Need help powering my cree bikelight with li-ion

[moon lander]

Hello, im in need of a little advice on how to power 4 cree leds with batteries. My current setup has 2 cree p4s in parallel powered by 1 18650 li-ion in direct drive. I know this is not ideal because if the vf of one of the leds is lower than the other, they will draw different currents, resulting in one brighter/hotter than the other. The setup works well however, 1.85amp draw from battery. Good heatsink on this light keeps them cool (the bike itself connects to the heatsink and draws alot of heat away.

Id like to add 2 more leds for a total of 4 Cree p4s. Id also like to drive them harder (at least 1 amp each). ideally id like it if they each drew the same current as well. runtime isnt terribly important at the moment but it cant hurt. here are some possibilities ive been considering.

#1 create a 2s pack- charge the batteries seperately to the same voltage, and then wire them is series, and wire the leds as 2 strings of 2 (2s2p). i realize i will have to make sure the voltage of the 2 cells stays the same, and charge them separately. i dont have the dough to get a balancing charger.

#2 create a 2p pack- charge the batteries separately to the same voltage, and then use them in parallel with 4 parallel leds. problem here is the led vf right? they may all run at a different current.

#3 run all 4 leds in parallel from 1 18650. with a short runtime, this is possibe right? does anyone know the max discharge rate of these unprotected 18650s from DX?

another option is to get a driver, but do they make a driver that will power 4 leds? or could i use 4 of these 750ma drivers from Kaidomain all on the same battery? or maybe 2 to a battery?

if anyone has any answers to these questions, i will be eternally grateful. im a bit of a noob when it comes to batteries (i do understand the dangers of li-ion) im happy to do lots of research, just need a starting point. thanks so much for your help.

 

[Timson]

Moonlander,

I encountered much the same problem when I built my Quad Cree mag....Really wanted to run regulated, but found my options limited with my cell choice.

I ended up running 2 x 18650's in series, direct drive to 2 series/2 parallel emitters - current draw well within the 18650's capabilities.

My emitters are all from the same reel / bin....... I can't perceive a difference in output between each parallel pair.


Tim.

 

[VidPro]

#3 +
run 4 leds all parelleled , in direct drive from Parelell set of 6 X 18650s , charge the parellel set as a set. then when that is just a bit to much power, put slight resistance on each parelell led unit, like .4ohms.
then you got very minimal losses, 3-4 hour runtime, to 50% and diminishing runtime for HOURS longer, if your out for longer than you suspected.
put in a 2 way switch, with a 4ohm fat resister to do a low mode that will last much longer.

WHY?:
wires alone could provide enough resistance with such fat amps, assuming you put the battery back on the bar or in water bottle type holder.

there is nothing to break or fail, it will be at least as efficient as adding something in.

you wont be running li-ion in SERIES, so you wont have much issues with that (balance and duel chargers or seprating the batts).

the runtime will be extended by not loosing 30% in some curcuit, and by running lower when the batteries are weakened.

you have enough total capacity that the batteries are running 1/2C, which is a happy rate for them.

the batteries will not be out of balance because they are parelleled

the batteries will be hard to "deep discharge" because the white leds will slow down a lot before the battery capacity reaches critical states.

ohh and SAME batch, same BIN is important, specially if your not running the added small resistance to each led. Same temperature will be important too. so you wouldnt want to do that, where temps on some will be much higher than others.
use a fuse or breaker near the battery pack

 

[AndyTiedye]

I am considering running 3 LEDs in parallel with a FluPIC on each one.
That should avoid any problems with mismatched Vf and one LED hogging too much of the current.
It also won't leave me in the dark if one LEd or regulator fails.

 

[moon lander]

thanks tim and vidpro, thats great advice. ill try parallel emitters and 2x18650 parallel at first because i only have 2x18650. if i notice any differences between emitters, ill parallel the bright ones to the dim ones and switch it to parallel/series with 2x18650 series. love the idea of a high/low switch for extra runtime. andy id love to go with what your doing eventually when i get into using drivers. thanks guys.

 

[VidPro]

you only got 2 battereis and 2 emitters that are the same.
i reiterate the Same batch, same age thing for the batteries and the emitters.

if you can assemble the whole thing with new stuff 4 well matched cells, and 4 well matched emitters. you would lose say 30$ that you could apply somewhere else.
any good place you buy the parts from, even un-binned, will be very close matching with little effort or special request.

that is kinda a disclaimer, you want everything matched, including the cells.

 

[chris_m]

A Fatman boost regulator will run 4 Crees in series from either 2s or 2p LiIon batteries (or even a single cell if you want) - 2s is more efficient since there is less difference between Vin and Vout. Any issues with cell balancing with 2s can be quite easily solved by using an external plug in balancing circuit - I use one based on the design at http://www.flyelectric.ukgateway.net/lithium-balancer.htm. I currently have a 3 Cree light I run like this, but will shortly be building up a very similar quad.

 

[VidPro]

also a N-flex, curcuit, is a Buck type, and will fall out of regulation and slow down, when used in the series type of configurations. the n-flex will take up to 16V input, and run up to 4 in series.
Boosting curcuits should be used with protected cells (like some cool new C ones) cause they can deep discharge series cells.

we have seen boosting curcuits and the Increased amperage they try to draw out of any series sets of cells when the capacity of one or more of the cells start to drop, its not nice to the cells.

 

[chris_m]

Yes, but he was looking to run off 2 LiIon cells, and with the nFlex you need at least 4 cells.

Personally I'm not convinced with the idea of using LiIon cells in any way without some sort of protection - I effectively have external protection since I use a uC circuit with my Fatman with low battery cut-off. If using something like this there is effectively no difference between a boost or a buck in the loading which is put on the cells.

 

[VidPro]

right and the more you stack up in series, the more out of balance it can be, without noticing that one of them is low.

i would still bet that more batteries (like ni-mhy) have been ruined by boost type curcuits which draw MORE amps as the battery becomes depleated, than by buck curcuits that fall out of regulation and slow down.

one slows Down when the battery is depleated, one speeds up at the worst times.
but as long as any cutoff still falls within the specs for ALL of the cells in the pack reguardless of the lack of balance at the time, then it shouldnt be a problem.

and to get a buck type curcuit to do the trick it still has to be very Close to the led voltage, so it WILL slow down.

i think that is the total simplicity of the 1 led 1 li-ion cell, they are already closely matched, its already a done deal. why waste energy doing something unnessisary.
if every component of the curcuit puts out light then you have wasted less . unless you need a pocket warmer.

 

[chris_m]

I agree if you leave the regulator to go out of regulation then the buck is better, but whilst you are still in regulation the effect on the battery is the same, and if you have a low battery cutoff then you never go out of regulation.

Yes a 1 LED, 1 cell solution is simple, but it's inherently unregulated which I still think is a pretty poor way to do things given the non-linear voltage/current curve of LEDs which results in a huge decrease in light as the battery voltage drops. Yes a regulator wastes energy, but with a well designed one this can be as little as 5%, which is pretty much lost in the noise given the advantage of a continuous steady bright output.

 

[VidPro]

i dont think resistered drive is a poor way to do anything.
been tossing curcuits on lights left and right, and replacing it with li-ion, and everytime, i get massive runtime over the "regulated then dead".
i would far rather have a light continue to go down, than be stuck on the street with a light that doesnt output any light.

it does go down, but most of my lights go to 50% SLOWER than the charts show for regulated, and run for HOURS longer after that. lots of these curcuits are highly inefficient , varying by a lot depending on the voltages of both items.
Do the math on the power supply, or with your meters, if you can find a curcuit that does anywhere near 5% loss , tell me which one it is.

every light i have tossed the curcuit completly out on, has never gone out on me <--- never, it always puts out some light , it needs a charge but it always works.

the fatman, and stuff like that are GREAT curcuits comparativly, the ones in the flashlights are very low regulated besides, they float up and down with the voltage like the ocean does the tides. Almost (but not quite) as much as rechargable batteries without em. and for good reason, they probably learned that torturing battiers on high amps when the capacity is gone, is not a good thing.

(oops, voltage not current, corrected)

 

[chris_m]

if you can find a curcuit that does anywhere near 5% loss , tell me which one it is.

Already mentioned it - look at the figures at http://taskled.com/techfatman.html. I've also put a meter on my system, and running 3 Crees off 7.4V LiIon the efficiency is up around the 95% mark running at 700mA output (<7.5W input for 7W output, hence still >2 hour runtime from well used 2200mAh cells).

It is always possible with a uC controlled regulator to get it to auto decrease the output near the end of runtime, so get the advantage of good continuous beam for most of the run and then keep the light shining as long as possible - my firmware has this as an option. However there's a point when "some" light isn't enough (remember we're talking bike lights here - you need lumens to keep riding at a decent speed), hence I'd much rather just carry enough batteries and have a warning flash when the batteries are running down allowing me to switch to a get-you-home mode if necessary. Oh, and my system also has a battery power level meter - much more useful than judging how much you've got left from how dim your light is!

 

[VidPro]

and if you start with enough batteries, (unlike trying to drive 3-4 off of 1 cell), you get your runtime and your lower battery discaharge rate, even when the boost start pulling the amps.

george always makes great curcuits, and current control is much better than that PWM stuff especially when the pulse rates they are choosing are so low.

but you also mention, some sort of needed cutoff, you need some battery level thing, you need some curcuit , increase costs, increased parts, increased labor, increased thought, and increased knowlege to apply.
and the resistered drive is SELF regulated, you dont have to monitor nothing, and you dont have to run batteries IN series at all.

and the average person can slap a resister setup pretty easily, sure we could stuff a computer in the thing, but it will only crash

 

[Calina]

Go with a generator and stand light, you will never be without light. ;-)

 

[moon lander]

vidpro, chris, you guys are awesome, thanks for all that info. so ive been looking at the fatman and nflex, they both seem pretty great. the nflex has a cool UI but looks like it could be a little complicated trying to program it with just 1 button. i love that the fatman can use an external taper pot, and its simple. also lets me use a simple battery setup (less series batteries), and more options for other types of batteries (like powering 3 crees at 750ma from 4 AA nimh cells).

there may be a problem with the fatman. i dont think i can power 4 crees without using series cells. id like to avoid using series cells so i dont have to worry about balancing them. i got this from taskleds tech pages about the fatman:

Examples on how to calculate battery/LED combinations:

Step 1: Determine output power

Power_output = Number_of_LEDS * Vf * Output_current

Step 2: Determine input power

Power_input = Power_output/efficiency

For calculation purposes we can assume efficiency will be around 90%

Step 3: Determine input current

Input_current = Power_input / Battery_voltage

Do this calculation for the lowest Battery_voltage you plan to run

Now, Input_current should be around 2A or less for optimal performance of Fatman. Fatman can run up to around 2.2A max. When running at high input power it is recommended to thermally epoxy the back of the Fatman PCB to a heatsink or the body of the flashlight.

heres my calculations on that: am i using the right value for vf of the leds?
power output = 4 leds * 3.7v * 1amp
power output = 14.8 watts

power input = 14.8w/90%
power input = 16.4w

input current = 16.4w/3.7
input current = 4.4 amps

input current is about 4.4amps, which is twice as much as it can handle. if i used 2 cells in series, its close to 2 amps with a freshly charged battery, but as the cells dip in voltage the amperage will exceed the fatmans max input current rating.

so it looks like i may be forced to use series cells afterall. is it hard to manually keep them balanced? if i charge them seperately and make sure the voltage is the same when resting, and make sure the cells are the same age, i should be ok right?
thanks again!

 

[VidPro]

that is about right, with either of them you would need to do series or series parellel on the LEDs themselves to get to a full drive current.
the n-flex maxes out at 1000ma, so you can get 2x500, or you run all the leds in series.
THEN you going to need Much and much voltage, many batts in series.

the fatman will "boost" so you wouldnt Require as many cells in series to run the leds in series, but it is still better if the battery voltage comes close to the leds .

with the n-flex your battery minimum voltage wants to be about 1v OVER your Leds max voltage.
ex: 14.4+v battery set, 4 leds

with the fatman your battery Maximum wants to be sligtly UNDER your leds maximum.
ex: 10.8V battery set 4leds

or therabouts, dont make me do math

and like i mentioned before, when the voltage on the nflexes battery goes below it will fall out of regulation and start dropping.
when the voltage on the fat man goes down, the current draw on the cells will increase.

 

[moon lander]

haha i would never make anyone to do math. thats just not cool. are you sure the fatman can drive leds in parallel? even so, wont the total power be the same and still too much? also, where is the best place to buy these drivers in the northeastern US?

 

[chris_m]

If you're using a Fatman, then there's no point in running the LEDs in parallel - won't help at all with your input power issues. From the ones I've had I'd suggest the Vf of the Crees is nearer 3.5V (mine are actually only 3.3V at 700mA), hence only 14V total, so you need a little less input power, though otherwise your calcs seem pretty good. Should just about be able to run at 1A with 2 series cells, though it will be a little out of spec when the cells are nearly discharged - I'd recommend dropping to 900mA, as you won't notice the difference in output and it will stay in spec. Alternatively run 3 cells in series (my plan with the quad I'm building is to use a 2 cell battery for 700mA or less in normal use, but a 3 cell for running at 1A).

You get the drivers direct from George at www.taskled.com - he's in the US so it should be nice and quick (even to the UK orders from him only take a few days). Also a forum there where it's worth repeating questions about current and power limits of the drivers as you'll get a response direct from George.

 

[AndyTiedye]

Added to Bike Light Thread Links