Just off the CNC - Update: It's Alve!

THANK YOU!
I'm an electrical engineer, and all of us EE's are forced to be mechanical engineers whether we want to or not..

PPtk

Oh, I'm far from a perfectionist... Too many years of getting things done, not getting things perfect. Perfect comes at the cost of time - not a luxury I usually have.. I always strive for 'as close to perfect as I can get in a reasonable time period'. Thank you though, I'm proud of my project - perfect or not.

I only wish it was me being lazy that has held this project up a bit. I can't tell you how much I'd like to sit on a beach with a Johnny Walker Blue and have people point and call me lazy :) Unfortunately, what has held this project up has been work - I've been absolutely buried the past month or so.

It's coming though - results soon!

PPtk

Ahhh . . . TIR, so that's how you're getting the throw you need. And, you're right about the Dry; mine pulls near 5 amps at the tail cap - don't know how much each emitter gets . . .

alotFoy

5 Amps? Holy crap. the DRY has 3 18650's right? If it does, that means that each emitter is getting 5 amps! (subtract a LITTLE for driver inefficiency, but it's a few percent at most)

I can't believe XM-L's are not just blowing up with sustained driving at 5 amps! It's also pointless, since past about 3.5 amps, XML emitters convert almost all of that extra power to heat, not light...

PPtk

P.S. - The driver circuitry I designed could easily support driving each XM-L to 5, even 6 amps.. Once I have everything debugged and working, I might crank them up to 4 amps per to see if it makes any (visible) difference.. It would also be trivial to put a few SST-90's in place of some XM-Ls. The only thing I would have to do is change the inductor and double the output capacitance (easy), and then my driver could happily give 9 amps to each SST-90 :) Heat would be a major problem though. It would be fine for short bursts, but there is no way that my enclosure could dissipate 720 watts of power without a shit-ton of airflow. If one needed a ton of light during a hurricane, though, it would be perfect :)

Many Drys (including mine) have a 20 second step-down UI. 20 seconds in "turbo" mode and it steps down to high and you're right; Match did a chart that shows the gains above 3.5 amps are small/don't justify the heat/emitter abuse. But, with 3 x 18650 the Dry does indeed feed each emitter with that much current.

Since there's no free lunch when dealing with extreme output, I guess that's why I'm so impressed with how you've addressed the necessary trade-offs.

amazingFoy

That makes more sense.. 20 seconds worth of 'turbo' mode can be dealt with thermally. Much beyond that, you'd be carrying a small toaster that happens to output a bit of light :)

As for addressing the trade offs, well, I realized many years ago that life is a compromise. Engineering a light such as this is no different - a set of compromises that work well together. Hopefully I have chosen well, and none of the variables turn out to be different than expected - but only time and 240 watts of input power will prove that :)

The other cool thing about the driver I worked up - it will take from 10V to 120V input. When I'm testing, I can run the thing from a 100V DC Power supply, and it will only draw roughly 2.4 amps - this way, I can test and debug without having to use massive input wires.. 18ga will be fine. Once in the vehicle, I'll be using 6ga to lessen the voltage drop. I designed the driver to have the highest efficiency at about 14V - typical of vehicle with the alternator turning.. At 100V, it will make a little more heat, but we're talking about a watt or two - nothing substantial compared to heat from the emitters.

Thanks so much for your kind words - I just might have to send you one of these to 'review' :)

PPtk

LOL

Oh yeah . . . would look killer on the front of my Passat.

And, as for the SST-90s . . . a big die that needs a lot of juice to perform and I doubt there would be any noticeable increase in brightness. I also agree that there would be little if any advantage using a few XP-Gs, other than less light. Once it's up and running, I predict you won't be saying, "I wish they threw better." There's going to be some reach for sure but I suspect aiming won't be an issue. It will basically be this insane horizontal of lightning from a distance and since it's flat, it might illuminate the shoulders of the road pretty well.

Actually, the more I consider aiming, the less sure I am about what you'll see from behind the wheel. I'm a novice so, this is probably all redundant to you. As is everything on the base being able to withstand over 170 degrees F sustained. Is that about right? A little high, perhaps?

Sorry - just can't seem to stop thinking about this . . . pretty damn awesome.

Foy

I felt the same way about the XP-G's - that's why I decided to kill them off. The might add a little throw, but it will be completely insignificant when placed inside the wall of light that should come out of this thing. SST-90's certainly DO put out a little more light, but they do it with (as you say) a bigger die, yielding much more spill and much less throw. They're also quite a bit less efficient (twice the output with 3x the input)

The beam pattern is not going to be horizontal, it will still be a circular output. The width of the enclosure is only 15 inches, essentially nothing once you get out more than a few feet from the light.

Every single component that I've selected (metals, plastics, electrical parts) is capable of sustained operation at 125C (257F). Even the PCB is built using upgraded material (FR4-TG170 instead of regular FR-4). My cut-off will be 100C, which is where the light will automatically start backing off to prevent damage - but this is a very cautious number. 125C really is no problem at all for the parts. Even 100C should never be seen though - the thermal analysis suggests that absolute max will be around 80C, and that's in a VERY hot 90F ambient environment with almost no airflow.

PPtk

15-inches. I guess I need to read a little closer. I had it bigger in my mind for some reason. For the light you're going to get, that's pretty small. Actually kinda funny when you think of how big some Cibies are.

Anyway, looking forward to seeing it . . .

thanksforsharingFoy

All understood still very keen, can't wait to see your finished product.

:) First signs of life. I had a little spare time tonight, so I started putting components on the circuit board and doing some basic testing. So far, all good news. The op-amp based current sense circuit works perfectly - I'm only using a 0.01 OHM Current Sense resistor (to mitigate power loss in the sense circuit), and the 50 to 1 op-amp works beautifully. At 3 amps, I'm seeing 1.5143V out of the sense circuit - well within the 1% component tolerances - and it's stable and quiet.

The DSP is soldered down - always fun soldering a fine pitch QFN by hand with an iron. 3 XM-L's are in place, and the first lumens have been emitted. I ran it up to 3 Amps for just long enough to measure the current sense output, but with the paper-thin circuit board not being attached to a heat-sink of any kind, it gets HOT in a hurry. 1.5 amps is about the most I can run for more than 5 seconds without risking damage.

All in all though - a successful beginning.

Pictures of the process:

This is the 64 pin QFN part drowning in water soluble flux. The tip of the soldering iron is in the picture for size comparison. 3 XM-L U2 Parts are also visible. Those are a real pain to solder by hand - the only real way to do it is with two irons. They look good though - nice and centered. At the far right of the picture, you can also catch a glimpse of the 50:1 Op-Amp. I love these TI parts - very stable, and very tiny.

Side view of the pads of the QFN. Soldered up nicely, considering it was done by hand rather than reflow. You can also see a few 0603 caps and one 0805 cap.

A picture of the emitters through the TIR optic. These Cute-3-XML optics from LEDIL are very well made, I have to say.
You can see how perfectly centered the emitters are in the optic though.

The emitters lighting up for the first time. This is insanely low current - 0.001 Amps (1 milliamp).

Emitters fired at 50mA with TIR Optic in place.

Emitters at 1.5 Amp With TIR Optic :)

Overall, I'm very satisfied with the results thus far. The THIN (0.3mm) circuit board was a good move - it allows heat to move from the emitters to the back side of the board (and into the enclosure) through the vias with incredible efficiency. There is no doubt that a THIN board like this with heavy copper and LOTS of via's will perform better than an aluminum star board. At 1.5 Amps, Its less than a second before the back side of the board is WARM - which is good - it means the heat is moving away from the emitters rather than staying trapped in them.

More tomorrow, I hope.

PPtk

Looking brilliant :)

Some of us might be interested (at least I am) to just buy the circuit board and stuff without the parts installed. You should take this also in consideration. The buyer will install the parts at his own risk. Less work for you, more satisfaction for the others.

Very nice project indeed.

Maybe a smaller light (bike light) for the future ? :)

Great build along, thanks for taking the time to share it with us.

This looks excellent.

If I can offer an idea for the future ... perhaps make the circuit board modular? 1 module for 3 XM-Ls? That way anyone can order 3, break 1 at their first try at soldering, and later use 2 for total of 6 XM-Ls. Or similar.

You get the picture.

Viktor

That's one thing I've enjoyed learning since I began this "flashlight thing" the need for a thermal pathway and the different ways to route heat away from the emitter/board or whatever. Nothing to absorb/move heat = very bad.

Again, I apologize if you've already answered this but am I right assuming the part of the TIR next to the emitter does not actually touch the emitter? Is it dished out a little or flat and mounted held away or . . . . does it touch?

I find myself wishing the TIRs were larger in diameter and maybe a little deeper . . . . so, I guess I have to ask; how big are the TIRs? Not a huge deal - just curious.

Foy

The need for a thermal pathway is so often either misunderstood, overlooked, or written off as too difficult. Without an excellent thermal pathway, you could have a heat-sink the size of an aircraft carrier and it will not matter - if you can't get the heat from your source to your sink, then your sink is irrelevant anyway.

The TIR's that I'm using are 35mm in diameter (for 3 emitters) and about 15mm high. bigger ones are available though! The Iris comes to mind ( http://www.ledil.com/node/2/p/2651 ) at 38mm in diameter and 28mm tall for a single XM-L.

The Khatod PL353CTL is also a big optic at 50mm in Diameter and 22mm in height for 3 XM-L's ( http://www.khatod.com/cms/pl353ctl_series___triple_lenses_for_cree_xlamp_xm_l_leds-2705099-2705099.html# )

TIR Optics are also different animals than reflectors. While bigger is still 'better' for the beam, it's less pronounced with optics than it is with reflectors. If the design is good, an optic shouldn't 'need' to be huge to produce a very nice beam profile.

The optics that I'm using (CUTE-3-XML) are hollowed out over the emitter - they don't 'touch' - but they're very close. I would be hard pressed to say that there's a full millimeter between the lens of the XM-L and the TIR optic itself. Even this small gap, however, helps to keep the heat from the emitter from dissipating into the lens rather than the circuit board.

PPtk

Have you seen any MPCBs suitable for the 50mm Khatod?

That optic looks like a killer piece for retrofitting triples.

FWIW, the 3-up CUTE optic is what I used on my last triple conversion. Very efficient and thermally stable.

man i'd love to install one of these on my scooter lol..seriously! would have doubts about supplying it with enough power though..

What an amazing thread!! It's really interesting to see the engineering thought process behind the design of the light. I wanted to go into mechanical engineering when I was I was in college, but my math just wasn't good enough. I ended up in IT doing business software instead. (once a nerd, always a nerd).

It will be interesting to see the beam pattern once you get it all together and running at full throttle. Although your design isn't conducive to a curved light to create a wide beam, having a set of emitters at each end of the light set outwards at 20 to 30 degrees might provide more peripheral light. Sort of like a built-in cornering lights. (Please understand that I am NOT arm-chair quarterbacking! Just thinking out loud.)

Can't wait to see the pictures of it in action on your Jeep. Aren't you concerned that the light beam will catch bushes or trees on fire as you drive past?

Thanks so much for taking the time to share this process with all of us!

Joanne