Need help on making a LED candle circuit brighter

Whaleboy01

Hello! I'm quite the novice at circuit design, only previously taking a stab at making an automated shutdown controller for a vehicle and mounted PC. Even that was gleamed from various circuits I found online. But this time I need to actually ask someone more experienced myself on this to see what everyone recommends. I have no clue when it comes to the high bright LEDs.

So I came across this lovely circuit in Nuts & Bolts for a candle flame simulation based on the Basic Stamp 1.

http://www.parallax.com/Portals/0/Downloads/docs/cols/nv/vol6/col/nv126.pdf

My intent is to build it with my 6 year old BS2 incorporating the 3 LED's per candle flame, the potentiometer, and the sensor for simulating the wind. Very exciting! I want to incorporate this into several wall sconces on the hallways in my house.

Now, one LED alone simply isn't bright enough for me. I would prefer to get close to the equivalent of a 35-40 watt light bulb.I discovered these that are warm white, 360 degree, and 10,000mcd

http://www.superbrightleds.com/moreinfo/component-leds/12mm-warm-white-led-360-degree-viewing-angle-10000-millilumens/843/

I figure 3 of these clustered is close enough to the brightness to be sufficient.

NOW, I'm assuming that driving these with the UNL2003 wouldn't work out very well.. Nor reducing the voltage via resistor? Or maybe that would be ok somehow? I plan on powering the circuit with 5 volt or 12 volt. Should I be using a MOSFET and would that still operate properly with the randomized dimming of the program?

Thank you so, so much in advance for the help!

Mike Green

If you look at the specification for your LEDs, you'll see that they draw 350mA maximum. The ULN2003 is rated at 500mA per output, so it should be fine (warm maybe, but fine). The LEDs have a forward voltage of 3.5. Two in series would be 7V. Three would be 10.5V. If you're going to use 5V for the LEDs, you could only have one per ULN2003 output. If you're going to use 12V for the LEDs, you could have three in series. You have to use Ohm's Law to figure out the resistor value like this:

Your power supply is 12V. The ULN2003 has a 1V drop across it when on. That leaves 11V. The LED has a 3 x 3.5V = 10.5V drop across it. That leaves 0.5V for the resistor. Ohm's Law says that R = E / I or R = 0.5V / 0.350A = 1.42 Ohms. To get that, you could use seven 10 Ohm resistors in parallel (10 / 7 = 1.43). Make sure that your power supply provides 12V and not something higher. If it's a higher voltage, you'll need a larger resistor.

You'll also need a much larger capacitor than shown in the Nuts and Volts article. That uses a 250uF electrolytic. You're running over ten times as much current through your LEDs. That means that you'll need at least ten times as much capacitance. You can always use ten 250uF 30V capacitors in parallel to get the same effect, but you should be able to find larger values at the same voltage rating and use fewer in parallel.

I suspect that this candle is going to be much brighter than you think.

Whaleboy01

Mike,

Thank you so much for responding! I've been doing more research since your post as it helped me get a better grasp on the circuit, as well as forcing to reread the original doc several times.

So, I decided to go with one LED per pin, for a total count of 6 LEDS per 6 pins from the stamp, allowing me to cluster 2 sets of 3 LEDs for simulated flame movement in addition to the flicker. My biggest issue here was reduction, from what I was reading it seemed like a bad idea to simply use a resistor per LED as they would generate considerable wasted heat. I would like to store the device in the attic so I don't want to take any chances with blown resistors, fire, whatever.

I started looking into alternatives (LED drivers, etc) and came across a simple (though somewhat inefficient) circuit using a LM317. They're really cheap and allow the reduction required for each LED on each pin.

I put together a quick schematic (attached below), including your recommendations and incorporating this circuit. However, I'm not sure if this will work properly?

The other issue I'm having is voltage required in this setup. I now know when putting LEDs in series there is always a voltage drop, so you have to add up each LED to get to your final required voltage (+voltage drops for whatever IC's). But what about on this circuit where they are in parallel? Is the 12v good enough? Should I consider using 12v for the LEDs only, than I separate supply for the stamp? What are your thoughts?

Oh, and thanks for the advice on the brightness! haha.. I would rather it be too bright than not bright enough at this point.

Mike Green

So, if you're using a 12V power supply and you're using a single LED in the circuit, how much power is going to be turned into heat in the LM317? I've given you all the steps in my previous message. Power (W) = Voltage (E) x Current (I). The same amount of power will have to be dissipated whether you have an LM317 or a resistor in the circuit.