A Watercooled Headlamp, Because Why Not?

There are extremely high powered LEDs out there, and most of the ‘creative’ uses of these are extremely high-powered flashlights, complete with heatsinks, forced air cooling, and beefy power supplies. [Christian] wanted to play around with one of these LEDs, but he wanted something a little more unique. He chose a headlamp, a build that is made even more impressive by the fact it is watercooled.

The body of the headlamp was milled out of aluminum, with a space for the LED in the front and channels in the back for coolant. Also in this enclosure are two buttons, a temperature sensor, and a port for the hose that carries the tubes and wires.

This hose connects to a large battery pack that houses four large lithium phosphate batteries and a boost converter built around an Arduino. The pack also houses a pump and reservoir that is able to keep the LED cool even at 130W.

29 thoughts on “A Watercooled Headlamp, Because Why Not?

  1. Is there actually a radiator on the watercooling loop? I can’t see one mentioned and he mentions it gets too hot after a few minutes… surely he didn’t consider one optional?

    130W is similar (if not more than) a computer CPU, and looking at various watercooling systems you’d want at least a 120x120mm radiator with a fan blowing through it….

    1. You don’t need nearly that much radiator area — it just depends on how much airflow you want to have. However, with a good multi-pass copper tube 120x120mm radiator (not for a computer, this was for a power electronics application) and a 24vdc fan that screamed about 260 cfm of air through the radiator (this is about a $75 fan) and a nice little mag-drive gear pump, one can dissipate about an order of magnitude more — 1800 watts (in this application) — while staying around 80C in 25C ambients. However, this is very loud! The fan we used was about 6000rpm full tilt (it had PWM temperature control) and something around 60dBa.

      For this thing, I am honestly surprised he didn’t just try to make it convective. There are a lot of misconceptions about watercooling because of the people that do it for PC’s– it’s mostly all hokey, in part because the actual parts are usually pretty expensive.

      Having channels like that actually significantly reduces the heat transfer — he needs a pile of pins that make the flow as turbulent as possible without excessively restricting flow. The water tank should’ve been long and thin, and made up one “side” of the box…with fins machined into it, so you can just convect off the large area.

      100 watts is not a lot of power density over the area of that LED puck. Interesting idea, but…well, anyways. I don’t want to be seen as a hater, but there are a lot of weird decisions in this project. RCA for charging? The list could go on. He should pick up a little milspec >10,000rpm rotron fan surplus on eBay — I think Aximax or something — that would work dang well!

      1. Yeah, I guess in a computer application you want a big radiator so you can have less airflow and hence less noise, as well as run it at a lower temperature over ambient.

        I haven’t really researched computer watercooling extensively – I’m perfectly happy with my Noctua air cooler, and it does perfectly well even with a heavily overclocked CPU.

        As for the example though, you can see the difference – I don’t imagine a 60dBa fan would be very nice in your computer, nor would 80 degrees radiator temperature be terribly good for your PC components – but you’re right in that it doesn’t even need to be that large.

        The combination tank/radiator you propose would certainly be a lot better than what’s there now…

      2. Did you not read the article? The heat transfer in the lamp is fine, he says it hardly warms up at all when pumping water from a bucket. Clearly the problem is the lack of a radiator in the pack.

    2. Reservoirs aren’t actually necessary in a water-cooling setup. They make it a lot easier to bleed out the bubbles, but if do that manually you can get away without one entirely. I’d strap the pump (a DDC-1T) on the opposite side of the helmet and use a largish radiator that’s optimized for low air flow. It would be thin but have a bigger xy footprint. You might be able to get away without any forced air if you could get it to stay horizontal. I always like to have a lot of radiator so the fins can spin real slow. The fans on my rad generally stay around 700rpm even when the dust builds up. More intensive things spin them up to like 1200rpm.

      I was going to say that it would be cool to make it (the rad/battery system) strap to your back. But that would get really hot. Sweat city.

  2. Awesomesauce! 15 years ago I built a bike headlamp for night riding in the winters of Edmonton; I used a 20 W track mount 12 volt bulb and ran it off a motorcycle battery. It was awesome. My friend built one, and we were wondering why his battery died after 20 minutes and his helmet melted – turns out he’d bought a 50 W bulb by accident. His was so bright I thought my lamp was dying. 130 W of LED power (vs the old incandescents I used) would be eyeball searing!

    1. Hopefully he didn’t go blind. Even if it wasn’t in his direct line of view, one sharp reflection and he’d be blind for a while (not to mention the implications of that happening during nighttime riding).

        1. @John
          Sometimes when the packaging on an LED bulb says “100 watt” there is small print above the number that says “equivalent to an incandescent bulb that draws.”

          Sometimes there isn’t.

          The difference is important and, after the money has already been spent, painfully obvious. :D

      1. The LED has no lens so the beam is very wide: it floods a large area around you but does not go very far or make a certain spot especially bright as lamps with lenses do. Reflections are no issue unless you’re looking directly into a mirror and you can walk past it without problems as long as you don’t look directly at the LED.

  3. I guess that knowing the inductance of the inductor, the voltage drop across the LED and the on and off times of the MOSFET, it would be possible to calculate the current and do away with the current measuring circuitry?

      1. You are right, but I have played around with these LED’s a while back. They are almost self-regulating when you undervolt them. Providing ~90% of it’s rated voltage will give almost no drop in brightness but a significant drop in current draw and heat output and you then do not need fancy current limiting circuitry.

        1. Don’t rely on that though. Temperature changes can cause the things to go apeshit on you. There should always be a current limiter of some kind in the circuit or you’re in for toasted LEDs (ask me how I know)

    1. In a Current Mode Buck Converter, Kirchhoff’s current law: Current going through the inductor is same as sum of current going to the filter cap and LED. Since the cap is only there to store/release energy, if you can control the inductor current, then you can control the LED current. BTW set up the output voltage of the converter to a safe value for the LED.

      Some Current Mode switcher uses a sensing series resistor for the inductor current and some use the voltage drop across the upper MOSFET (RDS-ON). Some even use the internal resistance of the inductor. http://www.microsemi.com/document-portal/doc_view/14646-an-7-a-simple-current-sense-technique-eliminating-a-sense-resistor
      “A Simple Current-Sense Technique Eliminating a Sense Resistor”

    2. In discontinuous mode (like I use here) you can do this by knowing input voltage, on time and inductance. However the sense resistor costs just a few cents and prevents you from botching things up if your converter accidentially reaches continuous mode where the non linear characteristics of the LED play a major role and current starts to run away.

Leave a Reply to Christian KnüllCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.