Voltage regulator question

I am making a LED blinder with 9 3w RGB leds and I am having some overheating problems. I am using individual PNP transistors to turn each led on and off. In order to make them work I have to tie together all of the emitters together and put them to the 5 volts off my arduino so that when I pull the base low it works. The problem with this is that it is causing uneven draw from my voltage regs. Each module (consisting of a reg, a transistor and an led all on a heatsink) was designed to regulate the power to only one LED but sharing power between the outputs of the regulators means that any minor difference in output causes a large difference in the current levels it is supplying. I dont want to run it like that for long because it will eventually burn something out, even with the heat sink.

I thought of a way to fix it but before I do all of the soldering I was wondering if somebody could tell me whether or not it will work. I was thinking that if i have a resistor in between the main line and the emitter on the transistor it would let path of least resistance take over and fix the problem. I was going to use some 1/8w 480ohm resistors just because I have a bunch laying around. The only problem I can see is that if it doesn't work out and it draws power through the resistors they would burn out under the 3w load. Heres what I was thinking of doing (not the whole schematic)

to arduino 5v<------------------------------------------------------------------------------------------------
| | | | Pretend like
its a resistor > > > > there are 9
| | | | of these total
these are the PNP's >|----- >|----- >|----- >|-----
| | | | | | | |
| | | |
To arduino digital pins--> | | | |

Theres also a LM317 regulator going to the emitters (after the proposed resistors) providing power and the collectors go the leds

Do you guys think this would solve my problem? Tell me if you need more info, thanks in advance

Something doesn't smell right here. You have a voltage regulator in series with each LED?? That regulator is essentially acting as a (variable) resistor, so it doesn't make sense to add more resistance.

Secondly, if you are in the 3W domain it's probably more efficient to use MOSFET's than PNP transistors and you will get less heating there.

Also, a bona fide schematic of what you are doing would be more helpful. The ASCII art attempt is appreciated but it's missing the regulators and still not all that clear.

The Gadget Shield: accelerometer, RGB LED, IR transmit/receive, speaker, microphone, light sensor, potentiometer, pushbuttons

I am running it like this. I have the main power line which is 12 volts and 12 amps. I have 9 3w RGB's in parallel to that, but in series with each LED I have a regulator and a transistor. In order to make the transistor work though I have to connect together the emitters and tie them to the signal negative. This works for the transistors but causes problems for the regs. BTW the transistors arent getting hot at all really, its all coming from the regulators.

My idea is to have a resistor between the signal negative and the emitter. This would cause resistance between each of the led modules but would not have any effect on the power going to the led from the transistor and regulator in a given module if that makes sense

Thanks for the replies but I tried out my idea an its working really well. If I have another problem I'll post a real diagram

but I tried out my idea an its working really well

So what did you do to fix it?

there is not a decoupling capacitor in sight.

Im a bit new to electronics and I'm still not really sure where capacitors are needed. I understand that they are used to provide extra current while the power supply is responding and also to reduce noise. I took apart the power source (a x-box brick I got specifically for this) and it has a bunch of them so I decided not to worry about it. This might be a bad approach but it seems to be working. I have a bunch of capacitors I just don't know where to put them.

In response to your question I just put resistors between the emitters and the wire going to the arduino plus. This causes the path of least resistance to keep one led drawing power from the wrong regulator and causing uneven heating.

I just put resistors between the emitters and the wire going to the arduino plus.

So you didn't do it right did you. You should have posted your schematic. The path of least resistance thing is a myth, all paths need to be considered.

I'm still not really sure where capacitors are needed.

see:- http://www.thebox.myzen.co.uk/Tutorial/De-coupling.html

Thanks for the link. Its working now but the NPN transistors I am using to sink the current are dropping a lot of voltage. They are making The whole thing quite a lot less bright, its because of the resistors I have between the base and my arduino but I dont want to get rid of them so that I dont burn out the arduino pin. How do I deal with this? Would replacing the power transistors with mosfets help?

How do I deal with this? Would replacing the power transistors with mosfets help?

Most of what you are saying doesn't make sense. Without a clear explanation and an actual circuit diagram (or at least a well drawn block diagram) it is going to be difficult for you to get useful advice.

download a copy of the schematic software from www.expresspcb.com and do up a schematic, it is easy to use.
Do a screen capture, paste into powerpoint, save as .jpg, and attach to a post.

Ill draw one up tomorrow during school and post it. Sorry its taken me so long to do so, its going to take a while and ive been avoiding it

Here it is guys. Sorry for the delay. Let me know if you need clarification. I did it from memory but I think I got everything right

Ok heres all the requested information

  • http://www.dealextreme.com/p/3w-led-emitter-on-star-multicolored-rgb-4530
    Red: 2.5V ~ 3.0V, 350mA Green: 3.2V ~ 3.8V, 350mA Blue: 3.2V ~ 3.8V, 350mA
  • PNP: BD436 Power Transistor (9)
    NPN: 2N5191 Power Transistor (3)
  • Resistor values posted on attached image, the resistors from the LEDs are 1 watt 4.3, 4.3 & 6.6 ohms for the different colors (blue, green, and red, respectively)
  • Yeah they are LM317T's futurlec, I tuned them all to 4.5 volts, which is actually not quite full power that the LEDs can take, I used 5.5 volts in my calculations but hey better safe than sorry
  • Yes I believe your talking about the ones coming off the regulators and they go to the power supply (X-box brick 12.4v & 12.4 amps)
  • Those resistors were my solution to the above problem, more in depth later in post
  1. I am using them to drop the levels that the resistors have to handle, and to get rid of any power fluctuations from the power supply, but considering that its for a xbox I don't think there will be any significant spikes or drops

  2. When I didn't have those resistors I was having uneven power draw problems. One regulator would be supplying power for several LEDs and because of it get ridiculously hot. I added the resistors in there to minimize the problem. My reasoning was that if there was resistance between each of the emitters the LEDs would draw the majority of their power from their dedicated regulator because of the path of least resistance. While it may have not totally eliminated the problem after I implemented them it was greatly diminished to a point where it is acceptable. I could not think of any other solution to this other than maybe diodes but I didn't want to mess with that.

  3. I think that my PNP transistors along the bottom are fine, I'm not detecting much voltage drop through them however the NPN transistors are causing a bunch of voltage drop which is a problem.

  4. sorry about that It was the way I took the picture, I wrote it going sideways not up

You ommited the value of the most important resistors in the circuit! The resistors for each LED segment.

No I didn't...

Resistor values posted on attached image, the resistors from the LEDs are 1 watt 4.3, 4.3 & 6.6 ohms for the different colors (blue, green, and red, respectively)

But your circuit is explicitly designed so that EACH regulator supplies power to THREE LEDs. I don't understand this issue?

Sorry I should have been more clear. It was in fact designed to supply for 3 LEDs but because they are all on the same star I referred to it as 1 RGB led. The problem is that it is supplying the power for other groups of 3 leds, ie it is powering 6 or 9 total, which is more than it was designed to handle. My solution was to minimize power sharing from one module (consisting of 1 PNP transistor, 1 regulator, and 3 LEDs all on one heatsink) to another.

Do they get warm?

No the PNP transistors aren't the problem. Those don't get warm or cause voltage drop at all its just the NPN ones for which both of those things are true.

How much voltage drop? How much current? Those 2N5191 are pretty low gain. It appears that you aren't fully turning them on ("saturating" them) and that is causing your voltage drop and overheating. You may have to make a DIY "Darlington Pair" and use another small transistor to fully turn on the 2N5191s.

Its kind of hard to tell but I would guess around 1 volt from the brightness differential. Thanks for that advice, that was what I was thinking of doing, I've got some spare signal transistors.

If it were me, I would abandon all those voltage regulators and just get a more suitable LED power supply. You can probably get a nice 5V or even 3.3V supply from a surplus place (or charity shop) for practically nothing. Keep it simple. Your scheme with all those regulators is WAY too complex and problematic IMHO.

In hindsight I probably could have just used higher value resistors but its all integrated and its working except for a few minor problems. I got the source I did because it was the cheapest one I could find that fit my requirements, I'm on a pretty tight budget.

Something I forgot to mention is that I'm having a problem with one of my LED's. When it has no signal but is still on the main power it is off as it should be, but when I give it signal from the arduino, regardless of high or low, it decides to stay on. I'm not really sure why and the fact that its off when there is no signal makes it seem to me that its not a shorting problem. If anyone has any idea what could cause this it would be great

Several things spring to mind.

  1. You have no capacitors on the input nor output of the regulators. This is needed to keep them stable.
  2. You are driving high power LEDs with a simple current limiting resistor. This is defiantly not a good idea.
  3. If something sticks on when turned off with a signal but is off to begin with it could be:-
    a) You haven't declared those pins to be outputs in the setup() function.
    b) A ground path is missing.

I am driving the LED's with both a regulator and resistors. I'm not sure what I could do to be safer than that. I definitely declared it because I just do a for loop in the setup to initialize all of the pins I use so it must be the ground connection is broken. Ill check that as soon as I set up all my stuff again. Thanks for the advice Ill let you know if it helps

I'm not sure what I could do to be safer than that

Lots of things.
You are driving it with a constant voltage regulator using a very low value of resistor to turn it into a constant current supply. This is not good. You need to use a constant current regulator so that the current stays the same as the voltage drop across the LED changes due to ageing, and temperature changes.