DIY 2cell constant current driver 6A

Flashlight Modding and DIY Parts
1

I wanted to power a XP-G2 in a UF-1405 with two cells. I know it can be driven from a single cell with 17mm FET driver, but the point is I want to power the led with 5-6A all the time and with diffenent batteries and bridges the current will be a lot lower most of the time. And so will be the output. So that is why I needed two cells for 8.4v and from that a current regulator to regulate the drain. And no special cells will be needed, full time turbo mode, etc. Much better in reality.

I found only 1 driver that can do this: 17mm MTN-MAXlp HP 1A-5.5A Low-Profile Buck Driver - 5V-18V
Which is fine, and in a perfect world I would get that. First of all it has no shipping to my country, and even if it had it will be 30-35$ total which is way too much for a few electrical pieces. And perhaps, it is unnecessarily complicated. If you have an idea for a cheaper chinese or EU version from this kind of driver please let me know.

So I thought to make an ultra simple circuit to satisfy my needs. And since this is budget oriented forum, perhaps someone will find it useful. The regulator is <2$, and the resistor is <1$, any board will do since the components can be floating in termal paste.

The idea is to use LM338t current regulator with one resistor that will dictate the current.

The LM338t should have a voltage drop of 3V if I am not mistaken. That means even if the vf of the led is 4.5V for a given current, the needed battery voltage will be 4.5+3=7.5V or 3.75 per cell. Depending on the needed curent, and the bin of the led. So in worst case conditiions, on low batteries, and very high vf led, the circuit should be able to deliver 5-6A to the LED. Which is nice. And since the voltages are really close to the needed, heat will be moderate.

This LM338 has a minimum passing current of 5A, and 8A typical. So it should handle 5-6A with a bit of a strech. (LM350 can be used for <3A scenarios, and LM317 for <1.5A) Operating temperature is <125C which means it will need to be cooled in the pill somehow but without touching the metal. The obvious choice is a silicone thermal pad.

Here is a calculator for the current and needed power for the resistor depending on the resistor. So 0.2Ohm resistor would be perfect for 6A current draw. The resistor would have to be at least 7.5W. So 2 resistors from 0.1Ohm @ 6W should do just fine. The resistors have a real world operating temperature of 100-120C. And will dissipate 7.5W of the power. At 6A at 4V led power should be in the 25W mark. And the efficency of the driver will be ~77%. Tailcap current should be ~4A.

For 4A LED draw, the resistor should be 0.3Ohm, power dissipation at resistor will be 5W. Led power will be 16W, and efficency ~76%. Tailcap current ~2,7A.

The efficency is not perfect, but the cost is very low. Especially as there is no need for special batteries, no need of bypassing springs, no need of high grade switches, nor ultra-fat led wires.

If I have made a mistake somewhere please correct me. Perhaps the idea is allready implemented in some driver, but I haven’t found it yet.

P.S. This can be used for any single led emitters XPL, XML2, XPG2 as is. With a little calculation and more cells even a 6V, or 12V even 24V or max 28V leds in series is easily possible.

2

You seem to only account for the power lost in the sense resistor.

If you also account for the power burned up by the regulator IC itself, the efficiency will be well below 50%.

At Vf = 4 V you start to go above 50% electrical efficiency only once the cells are down to 8 V.

3

If you are going to power your project with 2 or more cells and need cheap high-efficincy buck driver, better to use one from this:
http://www.ebay.com/sch/mtolya/m.html?\_nkw=&\_armrs=1&\_ipg=&\_from=

4

However, none of these are capable of more than 5amps. Or can they be resistor modded or otherwise?

5

Yes they are moddable. Seller claims that they can work at 4.8Amp for long time.

6

But you will need to use protected batteries only, there is no LVP in this drivers.

7

Use the two cells in parallel. Then you can have the current without the low efficiency and use a more common driver.

8

Hello!

This may be what you're after: FX6 22mm 6V - 8.4V 5A 4-Mode Driver Circuit Board for Cree XHP50 / XHP70 / MT-G2

Recommended reading: 2 Li-ion 3 amp Driver Super Output Current Mod "Now with Larger Toroid driving a MT-G2" -> I am doing my own mod on one of these. Got mine here.

Cheers ^:)

9

2 cells in parallel will have same problem - current will came lower when cells are half empty and led Vf at 5-6Amp is bigger than cells voltage.
Topicstarter want to use full current until cells are empty.

10

He wants to drive a 3V led with two cells rather than a 6V led so he needs a buck driver. The poor efficiency of a linear driver like the 338 won’t be much improvement over a single cell. Losing 15W in the IC should be a clue to that. If you want to realize the power gained from an additional cell you need to use a method that places power usage predominantly in the led. That’s going to be a buck driver with 2S cells or a number of choices with 2P cells. SmallSun ZY-T08 would be a fine choice for a host for 2P.

11

I’d say a similar problem to a lesser degree but still vastly better than simply pissing away half the power.

12

Once again.
XP-G2 have 4.0-4.2v Vf at 5-6Amp.
You can use any quantity of cells in parallel, but when voltage drops to less than 4v led will get less amps and when cell voltage is 3.0 you will get less than 0.5Amp neverless how much cells you have in parallel.

13

In any case he needs a buck driver not the lm338. The MtnE drivers are expensive because they use quality components and are made where labor isn’t cheap. They are still competitive with other small market driver suppliers

14

Calm down my dears!

I already gave mdeni a couple of good and nicely priced solutions, though the second one requires a custom inductor, a handful of additional current sensing resistors and a higher tier schottky (gonna use a 20SQ045 on mine, positioning the bigass inductor between the two boards :-) ).

No doubt she/he'll understand that extremely inefficient LM338 solution.

Cheers ^:)

15

I wanna food fight. :stuck_out_tongue: There are times when efficiency isn’t top dog and maybe this is one of them. I thought those drivers you mentioned Barkuti were for 6V LEDs and won’t work for a 3V led? There are lots of different things that go into the cost of a build and we don’t really know much about this one other than two series cells, a 3V led, and a possible driver but we can infer that because of the driver choice, the mass of the host head will need to almost double which could affect the size of the billet needed to machine it from and the machine time to do so. Those two together could easily swamp the cost of a more expensive driver.

That said, maybe usage will be limited to short bursts, or, it may be more appropriate to suggest the use of nimh cells where overhead voltage can be dialed in more closely(like with 4S nimh’s). It makes more sense to me to add mass to add supply than to add mass to deal with waste. 4 X 1.4V is only 5.6V vs 8.4 and fully discharged is still at 4V above the emitter Vf and doesn’t require anything more than a 7135 driver with a slave board without even a Zener mod, just a little attention to adequately heat sinking the board(s).

Or, maybe Mdeni is just going to go on as planned regardless of our input, I’ve been there myself, it seems to be the case that there is often an overlooked pearl so posting something isn’t a bad idea even if few seem to agree with you.

16

From Djozz tests the xpg2 is around 4.5V at 6A and the lm338 has 3V overhead so it falls out of regulation at 7.5V and is wasting from 23.5W down to18W on the way. In the meantime the led is contributing another 27W to the heat load. 3.75V includes voltage sag so resting voltage is higher and much less than half the cells capacity used before falling out of regulation. If the driver kiriba-ru suggests works and the seller is being honest about its capabilities then it would almost get there at better efficiency. Nothing seems to be a really good fit all the way around though. I have to think that a 4 or 8 X AA host would pin the donkey much better in several ways.

17

Thank you for correcting me. I probably missed something, and perhaps that was the IC power loss. MTN is the best for my application, but I am really amazed that there are very few alternatives from china. Especially on a forum like this. I will abandon the LM idea since you say that the losses are high. Losses are of no consern to me because I will be using 2x26650, but heat generation is bad with that much loss.

The Vf problem in XPG2 is real, and very big. One cell driving is not possible because the led will not draw 6A from that low voltage, and when the cell drops below 4.1V which is most of the capacity the current will be even lower. That is why multi cell operation is needed for a XPG2 to draw constant 5-6A current. Parallel cells will also deliver very low current below 4.1V even though they will increase the runtime. Series cells will improve the runtime, and on top of that drive the led as god intended. Especially considering the new batches XPG2 which have higher vf.

Very nice suggestions, the Russian drivers seem very good, but need modding. I will have to read upon it to find out more about the possible mods to increase current above 5A. If you have some links please tell me, I will gladly look into that.

Any other suggestions that will keep 5-6A constantly (from full to empty cells) into a XPG2/XPGHI are welcome. :innocent:

18

Looking around there doesn’t seem to be much in the way of multicell AA zoomies so if 6A is a must then forget about efficiency and wear gloves. A Maglight D with an Ahorton lens would allow for 4 or 8 X AA quite well but if you can’t get a MtnE driver then you probably have trouble getting a MagD.

19

The drivers I pointed out work with a single led, Rufusbduck.

Features reverse polarity and led short-circuit protections. There is the official explanation on how to tune-up the current output. Combined sense resistance is ≈45mΩ, sense voltage ≈225mV. If the driver indeed does ≈5A, the picture figures do not seem accurate to me. The R082 would channel close to 2'75A and the R100 a smidge above 2'25A. Another R200 on top of 'em and 6'1+ amps (27+W over an XP-G2, seriously?)…

The other driver is a well known one, once sold with a bigass inductor for 5+A in certain places. Kaidomain has it in 3 different versions. Smaller board footprint of the drivers I comment about here. Check out the thread above.

Seller “mtolya” has some nice stuff, like this one: LED driver LDRADJ29-4.8A, 4,8A, 12V, DC 8V-38V Multi-Mode for CREE XHP70 (6V) Bigass driver board, though.

Cheers ^:)