Any DC-DC adjustable regulator suggested to drive some LEDs

Question

Hello all,

I need a good adjustable DC-DC regulator suggestion. My main goal -for now- is to drive some LEDs in parallel, this is more or less what I'm looking for:

At least 600mA output with good efficiency (I think LDOs are not a good option here). I'm planning to drive 18 white LEDs (~3.2V, 30mA each), and of course if this can be used for other projects/applications that'd be ideal!
Adjustable: I need to tune the voltage a little down so the LEDs don't get too hot -adding a heat-sink is out of question- and as an added bonus this will reduce a little the power consumption
Very, very small package: I have very, very little room to work. This has to come in a very small package (SOT-23-5 or around that size) with very few parts required -and don't worry I can solder almost any SMD
Input Voltage: I want to power the thing with 3 or 4 rechargeable AA or AAA batteries (1.2V each).
US stock, around $2 each for few (maybe 4 or 5 of them)

At first I came with the ON Semi NCP1529ON Semi NCP1529 (NCP1529 datasheet) which comes in a nice and small UDFN6 (2x2mm) but tbh, I'm kinda stuck in the part where I have to select the right Inductor (L1) and C2 (Cout) for my needs -too hard to digest-, below a simplified diagram of what I have:

Luis

14rhb · Answer

Hi Luis, Have you had a look at the Texas Instruments WebBench software? There are tools in that for selecting LED drivers based on your constraints. I'll try and get you some links. Rod

Fred27 · Answer

I'd also agree with 14rhb that you'd be better looking at an LED driver rather than just a DC/DC. You'll get better control of the current through the LEDs without losing loads of power in series resistors.

What do you want to do with them? Right now apart from "in parallel" it sounds like you're just making a torch and I'm sure there's more too it than that.

genebren · Answer

Luis,

I did not see your simplified diagram?

This looks like a pretty good choice. I have worked with a few similar parts, and while I agree that component selection looks difficult, I have never needed to move too far away (if at all) from the recommended or typical component values (2.2uH and 10uF). If I can not find one of their recommended part numbers, I use the specification of one to search for a similar replacement part. The might be a pretty tough package to solder (UDFN6), but if you feel comfortable with that, then sure the space saving would be nice.

In a quick search, I found a similar part (AP3419, Diodes Inc.), which does not seem to be carried at Newark. This uses very similar typical value components (2.2uH, maybe a little bigger Cout).

Good luck with your design!

Gene

luislabmo · Answer

Hello Rod,

Nice tip. I tried after I saw your comment (first timer here), but unfortunately I didn't get any good match for my needs -I'll play more with it to see how it goes-, but I think I found a good pick from TI while looking at parts, TPS62290

Luis

14rhb · Answer

Hi Luis,

Looking at the TI webpage the two links I was thinking might help you, but you have probably now tried, are:

Light Emitting Diode (LED) Designer | WEBENCH Tools | TI.com

https://webench.ti.com/power-designer/

ps: TI are one of the best manufacturer's for providing samples if you need to test a design out.

Rod

shabaz · Answer

Hi Luis! I think it may be easier to go step-up, and drive a load of LEDs in series. TI has some parts that would provide around 40V, they are intended for backlighting (such as TFT panels). That's not a high enough voltage for 18 LEDs, so you could have (say) a couple of parallel chains of 9 series LEDs. That way, just a couple of resistors are needed, one per series chain. With the lower voltage all-parallel method, if there is no series resistor for each LED, unless all the LEDs come from the same batch, there will be brightness differences, it may not matter too much, but it could be noticeable. Modern LEDs seem to have less variation, but it could catch one out once these are being manufactured, so safer to have the resistors there to reduce this risk. EDIT: e.g. see LM2733, there is a table in the datasheet with component values for providing 30V at up to 110mA, that might just be enough to have two chains of 9 LEDs, or tweak the values for slightly higher than 30V.

Fred27 · Answer

I definitely think running them all in parallel is risky. There's a quick way to check though. I know Luis has a decent PSU as he sent me an adapter for the Keysight E36313A. (Thanks again, Luis!) Stick the LEDs in parallel, crank up the voltage slowly and see what happens. I suspect the LED with the lowest forward voltage drop will spoil the party and draw more current than the others. My guess would be at best uneven brightness and at worst that one LED blows. Two chains of 9 or three chains of 6 should make it easier. If you set the voltage to be just a little higher than the summed forward drops then you can get away with a low value series resistor and very little wasted power. Or maybe now an LED driver is feasible as you have only a few higher voltage chains to drive.

Fred27 · Answer

Regardless of the voltage I would have thought you're not going to get parallel LEDs illuminating properly without current limiting resistors. Below 3.2V you will get nothing and as you reach 3.2V one of the LEDs will surely start conducting before the others. I'd have thought the only way you would get away without limiting resistors is if you connected the LEDs in series, supplied just over 57.6V and relied on a supply that maxed out at 30mA.

This must be a problem that's been solved in a price and component count sensitive manner for all those cheap LED torches out there. Anyone know how that's normally done? Would the quickest and easiest solution be to cannibalise a dollar store torch?

luislabmo · Answer

Hello David, I see where you are going . I'm making sure to use LEDs with same part #/batch for my project, so in theory none should spoil the party. I see you love LED drivers , there is one catch with them, not good for driving LEDs in parallel ; if for some reason one LED stops working, current will go up for the remaining working LEDs which may spoil the party. I may add a resistor in series to avoid problems but then I will go back to the "a LED driver physically too big and won't fit" problem or the thing may get too hot if such thing happens. But, you and shabaz gave me a nice idea!. I might explore a step-up regulator and drive 2 or 3 chains in series with a resistor. Luis

luislabmo · Answer

Hi Shabaz! I'm planning to use same LEDs (part # / batch) so that is why I was planning to drive the LEDs in parallel, I completed my circuit design using the TI TPS62290 (the biggest component here is a 3x3mm inductor) but, I like your idea of using a boost/step-up converter to drive 2 or 3 chains in series with a series resistor. I will definitely explore this option too -the only downside I see here is efficiency but I think it may work too. Thanks for your input. Luis

luislabmo · Answer

Hello David

The problem with LED drivers is they are big for 18 LEDs and need many traces going around which takes some space -my main constraint-, that is why I'm trying to approaching the problem with an adjustable DC-DC regulator and not planning to use current-limiting-resistors as I will be providing a safe voltage below the rated voltage of the LEDs.... and Yes!, I'm making a torch but of an special shape and size for Halloween so I need the power circuit small.

Fred27 · Answer

I'm only a fan of LED drivers (in this case) because they control the current rather than voltage for you. Each output shouldn't be driving paralleled LEDs. Nothing should, whether it's a current or voltage source. Paralleled LEDs are never a good idea and that link you provided is exactly my point. It applies even more so to the voltage regulator you were proposing as to a current source like an LED driver. A step up regulator with LED chains definitely seem like the way to go. Kirchoff's law tells you the each LED in a chain has the same current and hence brightness. A small resistor should give you a good balance between power loss and sensitivity to the LED forward voltage drop. Don't forget to show us the suitably scary end result.

genebren · Answer

Luis, You are correct in your plans and assumptions regarding this approach for lighting LEDs. Leds to have steep IV curves, meaning that large changes of current can occur for a small change in voltage. Given this curve it is possible to control the current flow in the LED (an hence it's intensity and power dissipation) by varying the voltage applied across the LED without the need for current limiting resistors. I have worked on battery driven medical products, using LEDs, that used the very same technique for illuminating multiple LEDS at high intensity levels. I hope that you are able to find the right parts to pull this off. Gene

luislabmo · Answer

Well, to my knowledge , the current limiting resistors are required when driving the LEDs using a voltage higher that their rated voltage (since their current goes up exponentially ). When using an implementation that actively limits the current and/or feeds regulated and clean voltage within the rated voltage of the LEDs, one can skip the resistors. In my case i'm just planning to lower the voltage 0.1V to 0.2V this will lower their brightness but just a little (not noticeable tbh), but enough to keep them at a lower temperature -that won't melt my project. I've done it, but I'm improving it with an adjustable voltage regulator Luis

luislabmo · Answer

Hello Fred27 The " suitably scare end result " is complete!. Thank you ALL for all of your contributions. Luis

Any DC-DC adjustable regulator suggested to drive some LEDs

Top Replies