Hello, I am working on an air-cooled quad XHP50.2 flashlight, will be very exciting when finished. I am stuck at the driver however as I am not so good with electronics.
I am aiming for 6amps per emitter, total of 24. Batteries will be in 2s configuration.
Ideas until now: 1. 2x 12amp LD-3 drivers to power 2 Leds each but it doesnt seem possible because of heat dissipation according to the manufacturer. I could heatsink the Mosfets but I dont know if that would be enough… I asked if I could use one LD-3 driver with which I control 4 external mosfets, one for each led, but it seems also immpossible…
2. Use a Skyray driver with upgraded mosfets and control them with a 17mm nanjg driver, I am using this setup in my 7x Skyray but in 1s configuration. Will I need to make the zener mod in this case? Low voltage protection will still work?
3. Use four 8x7135 drivers zener moded with 8 additional chips each to reach 6amps and control all 4 with the same switch.
Problem: Difficult to solder 32 additional chips….
4. Use two moded Convoy L6 drivers to reach 12amps, each will power up 2 of the XHP50s.
Problem: I will need 2 switches in this case, something I want to avoid.
Please give me your ideas/advices, I will be very thankful.
My input is 8.4v, 2s configuration. I didnt think about this as I thought it wouldnt make a difference because the wattage remains the same… Current decreases but volts increase…
I think a simple FET/DD setup could work for you. I think that is your option 2. I assume you are going to use a SRK host, so 2s2p 18650s? The XHP50.2 has a lower Vf so you have to worry about too much current. With 2s2p high drain 18650s you might get 35A so you might want to actually add some resistance somewhere in the circuit to get it down to around 24A. You would need approximately 25 mOhms extra.
If you used the first generation XHP50 you might get right around 26A without having to add any additional resistance.
To make these estimations I’m using a simple graphical tool that I describe some here.
You would need a way to lower the voltage to the MCU, either a Zener mod or a voltage regulator. LVP could work fine but the firmware would have to be properly configured for it.
Ollie, I believe there is something wrong with your amp numbers. Would you be able to measure again? It will take very short and hefty leads(like: 12awg&12inches) to take the measurement. Maybe someone can correct me if these numbers do sound right…
Thank you very much for the answers!! I will be using s 2s4p setup, 8 high drain 18650s (10A). I think for DD that is a little too much, and if one led goes bad the remaining three will be 100% dead too because I am wiring them in parallel. By limiting the current somehow to 24amps even in an event of led failure each led will take 8amps which is not a problem for a small period of time.
In that case I agree DD is probably too much. The LD3 might be your best choice. You would need to heat sink the FET (s), and this might add quite a bit of challenge to the build. To really heat sink them well you would need to remove them from the driver PCB and make a good heat path to the flashlight body. I have done a couple builds where I’ve used the FET in its linear mode (as the LD3 does) which makes the FET dissipate a lot of heat. See these for some possible ideas:
Buck driver(s) could be an option, though I don’t know of one that would work off hand. But they would probably take a lot of space and might not end up being any more efficient than the linear FET route.
Edit: DD/FET with added resistance is still an option, and would be much simpler than the above options.
In a DD situation, if one of the LEDs died (became a short), the others would less than 8A each.
Ollie,
The heat produced by the emitters will most likely be too much to be on continuously. This is the case with most high power lights, and it’s not an issue of metal thermal conductivity, it’s an issue of dissipating the heat to the air quickly enough.
Yes ideally we would have a 100% efficient driver, but that is usually not possible. You suggested using cells with more internal resistance. This is essentially the same as adding resistance to reduce the current. Ideally a buck driver could work well, but it’s most likely not going to have an efficiency above 90%, in which case it won’t really be much more efficient than simply burning off the extra voltage by using a linear driver or DD with extra resistance.
@ Ollie: What do you mean remote battery pack? I will be using simply 2 battery tubes. I know my project is no pocket flashlight or EDC and was never meant to be, and noone in a right mind would use 14.000lm continiously no matter how big the host is. Even the most expensive/biggest flashlights have turbo timer. Thats why I will go with active cooling and a very nice heavy copper heatsink (no idea how much heat it will be able to dissipate, but better than 5-6 fins and passive cooling). My goal is to be able to keep a stable temperature of lets say max. 45-47C while I am in mid mode around 5000-7000lm. 100% will be used only for quick bursts.
@EasyB: Thank you for your links, they seems very interesting, I will look them now carefully. I spoke with the manufacturer of the LD-3 and he was really pessimistic about heatsinking the mosfets, although I believe its very possible it would work. I could try it but I dont want to waste 25euros only to possibly succeed. So I found another solution.
Final Solution: TA driver 46mm with Narsil firmware. Max possible current at DD according to the manufacturer is 26.3amps/4=6.5amps per emitter. I asked him to set the maximum current from the FET to 90% so I believe it will be ok. The FET will not be working at its maximum capacity and its heat will be hopefully less. I will be heatsinking the mosfet with a simple chip heatsink, probably I will build my own from 0.2mm copper sheet. In case I want more or less current I will change the parameters of the firmware ( never tried it though :+1: ). The Narsil firmware allows to set the brightness to any value using the ramping mode, so one needs to find only the sweet spot between heat and output.
Thanks to those who contributed and spent their time to help. Any additional ideas are of course always welcome!
I am in the process of collecting the parts now, so it will be a while before I have something to show. But I will upload the building process and finished flashlight when its ready!
Nice!!!
Can’t wait to see it!
In Narsil you can set the thermal stepdown yourself and doing so you calibrate the MCU, pretty sweet in a light like this. (And easy, even I can do it with the side switch )
That FET setup should work fine. It is PWMing the FET output, so in this case the FET doesn’t produce significant heat; you don’t have to worry about heatsinking the FET when it’s being PWMed.
My quick estimate of the current, with 2s4p Sanyo GA-like cells, which have ~50mOhms resistance, and 4 XHP50.2s in parallel is around 35A at full charge, so you might want to reduce the max duty cycle to something more like 75 or 80%.
Regarding heatsinking the FET when used in linear mode, like on the LD3: it is entirely possible to heatsink it. It just has to be done correctly, by soldering the drain of the FET to something with a good thermal path to the flashlight body. It would have to come off the driver PCB; there is just not a good enough thermal path when the FET is on the driver PCB. These FETs that we use on our FET drivers are rated to dissipate more than 100W, as long as the drain is properly heat sinked.
It seems you’re wanting to do something similar to a light I want to finish - when I can afford the emitters. It will use a 32x 7135 driver and 20x 7135 daughter board from RMM to power 4 xhp70.2s or mt-g2. 52x 380mA 7135s = 19.76A. 2s4p 18650 but it will be able to use 21-70s.
I am not going to use a Maglite host, I dont think I reffered to it but I am using a soda can type of host. Quad XHP50s in a Maglite would be I guess only for demostration purposes of 10sec bursts…
It would be very interesting to see it, please upload some pics when you have it. Consider also using XHP50.2s, they are cheaper and can handle if I remeber correctly up to 9amps, so your driver is compatible. Apart from that you can use XML-sized reflectors/centering rings which are much more common than XHP70 ones, plus you have more throw. Think about it.
)
