Driver for 3x XP-G in series

I am wanting to build a light with 3 XP-G LEDs wired in series. I am want to run the light off of a single 18650, What would be the best single mode driver to run this so that each LED is powered at 1000mA or close to it? I would also like to have the driver no larger than 17mm but since my light host has not been built I can change it to fit something a little larger if I have to.

Also please help me understand how this works to makes things easier for me in the future:
If I understand this correctly each LED has a forward voltage of 3.3 at 1000mA. When wired in series I would need to have a Vout of 9.9V at 1000mA. So to run a 18650 (4.2V fully charged) I would need to have a boost circuit driver... correct?

Thanks for any help you can give this light building newbie!!!

I'm no expert but a Fatman boost driver would be perfect I reckon.

http://www.taskled.com/fatman.html

That is one of the ones that I looked at but wasn't for sure if it was right, so thanks for pointing it out!

The big problem for me is the .8" (20.32mm) diameter is definitely larger than I wanted. I can and will rework my design of the host if needed but would prefer not to of course. If you or anyone else knows of a smaller diameter board that would make my life easier

sprinkle said:

a single 18650,
each LED is powered at 1000mA or close to it?

Click to expand...

That's gonna be hard. Suprisingly few batteries will put out 3A continuous. You'll have a 5min show-off light, not a stable output long running torch.
Just FYI, despite enticing advertising many 18650's won't come close to your design limit.

there is a 'tri flu pic' (not sure spelling) that runs 3 emitters individually, as in not series or parallel but 3 separate connections. Looks really good, I'm eager to try one out, sold on CPF, ~$38 each

Linger said:

That's gonna be hard. Suprisingly few batteries will put out 3A continuous. You'll have a 5min show-off light, not a stable output long running torch.
Just FYI, despite enticing advertising many 18650's won't come close to your design limit.

Click to expand...

Is the 5 min. a limit of battery capacity? or is it because it will get too hot? or will driver overheat?

Do you have a list of batteries that could handle 3A continuous? I was planning on using an AW 2600mAh 18650.

How is the light listed in your signature (Moddoo__ triple__ xpe) able to handle being ran with an 18650? Is it that the LEDs are ran in parallel instead of series? If so I thought it was safer to run LEDs in series to eliminate Vf variances?

Thanks for your reply and for your help!

The fatman is a nice driver, but it is not rated by georges80 (the designer) to run at 3 amps input - more on the order of 2 amps. (see bottom of this technical description page here: http://www.taskled.com/techfatman.html

The intention is that the fatman driver would be used for 2 LEDs in series.

The closest driver with an attempt at that rating is the shark boost driver, but even then, Wayne (the designer) really suggests similar kinds of use for that one as well, although it certainly has been pushed harder in the hobby world. It starts to get into the range of needing serious thermal management of the driver as well.

There really aren't many drivers in that category on the market, at least that I have been able to find. If I were going to do a 3X XP-G project like that, esp. for hobby use, I would wire them as 3P with a small resistor (maybe 0.5 - 1 ohm) in series with each battery to LED connection. There is no driver out there more efficient than this setup, and your runtime will be approx. 15 - 30 min max no matter what you do.

It is a good project for really maxing out the power balance of battery and LED capability. Expect to throw away your 18650s about every 10 - 20 charge / discharge cycles, as this kind of use is a push. It is a small price to pay for Indy car performance from a small engine. This is the price of safety when having a lot of fun.

HarryN said:

The fatman is a nice driver, but it is not rated by georges80 (the designer) to run at 3 amps input - more on the order of 2 amps. (see bottom of this technical description page here: http://www.taskled.com/techfatman.html

The intention is that the fatman driver would be used for 2 LEDs in series.

Click to expand...

Thanks for the info. I guess this pretty much rules out the fatman.

The closest driver with an attempt at that rating is the shark boost driver, but even then, Wayne (the designer) really suggests similar kinds of use for that one as well, although it certainly has been pushed harder in the hobby world. It starts to get into the range of needing serious thermal management of the driver as well.

Click to expand...

The Blue Shark was the one that I had figured on using if I couldn't get a better answer. I see that it is rated at 4A max and I should be below that though only barely. I have a pretty good heatsink designed for my light and was planning on using thermal paste or similar (still researching best method) to bond the driver to the heatsink.

There really aren't many drivers in that category on the market, at least that I have been able to find. If I were going to do a 3X XP-G project like that, esp. for hobby use, I would wire them as 3P with a small resistor (maybe 0.5 - 1 ohm) in series with each battery to LED connection. There is no driver out there more efficient than this setup, and your runtime will be approx. 15 - 30 min max no matter what you do.

Click to expand...

Yeah I am quickly finding out there isn't really a good solution for what I want to do.

Well right now I am kinda stuck on the series for this project at least... I ordered the LEDs from cutter already on a 20mm board wired in series. I really don't want to get involved into changing them out.

I am content on a 15-30 minute runtime as this will be a show off light and not a daily carry.

It is a good project for really maxing out the power balance of battery and LED capability. Expect to throw away your 18650s about every 10 - 20 charge / discharge cycles, as this kind of use is a push. It is a small price to pay for Indy car performance from a small engine. This is the price of safety when having a lot of fun.

Click to expand...

Haha I am known for trying to push the limits on stuff... I guess my start into flashlight building will be no exception.

Wow, there would be that much stress on the batteries that I would want to replace them after only 10-20 charges?


Would I be better off using say either 2x lithium primaries or 2x rcr123 li-ion even if the runtime is greatly diminished?

sprinkle said:

.

Wow, there would be that much stress on the batteries that I would want to replace them after only 10-20 charges?

Would I be better off using say either 2x lithium primaries or 2x rcr123 li-ion even if the runtime is greatly diminished?

Click to expand...

Please keep in mind that I am pretty conservative when it comes to Li based cells. I make the assumption that if my wife uses a light, and something goes wrong, life will be really bad.

I built a light called "breeze" which ran 2 x RCR2s x 1 Lux V at 700ma. It was pretty bright for its time, and it ran those RCR2 cells at a nominal 2C. Over time, it became clear that this was really more closer to 2.5 - 3C, which is on the upper end of safe.

There were some other RCR2 lights that pulled 4C, and would destroy those cells in 2 - 5 runs. The cells were actually excellent, but can only take so much, and since the risk of charging is usually a damaged cell under "charge" conditions, it is just a dice roll.

It didn't bother me to do a "toss them periodically", as they were still much cheaper than quality brand name primary CR2s. These were AW, protected RCR2s, which were quite good, but are no longer available for manufacturing reasons. This is one reason I never marketed the light, even though I have lots of parts.

Your project is similar, but not quite as intense, since you seem to still be pulling less than 2C. This is within the range of safer operation of a fully protected cell, but for instance, 4 XP-Gs would probably not be recommended.

I would not switch to primary cells, even if you can. The challenge with a light like this, is that you really want it to have good run time when you use it / go for a walk. In order to get this, you tend to change cells far too often, and start counting "run minutes". With R cells, you can just top them off anytime. For some lights, this does not matter, but for short runtime lights, it is a big deal. I learned this first hand.

Add up the cost of tossing your cells even once per month (about $ 20 / month) compared to tossing a pair of primary cells once per week. (about $ 5 - 10 / week) I suspect that after 6 months of use, your R 18650 cells will be depleted only about 1 - 2 X per month. The first 6 months don't really count, as we always use new toys a lot - right?

Keep going - it sounds fun.

So if I run 2x RCR123 I should be pulling under 2A correct? That with the Blue Shark boost driver I should be fine?

Since I am doubling my Vin I am effectively halving my Amp draw, so even though I have less capacity in my batteries 2x 750mAh = 1500mAh in the 2x RCR123s vs the 2600 mAh in the 18650s I should have a similar runtime. Is this correct?

Thank you so very much for taking the time to help out! lovecpf

Oh and not sure if this will end up a project that I would like to market rather just a fun little hobby right now. I have good resources to play fairly cheap though. I work with AutoCAD everyday so design no big deal and actually fun. One of my best friends is a machinist so I can work out trades for machining which is the big cost. And my boss is going to help me with the soldering until I feel confident with my skills to work on something this small on my own.

You are trying a very difficult project. You need to walk, before you run. Learn some easy stuff first, before you try this. There are a lot of bike lights that you can buy with P7 leds ready to go. To do what you want you need at least 2 18650 cells.
Don't give up...but start the learning process by reading more in the BIKE section.....

Ok, let me do this one on paper so I keep it straight and everyone can double check my work.

Power needed
- 3 x XP-G = approx 3.5 volts Vf X 1 amp = 10.5 watts to LEDs.

- Assume 80 % efficiency, so you need about 12 watts from the cells.

Power available

- 2S x R123s = 2 x 3.8V (avg) x (x)current = 12 watts.

x current = about 1.6 amps, so right on the edge, but ok. It looks on paper like 2C, but with only a few changes (cold cells, partial charge, etc), it gets close to 3C fast, which is not that pretty. Not exactly a long life option, but viable, and frankly, it is nice to have the option of running on primaries just in case.

Driver - need to look at all of the specs on those, but watch your max and minimum Vin closely.

edit - ok, looked at the fatman driver - seems like it will work electronically, with adequate heat management.

Shark drivers - I have less experience with these, and for some reason I could not get on wayne's web site tonight, but it might be ok as well. Keep in mind that if someone uses primary cells, the battery voltage will drop to about 2 - 2.5 volts / cell during a draw like that. If the micro controller on the driver really needs 5 volts to run, it could shut down in use, or not start at all on a cold day.

IMHO, what you are attempting is ok for personal hobby use, but I am not sure that it is a robust enough concept to tolerate all of the variation seen in say a batch of 1,000 lights. There can be some significant losses in a light body, connectors, etc.

Lighthouse one said:

You are trying a very difficult project. You need to walk, before you run. Learn some easy stuff first, before you try this. There are a lot of bike lights that you can buy with P7 leds ready to go. To do what you want you need at least 2 18650 cells.
Don't give up...but start the learning process by reading more in the BIKE section.....

Click to expand...

Thanks for the advice! Not really sure where the bike reference came from but I will check out that forum to get more info.

HarryN said:

Ok, let me do this one on paper so I keep it straight and everyone can double check my work.

Power needed
- 3 x XP-G = approx 3.5 volts Vf X 1 amp = 10.5 watts to LEDs.

- Assume 80 % efficiency, so you need about 12 watts from the cells.

Power available

- 2S x R123s = 2 x 3.8V (avg) x (x)current = 12 watts.

x current = about 1.6 amps, so right on the edge, but ok. It looks on paper like 2C, but with only a few changes (cold cells, partial charge, etc), it gets close to 3C fast, which is not that pretty. Not exactly a long life option, but viable, and frankly, it is nice to have the option of running on primaries just in case.

Driver - need to look at all of the specs on those, but watch your max and minimum Vin closely.

edit - ok, looked at the fatman driver - seems like it will work electronically, with adequate heat management.

Shark drivers - I have less experience with these, and for some reason I could not get on wayne's web site tonight, but it might be ok as well. Keep in mind that if someone uses primary cells, the battery voltage will drop to about 2 - 2.5 volts / cell during a draw like that. If the micro controller on the driver really needs 5 volts to run, it could shut down in use, or not start at all on a cold day.

Click to expand...

Thanks for taking the time to do these calculations! I really appreciate all of your help! I think I will have to drop the idea of 1A for this... I can deal with less power if that means safer!

IMHO, what you are attempting is ok for personal hobby use, but I am not sure that it is a robust enough concept to tolerate all of the variation seen in say a batch of 1,000 lights. There can be some significant losses in a light body, connectors, etc.

Click to expand...

Yeah this is definitely just a hobby thing and if I make more it will be in small runs and care taken with each and every light.

Ok so after seeing all these comments and doing a little more research it looks like powering these at 1A is not going to be a safe option. So see what you think about the following:

If I use the Blue Shark boost driver and adjust it so that the LEDs are only getting 700mA then i can figure:
7.2V(2x RCR123 @ 3.6V ea. - figuring at the dead end of the batteries) for my Vin
9.6V(Cree lists the XP-G at a typical Vf of 3.2 @ 700mA)
Then
7.2V*X = 9.6V*.7A(figure in 80% driver efficiency)
X = 8.4W/7.2V = 1.17A draw from the batteries. Since AW rates their RCR123 at 1.5A discharge rate max, I should be ok... right?

In fact if using a AW 18650 I should be ok.
3.6V*X = 9.6V*.7A(figure in 80% driver efficiency)
X = 8.4W/3.6V = 2.33A draw from the batteries. Since AW rates their 2600mAh 18650 at 5.2A discharge rate max, I should be ok... right?

I don't really understand why you want to use a boost driver and run the leds in series.
Personally i would go for a P7/MC-E Driver and drive the 3 leds in parallel.

Fichtenelch said:

I don't really understand why you want to use a boost driver and run the leds in series.
Personally i would go for a P7/MC-E Driver and drive the 3 leds in parallel.

Click to expand...

Well from what I have read on these forums is that is better to run LEDs in series to avoid having discrepancies of the Vf in the LEDs. So when I went to Cutter Electronics to order the LEDs I noticed they only sold them on a board either already wired in series or not wired up at all. So that led me to beleive that the info I read on these forum was probably right.

And please correct me if I am wrong, but if I was running them in parallel and running a good 2.8A driver P7/MC-E would this still not cause the same Amp draw from the batteries?

LEDs in parallel:
7.2V*X = 3.3V*2.8A (80% driver efficiency)
7.2V*X = 9.24W/.8
X = 11.55W/7.2V = 1.6A

LEDs in series:
7.2V*X = 9.9V*1A (80% driver efficiency)
7.2V*X = 9.24W/.8
X = 12.375W/7.2V = 1.72 (differences being from 3.0A vs the 2.8A on above driver)

I don't want to sound like a jerk... honestly I just really want to understand this better. The big problem I am seeing from the comments so far has been the draw on the batteries and safety not so much whether the driver can handle it. From the specs the driver can handle it and since I already have a 20mm board with the LEDs wired in series I figured I would use that. I also figured this would be a good thread to have out there since I couldn't really find any good concrete answers while searching for this.

I just want to use what is best for me with what I already have without having to wait another 3 months to get a different board from Cutter.

Thanks for your suggestion and any help you can give me.

sprinkle said:

Well from what I have read on these forums is that is better to run LEDs in series to avoid having discrepancies of the Vf in the LEDs. So when I went to Cutter Electronics to order the LEDs I noticed they only sold them on a board either already wired in series or not wired up at all. So that led me to beleive that the info I read on these forum was probably right.

And please correct me if I am wrong, but if I was running them in parallel and running a good 2.8A driver P7/MC-E would this still not cause the same Amp draw from the batteries?

LEDs in parallel:
7.2V*X = 3.3V*2.8A (80% driver efficiency)
7.2V*X = 9.24W/.8
X = 11.55W/7.2V = 1.6A

LEDs in series:
7.2V*X = 9.9V*1A (80% driver efficiency)
7.2V*X = 9.24W/.8
X = 12.375W/7.2V = 1.72 (differences being from 3.0A vs the 2.8A on above driver)

I don't want to sound like a jerk... honestly I just really want to understand this better. The big problem I am seeing from the comments so far has been the draw on the batteries and safety not so much whether the driver can handle it. From the specs the driver can handle it and since I already have a 20mm board with the LEDs wired in series I figured I would use that. I also figured this would be a good thread to have out there since I couldn't really find any good concrete answers while searching for this.

I just want to use what is best for me with what I already have without having to wait another 3 months to get a different board from Cutter.

Thanks for your suggestion and any help you can give me.

Click to expand...

Ok, forget my idea if you already have the board. i just thought about how i would start. and you still would have a high amp draw from the battery in parallel.

Fichtenelch said:

Ok, forget my idea if you already have the board. i just thought about how i would start. and you still would have a high amp draw from the battery in parallel.

Click to expand...

Ok so I had it right... yay! That means I am actually starting to understand this a little :twothumbs

Thank you for your time. I guess I should have asked before I ordered it wired in series. I got the "I want this now" itch only to have to wait 3 months on it to actually arrive.

Hi, I think you are on the right track, just consider to set up a test bench first to try it out before stuffing it into a light body. I have a large piece of Al plate I use to mount things on for bench testing.

If you end up needing more xp-gs to play with, I have a couple. I bought them from etgtech.com in December for a project, but ended up using a different LED for it. Some are individual, some are on stars. They are a nice LED, but that project has different goals than the XP-G is optimized for.

I am not really sure why people have Cree LEDs circumnavigate the globe when you can get them in LA with fast shipping, both plain and star mounted.