.. also not a big deal.. efficiency not incredible but doable.. 91% or so.

Thanks for the idea actually. i've been trying to figure out a good way to make luxeon lights for car tail lights.. and a cluster of 3 luxeons with a hotdriver would be a reasonable solution. (much more cost effective for me naturally).

Thinking about it.. another really reasonable solution would be to make each bank have it's own little LDO.. a nano board on each.. not nearly as fancy but current-limited..

In any event.. absolutely doable.. absolutely worthwhile.

-awr

High and low beams require different beam angles, so just changing the current isn't sufficient.

Greg

OK.. since the numbers work out much nicer for a 3x4 array.. i re-did it with 12.. too tired to re-work the math but i think that with 2 less emitters it would still be a viable solution, and lots easier to design the layout and the banks of emitters.

-awr

greg.. could just put a tiny servo motor in to change the angle.. then you can have a dial just like on a motorcycle.. in-fact having a dial where you changed from bright to dim and ranged the angle at the same time.. not difficult to design or program for anybody with RC servo experience.

-awr

www.edison-opto.com.tw

They have a 5 watt 200 lumens @ 10,8V@500mA led's

and 10 watt and 20 watt led's

The 3 watt edixeon emitter delivers 120 lumens @ 1000mA.

almost the same specs as the k2 , but the k2 can be driven harder.

m

Eliptical optics is what is needed, so excess light isn't throw at the sky/drivers eyes or in front of your bumper.

Also the K2 is not efficient, at least not at this point of time. What's coming out now appears to be 100lm @1000ma, and has a vf of L at that current. Some are even N vf. That's 3.75-4watts of input power. 25-27lm/watt. Add to the fact that it may be the 85lumens and not 100 typical, and figures get even worse. Only time you get better efficiency is underdriving the parts, which in that case you are goign to be looking at tons of emitters.

The luxV seems to be the most efficiecnt, but they have a short lifespan and in the hot enviroment (summer, close to a hot engine) they would suffer even more. I would defianltly use some active cooling. Perhaps even liquid cooling..

Probably LuxIII emitters on a copper heatsink with a fan or liquid cooled would be a reasonable approach. I'd use artic alumina epoxy to isolate them, and test each one. If you run them in parallel, or sets of parallel on seperate heatsinks isolation isn't too much of a concern. Use eplitical optics with a horizontal beam.

Hmmm...

Right now, the driver circuit seems like the most tricky area. I wish I knew more about mosfet circuits because I know I could build a high power circuit for cheap if I used todays high power fets. A while ago, I found this site:

http://www.lasertagparts.com/lightbrain/2002_twin.htm

They sell simple bike light dimmer circuits that uses high power mosfets. I believe that they may not sell their products anymore because the site hasnt been updated since early 2005. These products can handle up to 100W max. The only thing that wouldnt be practical for me is how they oporate. They have different brightness settings and other special battery saving devices. The Hotdriver seemed like a nice circuit. I again just need something cheap that I can set 2 different current levels with voltage coming in 2 different imputs (hi + lo). I hope I can adjust the current level for each input via a cheap potentiometer. This way I can fine tune the outputs for the different beams. It is so hard to describe and it is difficult to find different circuits online that would work. Please help.

Anybody with experience with simple mosfet circuits, please help. I thought about the LM317 voltage regulator that can also adjust current when wired differently. They have beefed up models that can handle 3A (LM150) and 5A (LM138). I thought about having two circuits that work together that can give the desired results. One circuit can provide the exact voltage I need. The other can adjust the amout of current to the LEDs while using the correct voltage.

I thought a mosfet hooked up to a 555 timer would work. I Put together a 555 dimmer once before and it is powerful and easy to build. The only problem is that as I slightly dim the circuit, the light seems to flicker a bit because of the low frequency pulses from the 555 timer. I see there are some PMW IC chips out there, but I dont know what is right for me and how to use them exactly. Again, any help is apprecieated.

Thanks Mark for the idea. I looked at Edison Opto LEDs in the past and wasnt blown away. Now, their three and five watt LEDs are very bright and efficient. Also, I seen that they have the 10W and 20W LEDs! I saw an announcement somewhere that Edison Opto is working on a 40W LED now! Holy crap! I think that those LEDs will work nice, even though the LED dies are a bit larger. I thought of an ingenious way to modify the elliptical Fraen optic to work with these bigger emitters with bigger light emitting areas. I think that I really want these LEDs, only my big problem is where the heck do you get them? I do not see any distributors on the internet so far. I want to call the 800 number, but I realise that they are about 8 hours behind us (Ohio) in time. I hope they arent too expensive and they sell in quantities fewer than 10.

Ok, the White Edison Opto LEDs have the following specs:

High Power
Model__Vf______I________Power___Flux____Efficiency___# Needed/Lumens
--------------------------------------------------------------------------
1W____3.7V____350mA___1.3W____55lm____42lm/W____28/1540lm
3W____3.8V____700mA___2.7W____90lm____33lm/W____17/1530lm
....\___________1000mA__3.8W____120lm___32lm/W____13/1560lm
5W____7.5V____700mA___5.3W____160lm___30lm/W____10/1600lm

Ultra High Power
Model__Vf______Current___Power___Flux____Efficiency___# Needed/Lumens
--------------------------------------------------------------------------
5W____10.4V___500mA___5.2W____200lm___38lm/W____8/1600lm
10W___17.5V___645mA___11.3W___400lm___35lm/W____4/1600lm
20W___18.5____1120mA__20.7W___700lm___33lm/W____3/2100lm


So there ya have it. Not bad specs. The high Vf can be worked with if I really need to. A step up converter would help out (like the Shark). They may not be as awesome as Lumileds (or something) but they look like they will get the job done. Anybody have any info on Edison Opto?

Well, I could go with a bajillion 1W luxeons right? I found this place selling Q-binned batwing (but possibly lambertian) luxeons for $1.25 each. I dont know how many is left though. Here is the link...

http://www.tek-tite.com/src/product_info.php?id=2985

30mm 15X30 degree Fraen optics with holders sell for $1.75 a piece at Future Electronics. That means I need about 50 1 Watters. It will cost me about 150 per headlight if I go this route... Never mind. This is why fewer emitter will save me mula.

Well, I plan on tilting the headlight module in some manner, or even have different sets light up with different optics used for either high and low beam. I do like the olliptical optics because of the beam pattern and the price (compared to the IMS 20mm and 27mm reflectors). Luxeon III's are very appealing because of the price, until the K2 come out. I hope to start this soon since I have a weird summer job approaching for a month, then I have to start fall courses again in late august.

Right now, since my friend is planning on getting a beefed up alternator for his truck for his sound equipment, efficiency is not a big concern for me now. It only means a difference of 20W or a few amps possibly. Heat is taken care of. Right now, a one of a kind LED headlight just needs to make and throw a lot of light.

Andrew, where did you say that you found those cheap reflectors? Whats the quality like? Beam angles? Price? Anything to save a few bucks would be nice.

Finally (sorry I leave long, drawn out responses), How big of a concern is isolating every emitter or star? Some people go nuts about this. I have two flashlights that use clusters of Luxeons that are attached to an aluminum heatsink. I am not bashing the idea, but what does it do? I never saw any negative effects. I dont even think that there is any continuity between the aluminum base and either of the + or - pads, right? I wont fret over it too much. I will probably epoxy them anyways. Again, thanks again.

I think to use several LEDs instead of a single-point light source you'd have to route the photons through a fiber optic. This way you'd be able to use relatively available reflector technology. With the cost and availability of large-gauge fiber I'm sure this would significantly reduce the R&D you'd have to do in order to make this happen in a one-off application.