DIY Light Bar - 40 x XM-L2 on copper.

Any updates to this project? I am interesting in building something like this for my father, but maybe scaled down a little bit. He's been wanting a nice light bar for his truck for a long time, lots of nighttime driving on farm roads, etc. Wanted to see if there was any long-term reliability issues with the driver and other components.

Im still waiting on reflectors to arrive before moving forward. I doubt there will be any issues with the drivers, as long as they are used as intended. The circuit shouldn’t be energized unless connected to a load (sufficiant load resistor or LED’s). It isnt protected from a dead short so care must be exercised. The fet requires a large heat sink, and size would depend on the voltage differential and what must be dissipated as heat in watts. The guy who built the drivers stands by his product and has been very instrumental in giving me the confidence to make this total investment. He makes nothing on his sales and puts them together as a meticulous enthusiast to please others.

A lot (if not all) guys are going with TIR optics in their light bar builds, which isnt the best option for a thrower. Although they do throw moderately well when several are combined with overlaying beams, they wont come close to competing with reflectors that properly focuses the emitters. The reflectors greatly complicate the build, but I have a few ideas how to overcome the limitations without needing a precision reflector alignment girdle to index them and hold them in place. I already have a pair of 100W HID wide angle lights that provide plenty of wide area coverage, so the LED’s are only for longer range. Otherwise, the TIR’s would have been sufficient and a whole lot less complex.

Another obstacle might be in rearranging some individual LED’s to balance the vF of the parallel strings once the heat sink warms up. Its going to depend more on the evenness of the temperatures across the sink and taking voltage measurements of each string once the sink heats up. Im hoping that in using this mega 20 pound well designed sink, combined with copper noctigons and all the emitters from the same batch - should help alleviate some of those potential issues. In testing, Ive found that vF can fluctuate greatly with temperature. Another good article from the reef guys takes a lot of the guess work out and provides some innovative solutions: JP's DIY LED Fixture - Reef Central Online Community but this is operating in a static environment… not dynamic with changing air temp and speed passing over the sink. Ideally, the diver would be of a boost design and provide for huge serial strings (75 - 150VDC), but that doesn’t yet exist.

This isnt going to be a quick build. I still need to rearrange my existing aux lights, wiring, relays, switches, mounts, swap in a different alternator for +200 amps @ low RPM; along with the custom brackets, pulleys, fuses and wiring to support it. It will all get done… eventually.

Sounds good, I look forward to hearing your results. I have been playing with a few different TIR lenses due to their smaller size which would make things a bit more compact and easier to mount. There are a couple that throw decently well but none of them are as good as a reflector IMO.

You are right about the emitter vF...as they warm up it gets noticeably lower. If you put an XM-L on a poor heatsink with a weak cell you can watch the amps jump up as the LED gets hot.

Do you really foresee needing to balance emitters in strings? How much variance have you seen? I figure that as long as they're within 0.25A-0.5A of each other I wouldn't worry about it too much at the level you're going to be running them.

There are some decent high-wattage boost converters I've seen on eBay and tested on Youtube, but I worry about reliability vs. this simpler (and well engineered) linear driver. I think that if you had a few of them to spread the load and replaced and upsized the capacitors they may work alright. I may have to get one to try it out, just to see. It would definitely simplify wiring but it may also bring out other issues. I'll probably just end up using this driver.

Where did you get your heatsink? I have bookmarked a few places on eBay and at heatsinkusa.com

I wont know what to expect until I have it up and running, then take several voltage measurements of each string at different temperatures (while wearing welding goggles). :bigsmile: The reef link illustrates what often happens in very meticulously built large arrays that operate in a static environment (which makes them even easier to vF balance). Its a long thread but worth the read. It doesnt take much to wind up with a string that hogs up most of the current.

While individual emitter vF might vary, the greater problem I foresee in this particular type of build is that a fully populated sink typically wont cool 100% evenly across the entire surface of the sink. Since I will be running 5 parallel strings of emitters per driver (4S5P) x 2, there is the potential for some strings to ultimately run at a higher temp, which would create a situation where they keep drawing more current and creating more heat in those locations (which will lower vF even more and cause them to draw even more current and create more heat). Im sure you already know this but thought Id mention it for others. The XM-L2 on copper should withstand a good amount of abuse in a worse case scenario (6A), but Id like to have even output from each emitter string, regardless of the sink temperature. They will be running above 3A each so it narrows the margin for error. After the emitters are in place, I’ll do my best to wire them in such a way that each string shares a symmetrical pattern on the sink from each group of 5 strings. I might interlace them with the other group of 5 from the second driver. Hopefully, that will create an even heat distribution pattern on the sink so they all share the same vF gradient as they heat up and cool down with the varying conditions. At that point, I can vF balance the strings if necessary and be done with it.

While I was trying to vF match my 50 emitters, I became frustrated at how quickly vF lowered directly after power was applied and also how small sink temp changes made a significant impact. After coming across the reef forum, I decided that the best way to test would be after all the emitters were wired to the actual sink and while operating in that environment.

After spewing all of that verbiage… maybe it doesnt matter at all. :bigsmile: Now wouldnt that be nice? But the reef posts indicate that it probably will matter.

Like you, I checked several other drivers and have a good many of them saved in my ebay favorites and other places. While a few of them claimed to be suitable for automotive applications, none of them made mention of voltage spike suppression or thermal management of the driver, or thermal monitoring to throttle power to the device it would be powering. The boost drivers are only rated at the higher amps when they boost only a few volts above vin… which nullifies their advantage for large arrays. Given the low cost differences to power ALL the emitters and the lack of support from the others (and how the Chinese often blatantly lie about their specs), it became clear that there is only one provider capable of delivering such a driver for an array of this size. On top of that, he’s a serious enthusiast that went through a great deal of time and growing pains in developing the driver to make it work as well as it does (check the cpf link in the first post). He also went far out of his way to answer all of my questions to educate me in how to best approach this. Im also engaging in a parallel build project for a landing light array in a custom show plane… so this must be reliable and not endanger the electrical system.

If you find other drivers, please post. Id love to find a cheap boost driver that could efficiently convert 14.5VDC to a stable 36-72VDC @3-4A, not overheat, be reliable, offer voltage spike suppression, driver thermal management and slaved thermal management.

I bought the sink from ebay but it was a one time deal. I prowl often, so If I find another one I’ll let you know.

That makes sense, you could have a positive feedback loop runaway situation. Your interlacing idea sounds like a good yet complex solution. Like you said, it will be hard to tell what really happens until you get it all hooked up and test it. It could be a bigger problem than anticipated or it could end up being much ado about nothing.

His driver is definitely well thought out and understood at least by its seller...the Chinese boost / buck converters are definitely suspect but I wonder if a few additional part tweaks (replace capacitors with higher quality components, etc.) would make them usable. The downside of a boost driver is that if something goes haywire you could end up frying all of the emitters, which would be almost impossible with your setup.

There is a First Person View FPV, RC pilot and whizbang that did a flying searchlight to fly his airplane at night and he worked out some similar problems to what you are facing. There are some big differences along with what he has done but I bet you could incorporate some of his thermal management techniques to protect your rack. He created a driver and a pump for liquid coolant that all fit in the remote control airplane. You should chat him up, I bet the two of you could really make something incredible. Here is the link.

I wonder if you decrease the number of emitters and ramp up the amperage while using active cooling if you might be able to accomplish more with less. Good luck, I hope you get the information that you need.

Thanks for your ideas and link. Although hugeone’s design challenges are quite different from my own, there were great lessons to be learned from his build thread that I might utilize in a future bicycle light. As RMM and I had mentioned, the vF mismatch scenario might or might not materialize. Either way, I can follow the well explained procedure laid out by our fish tank buddies in the other forum to vF balance the strings and get back on track (if necessary). Because of the voltage requirements of the linear driver, Im stuck keeping total vF lower than input voltage if I want to run in regulation. Given what the drivers can provide, watt per lumen efficiency and the size of the sink, 40 emitters in 4S5P x 2 is the ideal scenario. The large rectangular shape of the sink will likely have the largest influence on vF since the emitters in the center could hypothetically run warmer than the ones towards the ends of the sink. But this is just more speculation. Im using a massive overkill sink in combination with a CAD designed emitter placement stencil to precisely locate each mcpcb on the sink and to keep the distribution perfectly symmetrical. Hopefully, this will also result in symmetrical heat distribution… and I think it will.

Flying season has just started, which has been eating up a lot of my illumination play time.

Any updates?

40 xml2’s you’re nuts :bigsmile:

FYI 26.5mm Carclos do throw! Maybe not as good as say a C8 but, they still throw. Here is 24 xml 5000k t5’s, 18 spots, 6 elliptical (4 of my 8” light bars).

Middle dune is around 250 yards. The far one is over 500!

It’s like rolling daylight :smiley:

You got your lights working again! Did PTK fix the driver issue or did you have to sub? PM me if you must. I see you’re still running around in that crazed friggin rig of yours, and it apparently hasnt been rolled or blown apart yet. :bigsmile: That is just so bad ass and I love those vids you posted! I just got my vette turning 1.23Gs and finally sticking to the tarmac under WOT, so I feel your LS1 passion when you can keep it running in the power band during most of a run.

Your lights are kicking some major ass, but they are also projecting across a white reflective surface. Still, its very impressive - especially considering the TIR’s. Id never have guessed. I spent a great deal of time researching reflectors, optics, optical efficiencies and radiation patterns for various models. Then I ordered a bunch of samples and took them afield to see how my eyes interpreted each. I already have a pair of 9” reflectored 100W hid cornering lights I converted that pump out 160 degree arc to the sides and 200 yards forward, so I decided to make the LED light bar a pure spot to compliment the existing setup.

Sadly, its shelved at the moment. It turns out that the engineer that designed the drivers figured out a way to slave the thermal throttle controller so both drivers would act as one, which changed my core design. I just havent had time to get back into it.

There is also a parallel project Ive designed for a single driver 16 x XML2 aircraft landing light array for an unlimited category experimental that placed 2nd in national competition. He just wants something that will annihilate the off the shelf lights while saving weight. It will be active cooled with a 1/8” thick copper backbone plate mounted to a CPU sink (8 x XML in each wing with cockpit mounted driver fed via long 18ga leads to each array). Id like to at least get this one up and running (on the bench) first to iron out the bugs because it will be far less complex and not need to be water proof or survive massive G impacts. Lessons learned here will apply to the 40x.

Here’s where I could use some ideas and help:
For the 40x, Im a bit stuck trying to figure out how to mount 40 reflectors in a high G-shock environment without having to machine a retaining girdle to retain, secure and index them. That would cost a fortune to have done. Another member was kind enough to CAD an emitter placement stencil that I can have a shop cut out of vinyl and stick on the sink to allow me to precisely place the emitters and drill the holes for the mcpcb mounting screws. Thats 80 holes to drill and tap. Sheesh! I was thinking about JB welding emitter centering rings to the reflectors. Then gluing flat orings like these to the front of each reflector. Then mounting the large rectangular glass lens in a frame, and using a compression fit of the glass against the 40 reflectors with orings to secure it. I could also probably glue all the reflectors together with JB weld putty from the back side to stabilize it as a solid unit, then the glue mess would be invisible from the front of the unit. Its starting to sound like a redneck project which makes me take a few steps back. There’s plenty of room for error without an indexing girdle and I dont have the orings I linked. I already have all the emitters on copper, drivers, wire, reflectors, copper cut to spec and sinks… so the investment has been made.

ANY ideas would be GREATLY appreciated. Thanks for reading… its late and I tend to ramble when Im tired.

Love the Corvette!

:love:

The big lights(8” 6xml’s) have always worked well. It’s the new triple round xml2 light that is the problem.

It is probably not cool, and it would be a lot of work but I have been thinking about this light bar a lot, and I have an idea about the lens issue. It all depends on if someone can drill and tap the top your bar or the opening that the reflectors shine through. For simplicity to drill and tap to take the lens, bezel and gasket of a Solarforce L2(whatever). Those would be easier to waterproof than a channel of lights and reflectors. But that might increase the mass that I think you were saying you wanted to keep down.

Any updates to this project?

I could really use some help from someone with AutoCAD experience. I have a friend with a CNC machine that tells me he can cut out all of my parts to complete this project if I can provide input files and material. Can someone with AutoCAD or similar please help me with the drawings/input files? There isnt a whole lot to it. I could provide some detailed specs if someone could please help me.

Another member helped me get this far with a rough draft to conceptualize the project, but he no longer has access to AutoCAD.

There need to be a few revisions to the basic drawing along with a few other simple parts milled out.

FP I can try to help you

Done, PM me for the DWG file.

I can try to help you: 01:50 (new)

Done: 02:02 (new)

DAGGUM!!!

Thank you for offering to help me Daniel, but its not all that simple.

The drawing is a reflector placement grid for visualization purposes only.

1) It needs to be revised to make the reflectors equidistant from the edges of the heat sink while maintaining equal spacing between each reflector.

2) The heat sink will use 40 x 20mm Noctigon boards, which will need to be tapped and threaded (2 screws each board) to the heat sink in the center locations of each hole in the drawing. So, the only holes that would be drilled in the heat sink at this time are 80 small threaded mounting holes to fix each Noctigon board (40 boards require 80 small threaded holes to mount them).

3) The reflectors will be retained by a front aluminum mounting plate with 40 holes for each reflector (with a ledge in each hole for an oring and reflector to press against). The mounting plate will be a bit larger than the actual heat sink. I still need to get precise measurements for you. I will probably need to come up with a drawing and post it here for you to get a better idea to CAD from. Each reflector OD is 1.6142”. The holes will need to be slightly larger for differences in reflector size and I still need to find out the ID, depth and width of each shelf in each reflector hole to hold the reflectors in the holes. Also need to measure the thick aluminum plate sitting in the shop.

4) The reflector mounting plate will be fixed to the heat sink with 4 rectangular pieces of aluminum plate to box it in.

Im sorry if I lost you. Its not so easy to explain. This is all preliminary and Im in no rush. Are you still able to help? Ive never attempted anything like this before and know nothing about CNC equipment, so Im not sure what questions to ask the owner. Have you done this type of work before?

Thanks again.

20mm Noctigon

To help visualize, check out comfychair’s reflector retaining plate with the ledges in each hole to hold an oring and reflector. Mine would be rectangular with 40 holes as described above.