Flomotion had a very interesting idea of running an LED underwater, so on a quiet Sunday afternoon, I had a go at it.
I’m using a dedomed XM-L (not XM-L2) on a 20mm Noctigon copper PCB with a 15V 34A voltage controlled PSU and a Fluke 117 DMM on current measurement mode.
I was impressed by how well the LED handled the current, thanks to the large specific heat of water (4.186 joule/gram °C) vs copper (0.390 joule/gram °C). This is per mass, not per volume.
Another thing to note is how the lowering of the Vf as temperature increases accounts for the increasing current at about 6.60A. At other intervals I just increases the voltage of the PSU.
Pure water doesn’t conduct electricity, the impurities do the conducting.
Many years ago I had a job in a power station running steam turbines. After the turbines had been shut down for planned maintenance they would be filled with condensate to flush it out and turned by a small electric motor, the water was as near pure as it got going into the turbine. Some of the water was drained off every hour and the labs would test the conductivity of the water. As the conductivity approached 0 the turbine would be declared clean, drained off and gently brought up to operating temperature and pressure.
Interesting. I wonder how far you can push it if you keep the temps down. In CPU (computer) overclocking they push extreme numbers by smothering the CPU in liquid nitrogen
Maybe we can try it using a peltier http://goo.gl/IJ2zLJ
When a emitter+PCB is inside a light, I keep hearing that if the body of the light gets warm/hot, that’s a good sign that the heat is transferring well. How would the 50-55C that you measured compare to the temperature in a well heatsinked flashlight?
The specific heat of water is 4.18 units/volume and copper is 3.45 units/volume, so water does have a slight edge, but are very similar.
The top half of the water is about 80-100mL, so if there was a copper heatsink of 80-100 cubic cm, it would have a very similar temperature of 50-55C, or a little higher.
The copper however, should have fins and would be able to dissipate that heat into the air.
Since the FET drivers came into being, most of my lights use an XM-L2 at just over 6A. Small lights to large lights the emitter seems to do well, always on copper.
I tried a build where the existing emitter was pulling 8.47A (SBT-70) but an XM-L2 de-domed couldn’t handle it, so I had to reduce power. Ended up using 6.62A to the emitter in that light. (emitter amps, measured from a loop soldered into the negative lead of the emitter with a clamp meter)
The top 18650 cells will drive an XM-L2 to around 7.12A in direct drive, tested on a Noctigon which itself was thermally attached to a 3x3x3 cube of 6061. So these FET drivers are getting close to direct drive with several of mine doing 6.2A or so.
I tried isopropyl alcohol. I got it to boil with four new alkaline D cells. I could see the shadows of the bubbles. The alcohol was not conductive at 3 volts, even if I added salt to it and even though it has some water in it. What ended the test was when the head came apart due to the alcohol softening the epoxy. Water starts to separate into hydrogen and oxygen at around the voltage of an led. The oxygen combines with metals if present.
Someone else found an add. for a silicone oil cooled led light bulb and liked to it in the same thread.