2,000 Lumen Bike Light Build

[zinetti]

I ride my mountain bike a lot to stay sane - work and no play gets old but during the winter and daylight savings time, obviously any after work riding is less than sunny. I vowed not to spend $500 on a bike headlamp so I set out to build my own out of billet aluminum, 4 x 180lm K2s, Luxdrive Buck Puck @ 1000ma and 2 x 9.6v 2200ma batteries. This setup has worked great for over a year. One of the LED leads broke so i started researching replacements, who wouldn't want to upgrade at this point right !? I found the Cree MC-Es and want to replace the K2s with these. After debating series vs parallel and the drivers and batteries I would need to accomplish the large wattage I believe I have come up with a package that will work but would like the experts here to verify what I have come up with. All feedback is appreciated.

4 x MC-E - Running each element at 500ma ~500 lumen total per each LED
Arctic Alumina epoxied to parallel wired aluminum star boards
Have not decided on 4000k or 10000k, ordered one of each to test
Seems like the bluer 10000 K2s didn't light the trail as much as the ~5000k did, maybe just me

4 x Parallel star boards to mount Crees on
Run each LED module in parallel internally then wire the 4 stars in series
**Found the parallel star boards at Deal Extreme but they are backordered – anyone know where I can get them now?
Read that diodes in parallel can cause unbalanced current – also read that if 3 or more in parallel then balance remains pretty consistent
The other issue is with a single driver driving all 4 x 4 = 16 in series means I would need ~55v forward – seemingly unmanageable

CCHIPO driver - Outputting 2A max with remote dimmer of 1A or 2A
Awesome Driver description page TorchBoy - thank you!
Want to drive each LED to 500ma max – more seems like waste of power and heat
**What resistor values will I need to get 1A or 2A from CCHIPO?
**Is this a best driver option?

Battery - 25.9v 2600mah Li-Ion or my existing 2 x 9.6 2200 NiMh in series
Looking for about 2 hr run time at full power
Can CCHIPO receive higher voltage than it outputs – I estimate 13.6 v forward needed for the 4 LEDs
**Better to go with 2 x 14.4 v Li-Ion batteries in parallel?

Optics – I believe the same K2 lens holders and polycarbonate lenses will work, they seem to fit
Please correct me if I am incorrect

Heat - I have an all aluminum body 2.1" x 2.1" x 1" that these mount in
The K2s made this unit warm to the touch but never HOT
The much higher power of the Crees will no doubt be an issue
I now have a CPU heat sink with 25 sq in total surface area that will be thermal pasted & screwed to the unit
**Any way to determine if this will be enough to dissipate the heat created at full power with ambient temp of 50 degrees F with some air movement?

Sincerest thanks for any guidance provided,
Jason

 

[DM51]

Welcome to CPF, zinetti. I'm moving your thread to the Bicycle forum for you, where I hope you will receive some good replies.

 

[MtLuke]

4 x Parallel star boards to mount Crees on
Run each LED module in parallel internally then wire the 4 stars in series
**Found the parallel star boards at Deal Extreme but they are backordered – anyone know where I can get them now?

Try using LED supply, I know they have MCE's in stock, I think they're parallel.

Eric S

 

[trout]

Might be worth waiting for the HipFlex driver from Taskled. to hit the streets

https://www.candlepowerforums.com/threads/212737

 

[zinetti]

Thank you Trout, that will make the perfect driver! George seems like a great guy too.

Much appreciated!

 

[zinetti]

Thanks MtLuke but LED Supply said they did not have just the bare boards, could order but would need to order more than I would need.

Jason

 

[znomit]

If you want a parallel board its easy to jumper an individually addressed board.

 

[FrontRanger]

Run each LED module in parallel internally then wire the 4 stars in series

I would not do this...

Read that diodes in parallel can cause unbalanced current – also read that if 3 or more in parallel then balance remains pretty consistent

... for the very reason of current imbalance. Here's the problem: in forward bias, the current is approximately proportional to e^(Vd/Vt). At room temp, Vt is about 26 mV, so even a 5 mV mismatch on two diodes is a large difference in current. Putting more than two in parallel will help to bring the diode voltage closer to the statistical average for that type of diode at a specific current. It will help somewhat but some of the diodes will still get significantly less current than others, making it wasteful. A better configuration for the same nominal voltage is to make four series strings of four LEDs each, then connect the fours strings in parallel. This averaging of the sum of four LED voltages per string gives you an improvement by sqrt(4) = 2.

If you really want to equalize the currents, you can drive your intended current through each of the 16 diodes and measure the voltage. Divide them up into four groups such that the sum of each of the four is as equal as possible (use a spreadsheet). The main drawback with this is that the wiring can be messier, and may make it more trouble than it's worth.

Good luck with the build.