Best way to power A white LED

[Overclocked]

What is the best way to power a LED under the following conditions?

a) Where Battery Life is of Most concern and LEDs are at moderate brightness
b) Where battery life is not of concern (or infinite) and you want the maximum brightness from a LED.
c) Where Battery life is of concern but you want the Max Britghtness from the LED's while conversing battery life.

Or, Which is better for powering White LEDs under condition C.

1) PWM
2) Just Using Resistors
3) Boost Converter
4) Constant Current Source

I want to re do my flashlight, but I want it to be efficient yet get the most brightness. This time instead of a 9V, I am going to use 6 AAA Batteries in series (Some where Ive read AAA's have a capacity of 2200mAh) to get 9V. Or just use two 9v in parallel to get more current.

 

[HarveyH42]

Try this...

http://www.joulethief.com/kit.php

Gives schematic and full details. Haven't built one, but thinking of doing something with it...

 

[Hero999]

The switching regulator in the link below features current limiting. You could easilly omit the zener and just use it as a constant current source set at 350mA for a lux star LED.
https://www.romanblack.com/smps/a04.htm

That's how I would do it anyway, you're better off with six AAA batteries than two 9Vs in parallel.

Assuming 80% efficiency (worst case) the current from the battery will be 213mA, enough to keep the LED powered for over 10 hours.

In practice if the circuit is well designed the efficiency can exceed 90% so the batteries should last even longer.

 

[mneary]

In approximately the same volume, you're probably a lot better off with four AA batteries instead of six AAAs.

You may find still better energy density in a single D cell and the joule thief. (If my reading of the Duracell data sheets are correct, about three times as much).

I haven't seen a 2200mAH AAA, myself. Lots of AAs, though.

 

[mneary]

a) Where Battery Life is of Most concern and LEDs are at moderate brightness
--------adjust joule thief (boost converter) to moderate brightness.
b) Where battery life is not of concern (or infinite) and you want the maximum brightness from a LED.
--------adjust joule thief for maximum rated LED current.
c) Where Battery life is of concern but you want the Max Britghtness from the LED's while conversing battery life.
--------adjust joule thief for a little less than maximum rated LED current.

If you use a battery of 4V or more, substitute buck converter for boost converter. If you have several LEDs in series, the boost converter is almost always used.

 

[Overclocked]

I was going to connect the LEDs in series, and I may use a specialized Controller or just design a SMPS Boost converter from Scratch. Personally, Id rather sit down and design something, But thanks for the Idea's.

Ive got about 100 LEDs, But I think I might use something like 20 to 25 LEDs. I have seen the S-Flux (or piranaha) LEDs for cheap, they have a wider angle, but are physically bigger. Would S-Flux LED's be better suited for a flashlight?
(S-flux LED's are these things:**broken link removed**)

 

[audioguru]

The voltage from alkaline batteries drops like a rock and so does the brightness of a simple flashlight.

The voltage from a rechargable Ni-MH battery begins lower than alkaline but the voltage stays much higher longer. Then you can charge it and use it again many times almost for free.

 

[mneary]

The voltage from alkaline batteries drops like a rock and so does the brightness of a simple flashlight.

Agreed. I should have mentioned that.... you need a regulator to drive an LED from alkaline. Since the LED relies on current rather than voltage as the prime driver for ratings and for brightness, it is ideally a current regulator.

NiMH has a flatter voltage curve, which is also a solution, with careful design (resistance choice).

 

[Hero999]

I was assuming he was talking about using NiMH batteries and yes I don't think 2200mAh NiMH cells exist, I think he was confusing them with AA cells.

I was going to connect the LEDs in series, and I may use a specialized Controller or just design a SMPS Boost converter from Scratch. Personally, Id rather sit down and design something, But thanks for the Idea's.

Ive got about 100 LEDs, But I think I might use something like 20 to 25 LEDs. I have seen the S-Flux (or piranaha) LEDs for cheap, they have a wider angle, but are physically bigger. Would S-Flux LED's be better suited for a flashlight?
(S-flux LED's are these things:**broken link removed**)

It makes no difference whether they're in series or parallel.

The advantage with parallel is you'll probably be able to build a higher efficiency converter and it'll probably be easyier to do too. The disadvantage is that you'll need a resistor for each LED which is no problem as resistors are hardly expensive.

For example if you use the simple constant current regulator from my previous post to power 10 white S-flux LEDs at 27mA each you put a 22hm: resistor in series with each LED and just use one of them as the current sensing resistor.

Conneting them in series can also be done but you'll either need a buck boost converter (the output voltage will be negitive rather than positive) or a fly back converter and just make sure that the battery voltage will never exceed the voltage drop of the LEDs (they'll overcurrent) and that the output of your converter can never become open circuit as (the components on the output o fhte converter will be subjected to a high voltage).

 

[Overclocked]

Thanks, If I understand correctly, either a Constant current source or a Boost converter is used to power LEDs.

Im still trying to decide if I should use 5mm LEDs, S-Flux LED's or Luxeon Star LEDs. 5mm's are small (when compared to the other 2) and still give a pretty good brightness. S-Flux LED's are a bit larger but have a wider angle. Luxeon Star LEDs are extremely bright, and have the same angle as a 5mm LED but requires the power of about 10 LED's (assuming Each LED gets 20mA each).

But the wider the angle, the bigger area you can see. If 5mm and S-Flux LEDs are compared against each other, they still have the same brightness and require the same current, but S-Flux (as mentioned before) have a wider angle. For this application, it seems S-Flux LED's are ideal. If I wanted a large amount of light in a concentrated Space (or if I wanted to light my room..), I would go for Luxeon Star LED's.

So S-Flux it is then!

 

[Hero999]

Here's a simple constant current LED boost converter. I modified it from a simple boost switching regulator. Efficiency is typically around 60% to 80% but you can probably make it even better if you know what you're doing.

R1 = 10k but it needs to be less than that for higher power levels.
R2 = Current sense resistor.
C1 = 4.7nF
C2 = 100:mu:F
D1 = 1N4001 but a schottky barrier is better.
Most small NPN transistors will work for Tr1 and Tr2, 2N2222A, BC549 etc. but check the power rating.

The problem is the voltage drop of the diodes needs to be more than Vin or else the current from the power supply will flow though D1 and destroy the LEDs.

 

[mvs sarma]

i have seen aaa nimh in 1100mAH capacities. mostly either varta or camlean

 

[Overclocked]

Here's a simple constant current LED boost converter. I modified it from a simple boost switching regulator. Efficiency is typically around 60% to 80% but you can probably make it even better if you know what you're doing.

R1 = 10k but it needs to be less than that for higher power levels.
R2 = Current sense resistor.
C1 = 4.7nF
C2 = 100:mu:F
D1 = 1N4001 but a schottky barrier is better.
Most small NPN transistors will work for Tr1 and Tr2, 2N2222A, BC549 etc. but check the power rating.

The problem is the voltage drop of the diodes needs to be more than Vin or else the current from the power supply will flow though D1 and destroy the LEDs.

But how much voltage do you actually need on the output? If I run them all in series, Im assuming # of LED's * Vled would give how much I need. If each LED drops a voltage of 3.5V and I have 15 LEDs (since I dont know how much space s-flux uses, I'll assume 15 can fit on a round PCB board) that would be 52.5V! Of course, If I connected 3 in series and then connected 4 banks of 3 in parallel I would only need 10.5V.

So, I'll go with a previous suggestion for using D sized batteries (These are just design considerations- I might not use Alkaline Batteries, but rather NiMh. But for Assumptions sake, I'll ne referencing to Alkaline's) and looking at the data sheet for Energizer Batteries, They have a capacity of 20500mAh!! (which is 20.5Ah- Correct?) If a boost converter consumed even 1A, it would last 20hrs, @.5A 40hrs,@.25A 80hrs.

If I used that assumption (assuming each LED requried 30mA each), I would need 90mA per branch, which if there are 5 branches total, I would need 450mA total, in which .5A would be sufficient. So it would last 40hrs on 2 D batteries.

 

[Hero999]

The output voltage is simply the forward voltage multiplied by the number of LEDs. This circuit is most efficient when the output voltage isn't much higher than the input voltage but you want to ensure that the output is never below the input so you do need a reasonable margin for error. Yes, four banks of three is probably ideal, and six cells will give 7.2V which would be perfect.

The current supplied by the battery isn't the same as the current used by the LEDs if you're powering them from a switching regulator. If you're using a step-up converter then the battery will always supply a larger current than the LEDs are using and if you're using a step-down converter the batteries will always supply a smaller current than the LEDs are using.

For example:
[latex]V_{IN} = 12V[/latex]
[latex]V_F = 7V[/latex]'Total forward voltage for all of the LEDs in series.
[latex]I_F = 0.3A[/latex]'Total forward current for all of the LEDs in parallel.
[latex]P_{OUT} = 7 \times 0.3=2.1W[/latex]
[latex]P_{IN} = \frac{2.1}{0.8}= 2.625W[/latex]'Assuming 80% efficiency.
[latex]I_{IN} = \frac{2.625}{12}=0.21875A[/latex]

If you're going to use a NiMH D cell then make sure it's a real D cell, lots of NiMH D cells are AA cells in disguise.

 

[Overclocked]

The output voltage is simply the forward voltage multiplied by the number of LEDs. This circuit is most efficient when the output voltage isn't much higher than the input voltage but you want to ensure that the output is never below the input so you do need a reasonable margin for error. Yes, four banks of three is probably ideal, and six cells will give 7.2V which would be perfect.

The current supplied by the battery isn't the same as the current used by the LEDs if you're powering them from a switching regulator. If you're using a step-up converter then the battery will always supply a larger current than the LEDs are using and if you're using a step-down converter the batteries will always supply a smaller current than the LEDs are using.

For example:
[latex]V_{IN} = 12V[/latex]
[latex]V_F = 7V[/latex]'Total forward voltage for all of the LEDs in series.
[latex]I_F = 0.3A[/latex]'Total forward current for all of the LEDs in parallel.
[latex]P_{OUT} = 7 \times 0.3=2.1W[/latex]
[latex]P_{IN} = \frac{2.1}{0.8}= 2.625W[/latex]'Assuming 80% efficiency.
[latex]I_{IN} = \frac{2.625}{12}=0.21875A[/latex]

If you're going to use a NiMH D cell then make sure it's a real D cell, lots of NiMH D cells are AA cells in disguise.

Dont you mean 10.5V? Assuming 3.5V for each LED, and 3 LEDs in each bank, thats 10.5V @30mA per bank.

 

[audioguru]

Energizer's alkaline D cell is 20,500mAh if the current is only 25mA and if the end voltage is only 0.8V. The voltage keeps dropping and dropping and dropping and dropping and dropping and dropping.
With a 600mA load and lots of breaks to rest, it is only 9600mAh down to 1.0v.

 

[Overclocked]

Energizer's alkaline D cell is 20,500mAh if the current is only 25mA and if the end voltage is only 0.8V. The voltage keeps dropping and dropping and dropping and dropping and dropping and dropping.
With a 600mA load and lots of breaks to rest, it is only 9600mAh down to 1.0v.

As far as I'm concerned, flashlights are used for emergencys (or to see). Power Outages only last for a hour at max, So this should last a long time.

Im having a hard time finding a controller, but I have found one:https://www.mouser.com/search/Produ...virtualkey52230000virtualkey522-ZXLD1101ET5TA

Not exactly a boost converter, rather a constant current source with a limit. According to the datasheet, only one resistor needs to be changed to change LED current. The data sheet shows up to 6 can be connected in series with each other, but I'll only be hooking up 3 in series, with 5 banks in parallel. Max Output Current in .5A, which is more than enough.

I wanted to design something from scratch to prove to myself Ive moved up to a next level and can design a SMPS from scratch, but I guess this is OK too... The only other thing that worries me is what if this isnt bright enough for a flashlight application? And Technically, I dont need SMD...Id rather have the challenge of Through hole

ADD: I found the same one but with a lower Supply Voltage
https://www.mouser.com/search/Produ...TAvirtualkey52230000virtualkey522-ZXSC400E6TA

 

[Hero999]

Dont you mean 10.5V? Assuming 3.5V for each LED, and 3 LEDs in each bank, thats 10.5V @30mA per bank.

I was just giving an example of how to calculate the input current to a switching regulator. Try changing the forward voltage to 14V and the output current to 0.15A but keeping the efficiency 80% in my example and the input power and current and output power will alll stay the same. Now try calcultating it for your LEDs, batteries and the switching regulator of your choice.

Im having a hard time finding a controller, but I have found one:https://www.mouser.com/search/Product...-ZXLD1101ET5TA

Not exactly a boost converter, rather a constant current source with a limit. According to the datasheet, only one resistor needs to be changed to change LED current. The data sheet shows up to 6 can be connected in series with each other, but I'll only be hooking up 3 in series, with 5 banks in parallel. Max Output Current in .5A, which is more than enough.

I wanted to design something from scratch to prove to myself Ive moved up to a next level and can design a SMPS from scratch, but I guess this is OK too... The only other thing that worries me is what if this isnt bright enough for a flashlight application? And Technically, I dont need SMD...Id rather have the challenge of Through hole

ADD: I found the same one but with a lower Supply Voltage
https://www.mouser.com/search/Product...22-ZXSC400E6TA

The second controller is probably better as it gives you a greater supply voltage range. Always be ware that freshly charged NiMH cells can supply as much as 1.6V each so bear this in mind when considering the number of cells and the maximum voltage of your circuit.

If you still feel like growing your own, it's pretty easy to convert a standard switching regulator IC inot a constant current source; there's a good EDN article on how to do this and it should work for a variety of different switching regulator controllers.
**broken link removed**

 

[Overclocked]

I find it strange that I cannot find the calculation to find Lmin (the minumum inductance needed). The book I have only gives the formula for Buck converters, and the data sheet doesnt give the formula either.

Does anyone know the formula to find Lmin for boost converters? The one given for buck just uses Vin, Vout and Switching current. I thought the buck formula would work in reverse, but it doesnt.

 

[Hero999]

I would assume is the same as a boost converter, I can't think of the formula of the top of my head though. Doesn't the datasheet suggest a specific inductor?