A quick update. My attempts to make the second unit were futile because of poor led chip quality. I need LEDs from the same batch, because apparently what I got in my second and third shipments from fasttech are of poorer quality than my first ones. At the same voltage, the array draws a lot more amps and one-two leds die(breaking the circuit, and protecting the others). I’m using helluva thick wires and good soldering between them so the problem must be in the chips themselves. This is getting expensive. A new shipment would be $38 for 10 chips.
I added booster modules to keep the voltage and current steady at about 14.3v & 9A. You could theoretically power this from a 18650 battery pack. Yeah, it should be able to give 150w of power, but you could in theory. :bigsmile:
I padded the back of the TIR lenses with aluminum foil. From my light meter tests this increased the light output by about 15-20%. I need to redo them, because this seems too great and I don’t like saying bullshit. :~
One more thing I added(for testing mostly, it won’t have much - if any - use after I get them inside the headlight unit) is a 40A PWM dimmer. Two more tests with the working array here:
And a more conclusive burn test. You can feel the heat on your jeans from about 70-100cm way. :bigsmile:
Hey Boggy. Sorry to hear about the emitter failures. Were they mounted on copper? I quickly looked above and didn't see any details on the thermal path to the heat sink.
I know that after +5A things are getting really nasty on AL, and copper can let them run at higher current before going nighty-night. I am however baffled by the difference in voltage/current between my batches. The two modules are identical, same wiring, same boost/stabilizing module, same cooling/thermal path(I’m using Ceramique 2 as paste). So why does one work at +10A for a specific voltage, but at the same voltage the other goes way higher(actually, if I remember correctly I saw a max of 10.92A @ about 14.6V and nothing bad happened to the main unit).
I did manage to get the second module to work at about 14.1V and 8.5A, but as soon as I eliminated the multimeter wiring for the current, and wired them normal, they “popped”. And then I had no more LEDs for replacement… :_( That 0.1 ohm resistance in the multimeter and probe wires was all that was keeping them from becoming toast(custom 2mm/12AWG wires for the probes). I don’t know if they soldered them badly on the aluminum stars, or the chips themselves are of lower quality, but this is what’s happening.
Regarding cooling, that AMD cooler does a terrific job. It’s taking 130-150W of power and the heat never rises above 55 degrees centigrade (131 F). I can keep my finger on it and the aluminum pills after +20m of run time. It is however moving a LOT of air as you could probably notice from the burn test video above. The fan scales nicely to 14.5V regarding airflow.
I am pleased however of the small form factor, light output and focusing (despite the problems I mentioned). No 100w LED can match this.
Sorry to hear about the set back. What current do you believe that the leds were being killed, 5 amps? I'd be interested to see what you could get with them mounted on copper stars.
Just above 5A. 5.5A for the ones in good batches. Not bad actually for $3.72 AL star leds(I still can’t find them cheaper anywhere else). I would love to try copper mounted leds as well, but for the quantity I need this would be so expensive.
In theory, the voltage limit for the array would be 14.6v (3.65v * 4leds in series). That’s just perfect for a car’s electrical system. I just added the boost module because it cancels any fluctuations in the input (like when the engine is stopped and the voltage drops to ~12.6v), and it also acts like a small step-down module if the voltage would rise to about 14.8-15v.
I added two thermostats inside the radiator(on the second, faulty module. before I burnt it, of course :bigsmile: ). One is set at 75 degrees C and it will shut down one strip to give the fan the ability to cool the heatsink down more easily without loosing all the light. The second thermostat is at 100 degrees and will cut down the second array as well, leaving the fan running. This way, if mounted in a headlight and there isn’t a lot of cool air coming in(and maybe the heat of the engine interfering too), I won’t burn any of the leds. Also, if there is a fan failure, it will shut down without anything bad happening.
I have to be careful though, the thermostats are rated at 5A. :bigsmile: It was a snug fit in the radiator, but my fear is that they’re a bit far from the heat source, and the actual heat on the “led layer” would be quite higher. So I’m thinking of chaging them to 55/75 degrees thermostats. Or using screw on, KSD301, 10A thermos higher on the rad.
To eliminate any unnecessary stress on the fan if I will need to use it for quick flashes, I got delay modules that will kick the fan in only if the power is on for more than one straight second(adjustable, 0-10s). Gotta love what ebay offers for $1-2
Is it possible that the voltage mismatch on the new batch of emitters is because you mixed them with some older surviving emitters that were already burned in?
I understand that once an XM-L sees a very high current its’ Vf is lowered.
Never thought of this, you might be right! I will test one emitter that survived the ordeal and one from the new batch that I will order. If they differ more than 5-10% that must be it!. Sadly, I have no “virgin” XM-Ls around to test it right now.
One counter-argument to this theory is that the first module has seen quite high currents, and still performs better at high voltages than the other one.
A new shipment would be $38 for 10 chips.