Best way to power 3w RGB LED?

I'm not asking for any schematic/diagram here...just some general discussion/feedback (suggestions)

I'll make/post my Eagle files for 'review' once done...

I want to whip up a custom pcb that is basically a stripped don 'Arduino' with an on-board RGB led (to stick in some lamp project), and an HC-05 bluetooth module embedded.

anyways.. I started to think that using a high powered RGB LED (star) might be better way to go (for maximum brightness)..

the specs for the led are as follows:

Red: 2.5V ~ 3.0V, 350mA
Green: 3.2V ~ 3.8V, 350mA
Blue: 3.2V ~ 3.8V, 350mA

  • Star size: 20 x 20 x 1.6mm

  • Note: each color channel is able to take a maximum of 350mA of current, thus a 1050mA combined current input and the 3W rating

I have only designed/worked with these RGB led stars in a single channel fashion (1 regulator, 3 x transistors and only one channel/die active at any time, no color mixing).

This time I want to be able to color mix. (so have more than just one single die active at any given time)

Would I just plan on:

  • using a 12v wal-wart to power this thing? Is there a better approach here? I not too keen on trying to use a 12v input source that gets chopped down to 5v (or whatever) by the vRegulators

  • would I need to use 3 separate vRegulators? one for each color/channel?? (along with 3 transistors..etc)

In general what would the best approach to tackle this with? Any other suggestions?

I'm not very clear/comfortable on making a constant current circuit for this type of set-up..

I had used (which I was told later is not the best approach) an LM317 as my base

BC817: http://www.digikey.com/product-detail/en/BC81725MTF/BC81725MTFCT-ND/1305367

And the:
LM317: http://www.digikey.com/product-detail/en/LM317DCYR/296-12602-1-ND/443738

I had followed the datasheet and was my first/only constant current anything.

If there is a better way with a different chip that is even easier to set-up (not a bunch of other components needed is a key point for me, along with easy to follow/understand wiring, since I'm just a hobbyist) :slight_smile:

the best battery pack, that has enough voltage/current .....that allows for all components (arduino circuit, bt module and rgb led star) voltage drop.... close enough in the vF of the leds (and the vRegulator needs) so it doesnt burn un-needed heat...etc with a nice, EASY chip for making constant current circuit for the led as well..

thoughts? suggestions? feedback?

Thanks in advance!

-xl

I'm not very clear/comfortable on making a constant current circuit for this type of set-up..

Well get comfortable with it. It is the only way to drive a power led correctly. Everything else is a fudge that is inefficient, under runs the LEDs or melts them.

Have a look here, pretty much what your wanting to do but minus the BT. I have attached the schematic and you have a couple of spare pins to drive a BT module.

ATtiny-RGB Moodlamp.pdf (8.88 KB)

xl97:
Would I just plan on:

  • using a 12v wal-wart to power this thing? Is there a better approach here? I not too keen on trying to use a 12v input source that gets chopped down to 5v (or whatever) by the vRegulators

Why start with 12V? 5V is plenty.

Riva:
I have attached the schematic and you have a couple of spare pins to drive a BT module.

Yes that will do, it is a constant current driver.

However the actual current it drives at is dependent on getting those exact and very small values of power resistor.

Thanks for the reply guys.

@GrumpyMike..

haha.. I'm trying to get 'comfortable' with learning how to make a 'correct' constant current driver. Why I am askingfor advice/suggestion on an easy to use chip for this purpose. :slight_smile: Something that doesnt have alot of extras/fluff to side-track a beginner like myself to all this..

I have made this pcb in the past (as mentioned) using an LM317 chip:

but if I recall this was an old chip and an old approach (although it worked and what was in the datasheet)

I would just do this 3 times (1 for each channel/die of the led) if not for the recommendation there was a better way. (so I am trying to ask/learn) :slight_smile:

@fungus

sorry, I guess I was thinking (erroneously) that I needed to have a power supply with enough voltage for all three led channels to be on...

So if I used a 5v wal-wart as my 'source'..... would that mean running my Arduino without a vRegulator? (since I usually use some sort of vReg to keep a constant 5v feed to my Arduino circuit... and then a +3.3v regulator fed -from- the +5v regulator..to power any other components..in this case the BT module)

so just a 'raw' +5v power source (direct from wal-wart for example).. and then all my circuits in parallel from that?

@Riva-

Thanks for the schematic. I only briefly looked at it.... (and am already confused..lol).
which I'm sure stems from my lack of experience in this area.. :slight_smile:

but it uses BOTH a transistor and a moseft?..

excuse my ignorance on this question,...is normally how a constant current driver is set-up? (and not through some dedicated IC or something?)

I have not read the link yet.. but will do so in hope I keep learning. Appreciate it.

@GrumpyMike (again) lol..

Your comment makes me feel like this is not the best way? I'm also not clear what you mean by 'power' resistor? Are you referring to resistors that rated for like 1w and over? (would like to know so when I see the term again in use I understand the meaning)

Also (on a side note) what ever happened to that board I shipped out to ya in the UK? :slight_smile:

Thanks everyone.

Well...

That circuit is really only good enough to drive red or blue or green via the cc lm317

But if you switch red and say green on at the same time, red will suck up all the current due to it's foward voltage.

that is correct.. as stated this is NOT the current circuit/design... this is an example of the only type of constant current driver I have done before.... (and also as mentioned it was only designed to use 1 channel/color at a time)..

however I was told that using the LM317 it not the best approach. if that is not true, and this is fine... then I'll just add 2 more and make so I -can- color mix.... but if it is NOT the best way.. then I was/am asking what other IC can be used? or some links to a nice and easy to follow set-up of a constant current driver. The one posted doesnt seem a as easy or straight forward as using an LM317 though... (uses more components...etc)..

a quick google search shows that many people use this LM317 set-up in a similar fashion though??

so to summarize, yes you are correct.. that particular board design & schematic were designed to only have 1 channel on at a time. :slight_smile:

thanks!

The thing about both those sorts of constant current drives, the regulator and the simple FET is that both work by having a seriese resistor that has to take all the current. It has to be big enough to develop enough voltage across it to act as the feedback for the regulating device, regulator or FET.

This means that the actual current you get is determined by the precise value of this resistor. It is not easly changed if you want to change the current and you are stuck with the values of resistor that they sell unless you want to mess about with seriese or parallel combinations.
A better circuit involves the same sort of arrangement but with an op-amp in the feedback path. The gain of the amp means that you can have a much lower value of feedback resistor, so less power is wasted. Also the amp's feedback resistors can be used to set the current. You can even make one a small pot so you can set the exact value you want.

xl97:
@fungus

So if I used a 5v wal-wart as my 'source'..... would that mean running my Arduino without a vRegulator? (since I usually use some sort of vReg to keep a constant 5v feed to my Arduino circuit... and then a +3.3v regulator fed -from- the +5v regulator..to power any other components..in this case the BT module)

so just a 'raw' +5v power source (direct from wal-wart for example).. and then all my circuits in parallel from that?

Yes.

(Providing your constant-current circuit works with 1V of headroom)

You might want to look at the AMC7135 chip.

They're cheap and apparently you can apply a PWM signal to the VDD pin

(caveat emptor, it's not in the datasheet and I never tried it myself...)

xl97:
A quick google search shows that many people use this LM317 set-up in a similar fashion though?

Yes, but it is basically a "demonstration" project, quite unsuitable for serious use as you immediately suffer the dropout voltage of the LM317 plus its regulated voltage of 1.25V so you are losing at the very minimum, 2.5 to 3V to start with. This is entirely unsuitable for a 5V supply system, OK for supplying single LEDs from 12V where you don't care about the power wastage (and are correspondingly, heat sinking the LM317s).

The two-transistor (or perhaps one transistor and one FET) circuits are the way to go.
Dropout voltage less than a volt. By connecting the base bias resistor of the first transistor to your Arduino output, you are getting the driver function into the bargain. And yes, you may need a heatsink even then.

Sorry for the late reply here..

(didnt get an email notification)..

I'm planning to give it another go this weekend (finally).. so I'll have to try and search around and see if I can understand your schematic (which I appreciate) that you posted.

So would I be using/making one of these 'circuits' (as you posted) for each die/channel on the led?... right?

The 'O/I' is the Arduino pin...correct?

What are the components under the Vcc+

thanks!
-xl

What are the components under the Vcc+

Your LEDs.

The 'O/I' is the Arduino pin...correct?

Yes it means either a logic 0 or a logic 1

So would I be using/making one of these 'circuits' (as you posted) for each die/channel on the led?... right?

Yes. That circuit as two LEDs in series that is a good way to drive both at the same constant current if you have enough voltage to drive them.

Just tossing ideas out there, but the WS2803 chip has 18 constant current sink channels that can each do 25ma, with built in PWM. That's enough for 150ma per channel on an RGB. I'm working on a similar project but for me 150ma is bright enough, and it lets me have a smaller heatsink and use less power.

On ebay there are positively tons of cheap adjustable switching regulators. I love these things because they cost about $1.20 and take up to 35V in. I suppose one way of doing things would be to regulate to 7 volts or so, put two RGB LEDs on one heatsink, run each color in series, and drive them at 150ma with the ws2803, and use a 7805 to regulate down to the 5v that your AVR and bluetooth wants.

By paraleling the channels you also get a few bits more resolution, because you can control the sub-channels of the RGB channels.

@Grumpy_Mike-

thanks for the reply/clarification..

  • oh they 'are' leds.?.. (I wasnt sure as there were 2 and how it applied to my set-up/question)..

  • I believe I have only seen something like I/O (input/output) before... so I wasnt clear what it meant.

question: in the schematic posted, wouldnt BOTH leds be pwm'd at the same time since there is only 1 Arduino pin in use here?

if my goal is to have each die on the RGB led star then I need to have one of these circuit designs/layouts for 'each' channel (thats what you're saying 'yes' to...correct?).. that means only one 'led' per 'circuit'...

using the schematic design above as an example... this wouldnt work (as it is) being as there is two leds on one (constant current) 'circuit' and I want to individually PWM control each color.

(sorry if some of these seem stupid or obvious..but when/while still learning., when I see slightly different things, I'm not always sure if its how I'm supposed to do things...just an 'example' but not directly transferable to my project/question..etc. so I ask to can learn from solid building blocks/facts in baby steps)

Is that a 4.7k resistor? (or 4.7M?..thats a pretty big resistor then..no?)..lol

Hopefully I can whip up a variant to my LM317 constant current approach/pcb this weekend and send out for some pcb's

@EternityForest-

Thanks for the alternate approach suggestion.. so then all (6 per group) pins would would be connected together and go to one die/color/channel?

Something to keep in mind... but I think trying to learn how to make a proper constant current driver circuit will just help me in furture projects to come as well.... :slight_smile:

xl97:
using the schematic design above as an example... this wouldnt work (as it is) being as there is two leds on one (constant current) 'circuit' and I want to individually PWM control each color.

Yes you need a separate driver per LED colour (Red, Green & Blue) but most larger LED's don't have a single large LED but several smaller ones in series (see image) and this is what the hand drawn schematic depicts.

557903260_o.jpg

Hi guys-

so I'm finally back to trying to make a real PCB for this little personal project..

Re-reading the comments so far to get back up to speed..

I have a question about the suggested way Paul_B has posted.. and the approach Grumpy_Mike mentions.

Paul_B:

Grumpy_Mike-

"This means that the actual current you get is determined by the precise value of this resistor. It is not easly changed if you want to change the current and you are stuck with the values of resistor that they sell unless you want to mess about with series or parallel combinations.

A better circuit involves the same sort of arrangement but with an op-amp in the feedback path. The gain of the amp means that you can have a much lower value of feedback resistor, so less power is wasted. Also the amp's feedback resistors can be used to set the current. You can even make one a small pot so you can set the exact value you want."

Which make sense....about the resistor sizes....etc and being able to adjust the current amount is a nice touch/feature as well..

But I am not experienced enough to make a call either way... or know if one is easier than the other..etc

Being a constant current driver.. the led(s) will still just pull whatever they need...correct? (always problems with the current sourcing vs sinking stuff still) .. or would each 'circuit' (constant current feed) need to be tailored for the led/color is will be connected to? if applicable... in this case I believe all current needs are the same, but voltage requirements differ for the red die..etc

Is there a diagram or a keyword to use for when searching for a schematic/diagram that uses an op-amp as Grumpy_Nike suggests?

thanks!
-xl

Jumping back a bit, when you're designing your board make sure you're prepared for the heat that these linear regulator circuits generate. Assuming the cheap, overseas board houses, use the full 5cm x 5cm size and lots of copper fill. And even then you should try to design your board in such a way that leaves you open to add additional heat dissipation.

hmm..

good idea.

I was only going to be designing the pcb for me to mount a 3W RGB star led, and have a heatsink under that.

I wasnt thinking about the regulators.

I was planning on using a +5v wal-wart.. or possibly a +7.4v Li-Ion pack (is there a way to design to use either?... wait, now I'm just getting ahead of myself!)...haha

what type of heat dissipation are you talking about here? Like heatsinks? (what type of design modifications do I need to not use any ugly heatsinks!?) :slight_smile:

I wanted to use mostly SMD parts, and keep the foot print as small as I can.

Do both approaches suggested (2 transistors vs op-amp) have heat issues to contend with?