Need help with powering LEDs, ASAP

[ClockMan]

I have recently started adding lighting to my clocks with fiber optics and LEDs to light them. The backs of the clocks have limited space, so I am powering from 2 AAA batteries. (A 1" X 1" proto board will fit). Works OK on LEDs with smaller forward voltages (1.7-2.8) with 47Ω 1/2W resistor, but need to get higher voltages for the blue and white which are over 3.0V. Also, need help with getting more life out of batteries, lights getting dim after a few days, and out after 2 weeks. Would 1/4W reistor help? Tried an LDR, but they pull 15mA when in light and the LEDs pull 20mA, so using almost as much battery when LED is off. I am a beginner with electronics, please keep it simple. Appreciate any help.

 

[Hero999]

For higher voltage LEDs use a Joule Thief circuit.
**broken link removed**
Joule Thief Boost Circuit

Another option is to use a 74HC04 voltage doubler which requires no wound components, I'll post a schematic if you're interested.

Use a larger battery or higher value series resistor if you want it to last for longer.

 

[MOSFET KILLER]

There is a type of switched mode power supply called a joule theif. You could put those two batteries in parallel for more current and the joule theif would step the voltage up enough to power those LEDs.

Hero, you beat me by 2 minutes.

 

[ClockMan]

To Hero999
I saw the Joule thief, but not good with electronics. Would not know how to wind the coil for the expected output. What is the voltage doubler you mentioned? Schematics would help.

Thanks.

 

[ClockMan]

To MOSFET
I am interested in any options that would help. Could you supply a schematic of what you are talking about ?

Thanks.

 

[audioguru]

Go to the website of a battery manufacturer such as Energizer or Duracell and look at the datasheet for an AAA alkaline battery.
They can supply 20mA for only 40 hours when the voltage will be so low that the LEDs will be extremely dim. The voltage drops the entire time so the LEDs are continuously dimming. Use AA cells that have over twice the capacity (life during the same discharge current) as AAA cells.

 

[MOSFET KILLER]

Tim Williams has a good page on **broken link removed**.

 

[acmefixer]

Rechargeable Battery May Be the Solution

I have recently started adding lighting to my clocks with fiber optics and LEDs to light them. The backs of the clocks have limited space, so I am powering from 2 AAA batteries. (A 1" X 1" proto board will fit). Works OK on LEDs with smaller forward voltages (1.7-2.8) with 47Ω 1/2W resistor, but need to get higher voltages for the blue and white which are over 3.0V. Also, need help with getting more life out of batteries, lights getting dim after a few days, and out after 2 weeks. Would 1/4W reistor help? Tried an LDR, but they pull 15mA when in light and the LEDs pull 20mA, so using almost as much battery when LED is off. I am a beginner with electronics, please keep it simple. Appreciate any help.

There are about a thousand milliamp-hours in a AAA cell. If you draw 20 mA from it, it will last about fifty hours. If you draw 1 milliamp from it, it will last a thousand hours. So it depends on how little brightness you can put up with to extend the battery life.

One solution is to use rechargeable batteries. Another is to use solar cells and light to recharge the cells.

 

[Hero999]

To Hero999
I saw the Joule thief, but not good with electronics. Would not know how to wind the coil for the expected output. What is the voltage doubler you mentioned? Schematics would help.

Thanks.

I had a feeling you'd say that.

The two transistor Joule Thief circuit I posted uses a pre-made coil and an off the shelf pulse transformer or common mode choke can be used rather than the wound coil on the one transistor Joule Thief.

Another option is to use a capacitive voltage doubler, see attached, which requires an oscillator such as the 7555 or a CMOS inverter. Two AA cells must be used because the ICs I suggested need at least 2V to work.

 

[acmefixer]

I had a feeling you'd say that.

The two transistor Joule Thief circuit I posted uses a pre-made coil and an off the shelf pulse transformer or common mode choke can be used rather than the wound coil on the one transistor Joule Thief.

Another option is to use a capacitive voltage doubler, see attached, which requires an oscillator such as the 7555 or a CMOS inverter. Two AA cells must be used because the ICs I suggested need at least 2V to work.

The simplicity of the single transistor JT makes it unbeatable. Another point that is important for multiple colors is that each color has a different voltage drop so if they are in parallel it makes it difficult to get them all working from a single driving circuit. If you connect the red, green and blue LEDs in parallel without resistors the red LED will hog all the current because it has the lowest forward voltage.

The 7555 and CMOS inverter circuits must drive the LEDs in parallel. The single transistor JT will drive all 3 or even 4 colors when they are in series, and the current is the same in all of them.

To wind the coil, all you need is a AA cell, and some paper and tape. Here is a link to a JT wound with an air core. You can see how simple it is, just some wire wound around the battery. As long as the polarity of the windings is right, it should work. I would connect all three colors in series and reduce the turns in the feedback winding to half the primary winding. Use a 2N4401 and use a resistor of 10k or more to keep the current down and extend battery life. I'll try to tack solder one together tonight and post a pic.

 

[ClockMan]

Sorry to be a pain, but I need to start over. First, I appreciate all the repsonses so far, but the links are not helping much. I am a clock maker and a dummy with electronics. Will need things explained thoroughly how they work. Exactly what I need is a way to power a single LED from low voltage (1 or 2 AA or AAA, maybe 9V). Needs are long battery life with maximum LED brightness, don't want to be changing batteries constantly. Needs to be small, cheap and simple. All this needs to fit into the back of the clock along with the clock mechanism. Or, need another way to light the fiber optics. Would another light source be more efficient than LEDs?

 

[acmefixer]

To MOSFET
I am interested in any options that would help. Could you supply a schematic of what you are talking about ?

Thanks.

I soldered one together and took a pic, then **broken link removed**. You can read and click on the pic to see the actual simple circuit. It's very simple and gets the job done. You can follow the wires in the picture and build one up.

 

[ClockMan]

@acmefixer

That looks pretty good, cheap and simple. Should be small enough, too. What kind of battery life could I expect from this? Would I need to put a higher value resistor in for only 1 LED? Could I put a pot in line, adjust to get the brightness I want, then check the pot for the resistance value, then know what reistor I want? Like I said, I am a beginner with electronics. I undertsand volts, amps, reistance, but not familiar with coils, inductors, tranistors. Would appreciate explainations if not too much trouble, and description of how that Joule Thief thing works. Also, MOSFET KILLER said something about putting 2 batteries in parallel to get more current. Is this an option with this type of circuit? What if I wanted to add an LDR to automatically turn the light on at night? I considered the solar cell to recharge the batteries, but don't know if they make a solar cell small enough to fit my clocks, or have the power to give a good recharge. Also, would not know how to wire that either.
I really need to keep these simple and cheap, because if they work OK, I would like to start selling the clocks with the lights in them. They look cool, but can't make them profitable with expensive parts, and won't have happy customers who constantly have to change batteries or are unsatisfied with the quality/performance of the lights.

Thanks for the help so far, everyone.

 

[acmefixer]

@acmefixer

That looks pretty good, cheap and simple. Should be small enough, too. What kind of battery life could I expect from this? Would I need to put a higher value resistor in for only 1 LED? Could I put a pot in line, adjust to get the brightness I want, then check the pot for the resistance value, then know what reistor I want? Like I said, I am a beginner with electronics. I undertsand volts, amps, reistance, but not familiar with coils, inductors, tranistors. Would appreciate explainations if not too much trouble, and description of how that Joule Thief thing works. Also, MOSFET KILLER said something about putting 2 batteries in parallel to get more current. Is this an option with this type of circuit? What if I wanted to add an LDR to automatically turn the light on at night? I considered the solar cell to recharge the batteries, but don't know if they make a solar cell small enough to fit my clocks, or have the power to give a good recharge. Also, would not know how to wire that either.
I really need to keep these simple and cheap, because if they work OK, I would like to start selling the clocks with the lights in them. They look cool, but can't make them profitable with expensive parts, and won't have happy customers who constantly have to change batteries or are unsatisfied with the quality/performance of the lights.

Thanks for the help so far, everyone.

The battery life depends on the brightness. Current draw was 25 mA for this one, so 1000 maH divided by 25 is 40 hours. But it's not all that simple . As the battery voltage drops, the current drops, and the LEDs get dimmer, and the JT continues to run at a lower brightness. Build one and try it and find out. You don't even need solder, the circuit is so simple. Just use some fine wire to hold the joints together.

You can put a pot and resistor in series, say 4.7k and a 25k pot. Adjust it for the brightness and current you want. For your purpose, it would probably be best to use the batteries in series for 3V. It would be a good idea to add some form of control to turn off the LEDs when no one is around. A Daylight Sensor would be a good idea. A motion sensor would be even better but that would add complexity and be more expensive. The simplest and most effective way to extend the battery life is to flash the LEDs. If you turn them on for a small fraction of a second and then off for the remaining second, then the battery life could be tripled or more. Ten percent on time gives ten times the battery life. I blogged my Blue Blinky that ran 21 months and used a single AA cell. But you have to be realistic. The more you ask from the batteries the more often the batteries will have to be changed. In my blog I documented my Supercharged Joule Thief which has greater efficiency. I also have the SJT with a capacitor that makes it flash.

It's good to do a bit of research, but you'll never know if what you find is suitable for your purposes until you try it. So you're going to have to stop asking questions and build something.

See these URLs for more info.
Joule thief - Wikipedia, the free encyclopedia
**broken link removed**
And of course my blog, watsonseblog.blogspot.com which has lots of JTs of various kinds, with pictures, too.

 

[Hero999]

The simplicity of the single transistor JT makes it unbeatable.

I agree, the only time I'd really recommend a voltage doubler is when you're already using a couple of logic gates, a microcontroller or want to flash an LED.

You also might be able to buy a charge pump IC which can power an LED from two AA cells just by adding a capacitor.

Another point that is important for multiple colors is that each color has a different voltage drop so if they are in parallel it makes it difficult to get them all working from a single driving circuit. If you connect the red, green and blue LEDs in parallel without resistors the red LED will hog all the current because it has the lowest forward voltage.

You really shouldn't parallel LEDs anyway, even if they are the same colour the voltage drops will differ enough to prevent adequate current sharing.

This can be mitigated by using a series resistor for each LED or a small current balancing resistor if the LEDs are run off a constant current source.

 

[ClockMan]

To acmefixer

That's about the coolest thing I ever saw. Just 20 turns of 24 AWG wire around a 1/4" wooden dowel. Have a 3.3K Ohm resistor in parallel with a .1 microfared cap ahead of the transistor base. Running a blue LED from a single AA battery, and it is BRIGHT. Put my tester across the LED leads and says I'm getting 5V out of this. Is that possible? Remember I am clock maker and work with wood. This is the first time I tried this. How does this work? What kind of life can I expect out of the battery? Is there a way to adjust for best balance of brightness/battery life, like putting an (X) Ohm resistor ahead of the LED?

Thanks for all your help, very cheap and simple. Exactly what I was looking for. Now I have a starting point to try other options.
ClockMan

 

[ClockMan]

Sorry, read my meter wrong. Getting 1.8V across the LED. Still don't understand why this works. ???

 

[acmefixer]

To acmefixer

That's about the coolest thing I ever saw. Just 20 turns of 24 AWG wire around a 1/4" wooden dowel. Have a 3.3K Ohm resistor in parallel with a .1 microfared cap ahead of the transistor base. Running a blue LED from a single AA battery, and it is BRIGHT. Put my tester across the LED leads and says I'm getting 5V out of this. Is that possible? Remember I am clock maker and work with wood. This is the first time I tried this. How does this work? What kind of life can I expect out of the battery? Is there a way to adjust for best balance of brightness/battery life, like putting an (X) Ohm resistor ahead of the LED?

Thanks for all your help, very cheap and simple. Exactly what I was looking for. Now I have a starting point to try other options.
ClockMan

I wound 20 turns of 24 AWG enameled around a 1/4" wooden dowel and measured the inductance, and got 1.46 microhenry. That's really low for a JT, the frequency is probably in the AM broadcast band. The JT will work, but the capacitance of the LED starts to cause losses and it won't put out as much light as it would if the frequency was kept down below the AM broadcast band. Try to keep the length of wire at 10 to 20 feet, 15 or 16 feet will give you around 30 microhenrys. The freq should be somewhere around 250 to 450 kHz.

What transistor are you using? A good choice is the 2N4401 from Rat Shack or a lot cheaper on ebay. Another choice is the BC337-25. The 3.3k base bias resistor will determine how much current it will draw from the two AAA cells. Measure the current from the battery. You should use the resistor that will give you about 25 mA for a battery life of maybe 40 hours. I think 3.3k may be too low.

Another thing I highly recommend is a bypass capacitor across the battery leads, 10 uF electrolytic is a good start. I have had mixed results putting a capacitor across the resistor. I've found that it sometimes makes the circuit draw more current but not give a brighter LED. And it greatly depends on the size of the capacitor. You may find that for that small a coil, a .1 uf capacitor may be too low. I would try much smaller caps, .01 uF and below. I would guess that things might be better if the cap was .001 uF or 1 nF or 1000 pF. they are all the same.

Putting resistors anywhere in the circuit will only waste power in the resistor. You really don't want to waste power. Just keep it simple, and use parts you can easily obtain. I personally don't use air cire (or wood core - same thing) coils because they take up a lot more room. Small toroids are easy to wind and give a lot more inductance with a shorter length wire. The way it works is like the ignition coil in an old car, with a set of points that open and close. Every time the transistor turns off, the coil makes a spike of voltage. That 'boost' gives enough voltage to power the LED. That's an overly simple explanation. Check out some of the JTs I've built on my blog. Have fun.

 

[audioguru]

You will not accurately measure the voltage across the LED because it is pulsing at a high frequency. Your meter will measure the average voltage which might be half or less of the actual voltage.

 

[Hero999]

I've never seen an air coil Joule Theif before.

Audioguru is right.

You need an oscilloscope to measure the voltage accurately.