Emisar D1 with auxiliary flood beam

My dual beam trail running light was a good success for me. Having two beams really increases the utility of the light; you don’t have to carry two lights or compromise with only one beam shape. So it got me thinking of similar mods I could do. I came up with the idea of putting cree XD16 LEDs between the heat fins of an emisar light.
This mod did not have an immediate use like my trail running light did, but I think this light could be good as a leisurely walking light. The flood beam naturally points down near your feet, then you can use the relatively throwy and powerful sliced SST40 main reflector beam to check out far away things. The side flood beam can also naturally illuminate a scene when the light is tail or head standing.
Similar to my trail running light, I adapted a stock dual channel driver to control the two separate beams. I wanted the flood beam to be bright enough to very well light the ground while walking, so I wanted more than just the single 7135 present on the driver. Originally I tried to mount 4 7135 chips near the perimeter of the 20mm MCPCB, but the space was very tight and wiring all the chips without shorting something on the reflector was too difficult.
I then remembered I had bought some CN5710 1A regulator chips to experiment with.
A few people on the forum have experience with them.

They are similar size to 7135, but as Mike C explains, they regulate the high side instead of the low side like 7135s and our FET drivers. The constant current is selected by a resistance between one of the pins and ground. Similar to the 7135, the chip has a signal pin which can take a PWM signal to dim the output. However, the datasheet says the PWM frequency should be 2KHz max. How this chip would work with the high PWM frequency of our drivers was the main unknown for me.

With the CN5710 I could just use a single chip to drive 1A to the flood LEDs. The chip can actually be configured to supply more than 1A current just by lowering the Iset resistance, but 1A is the specified maximum, so I stuck with that. I soldered the large ground pad to the driver to dissipate heat, but I had to alter/mask the other pads on the driver since the connections on the two chips are different. I ran the driver PWM signal to the other side of the chip using a small piece of copper tape to line up with the CN5710 pin layout. I masked the other pad with kapton tape since I wouldn’t be using it. The snake of small resistors are the current setting resistors and add up to 1.72 kOhms. I used these just because I didn’t have a single resistor of that value.

I cut a small piece of silicone heat dissipation cube to put on the regulator chip and presses up against the shelf.



The XD16 LEDs are so small they fit between the heat fins perfectly. I stripped the insulation off some magnet wire I had and used that to wire 3 XD16s in parallel. I used arctic alumina to adhere the LEDs to the outside of the light. I drilled a small hole in the head to feed the wires through, then I plugged it with silicone. To cover the LEDs, I used a small piece of plastic (probably PET) from a “clamshell” package. I was concerned about the plastic melting or burning, but I tested it for 10min at 1A and it appeared fine.

I used 3 LEDs to split the current, but I was still concerned that the LEDs would overheat because of the thermal adhesive connection. A back-of-the-envelope calculation shows that the thermal resistance of the arctic alumina connection is about 60 K/W. At 0.33A the LED produces about 0.5W of heat, so it appears I should be OK. The lumen output at 1A is rock solid which supports my estimation.

The dimming and UI is not perfect since I did not change the FW at all. This FET+1 driver has smooth ramping which first ramps the 7135 channel to 100% then, while keeping the 7135 channel at 100%, starts the FET ramp. Only at the top of the ramp does the 7135 channel turn off. So the first part of the ramp uses the flood LEDs while keeping the SST40 off, which works nicely. But the high PWM frequency does not work perfectly with the CN5710. There is a smooth ramp up to about 0.35A, then it jumps to 1A. I guess this is because at higher duty cycles there is not enough time for the CN5710 to turn off.

Really cool mod EasyB! Some delicate work involved there. Great photos too.