Hi All,

I am currently doing a project to construct an underwater scuba dive flashlight.

The flash light will be using a 10W LED
(bought it from ebay http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=290739819764 )

Power source is 18Volt PowerDrill rechargeable battery (19.5V measured)
( http://www.kmart.com/shc/s/p_10151_...c&srccode=cii_13736960&cpncode=32-173320936-2 )

my question is, whats the most efficient way to drive this LED.

basic rating for this LED:
forward voltage: 9-12V
Current: 1A

I dont need to dim the LED, i just need it to be on and off. so i am not planing to use PWM to drive it (I also don't think using a PWM will drive the LED more efficiently).

I have couple ideas in mind.

1) Simply drive it with a power resistor (I think it will be getting a big power dissipation from the resistor, because the voltage drop across it will be about 19.5V-12V= 5.5V)

2) use a DC to DC convertor converter to convert 19.5volt to 12volt and use a smaller value power resistor to drive the LED. However, I am not sure what is the typical power dissipation of a DC to DC convertor.

Any other suggestion would be greatly appreciated!

Thanks in advance!!

 

LM2576-ADJ can be used for this LED.

Either in voltage mode, ignoring the fact that many people use current-mode converters (you just have to be careful when you adjust the voltage as the gradient raises very fast);

Or in current mode, you need some additional components.

 

By the way these LEDs need a large cooling grid. Perferably fan-assisted (see VGA coolers for instance). I have one of these 10W LED chips inside my computer case, supplied with 11.3V (that results in 1.08A current).

I don't use current regulation as such. It serves me for many months now, still 100% brightness.

You may want to use brightness regulation in order to extend battery life...?

The 12V information is simply wrong, resulting in too high current (nearly 2 Amps but also I don't get exact 12V from my ATX supply, it's a little less than that I think).

 

LM2576-ADJ can be used for this LED.

Either in voltage mode, ignoring the fact that many people use current-mode converters

I designed several battery chargers using the 2576 chip for constant current usage. It would easily drive the LED to any current level up to 3A.

see pages 9, 10, 11 of app note.

http://www.ti.com/lit/an/snva557/snva557.pdf

efficiency is decent, probably 80% ballpark at 2- 3A.

Just put the LED where the battery is and the converter will source constant current into it.

 

I designed several battery chargers using the 2576 chip for constant current usage. It would easily drive the LED to any current level up to 3A.

see pages 9, 10, 11 of app note.

http://www.ti.com/lit/an/snva557/snva557.pdf

efficiency is decent, probably 80% ballpark at 2- 3A.

Just put the LED where the battery is and the converter will source constant current into it.

There they are, these additional components

 

It's an extra op-amp and some resistors, you don't have to use a variable rate design as shown in the app note. Just set it for the current you want using resistors.

But understand, if you try to drive a power LED from a voltage source it will destroy it. Just drop it in the garbage now, it won't work.

 

A PWM is the most efficient way to run one of these LED's ,especially when you are going to be running it from a battery.You will be able to cut down the duty cycle on the pwm to about 60 - 70% and not see any change in light output .

 

<sigh>

PWM will cause a visual variation of light level. This is why it is so good for LEDs, it is basically proportional. Only a resistor (or equivalent, such as a constant current source) will keep an LED from burning up. PWM or not, the resistor is mandatory.

A SMPS constant current regulator converts the voltage into whatever it needs to be to set the current, and boosts efficiency as a result. This is where the 70%+ figures are coming from, conversion is much more efficient than resistance. The cost is in complexity.

I see it over and over, people both advocating voltage and PWM as if it regulates current, it does not. Only a resistor or current regulator does this, and the latter is programed with resistors.

A tutorial I wrote on low power LEDs, most of the basics remain the same.

LEDs, 555s, Flashers, and Light Chasers

 

By the way these LEDs need a large cooling grid. Perferably fan-assisted (see VGA coolers for instance). I have one of these 10W LED chips inside my computer case, supplied with 11.3V (that results in 1.08A current).

I will be using this underwater. The heatsink will be water-cooling.

 

A PWM is the most efficient way to run one of these LED's ,especially when you are going to be running it from a battery.You will be able to cut down the duty cycle on the pwm to about 60 - 70% and not see any change in light output .

If you drive the duty cycle to 70%, wouldn't the circuit dim the LED by 30%?
to my understanding, the light intensity is liner to the duty cycle.

I have only done research on this topic, but have not personally tried it myself to see the result.

 

A SMPS constant current regulator converts the voltage into whatever it needs to be to set the current, and boosts efficiency as a result. This is where the 70%+ figures are coming from, conversion is much more efficient than resistance. The cost is in complexity.

I did some research and found this current current LED driver (STCS2: http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00172746.pdf) .

Is this something i want to use to drive a 10W LED? Its rated at 2A, i will probably be only driving the LED with 1A.
I just wanted to make sure before i make the purchase.

Thanks !!!

 

It's an extra op-amp and some resistors, you don't have to use a variable rate design as shown in the app note. Just set it for the current you want using resistors.

But understand, if you try to drive a power LED from a voltage source it will destroy it. Just drop it in the garbage now, it won't work.

I don't understand at all how this should happen. I read the voltage/current graph from LED datasheet. There is no such information that current would rise uncontrollably for any voltage.

Sure if you use a dc/dc chip already, add the extra components for current control if you have a wish or need to do so.

 

I don't understand at all how this should happen. I read the voltage/current graph from LED datasheet. There is no such information that current would rise uncontrollably for any voltage.

OK, if you want to try to drive a power LED from a fixed voltage source, try it.

RIP, LED.


Look up "thermal runaway". The VBE of a diode drops as temp goes up, so driving from a fixed voltage, that increases the LED current as the VBE shrinks. That increases temp and the loop runs away, and PRESTO: smoldering LED remains.

 

<sigh>

PWM will cause a visual variation of light level. This is why it is so good for LEDs, it is basically proportional. Only a resistor (or equivalent, such as a constant current source) will keep an LED from burning up. PWM or not, the resistor is mandatory.

A SMPS constant current regulator converts the voltage into whatever it needs to be to set the current, and boosts efficiency as a result. This is where the 70%+ figures are coming from, conversion is much more efficient than resistance. The cost is in complexity.

I see it over and over, people both advocating voltage and PWM as if it regulates current, it does not. Only a resistor or current regulator does this, and the latter is programed with resistors.

A tutorial I wrote on low power LEDs, most of the basics remain the same.

LEDs, 555s, Flashers, and Light Chasers

I agree Bill the current regulation is mandatory, but pulsing the regulated current can save battery power with little or no noticable difference in intensity of light.

 

I can imagine the cooling will be problem. It will require mechnical skills. How can you transfer the heat from the backplane of the LED into the water?

And do you agree if you adjust the voltage very slowly, there will be an initial point where the LED will just turn on? Well this is reality. Or do you maintain the point there will be an initial point where current will rise apruptly and destroy the LED? This is not true actually.

I don't disagree that some people seek to have current regulation, and/or some applications will require it. I use said 10W power LED only controlling the voltage and it works for me for a number of months now.

 

What do you plan to house this all in? I hope you have an existing housing designed UW use. Pretty much anything else will leak and or crush in only several meters.

 

I agree Bill the current regulation is mandatory, but pulsing the regulated current can save battery power with little or no noticable difference in intensity of light.

You get the same exact effect just by reducing the current, LEDs are pretty nonlinear, so are eye balls. There is nothing mystical about PWM. When used with a SMPS it can be pretty cool, but that has to do with the process of conversion as opposed to limiting.

 

If you drive the duty cycle to 70%, wouldn't the circuit dim the LED by 30%?
to my understanding, the light intensity is liner to the duty cycle.

I have only done research on this topic, but have not personally tried it myself to see the result.

Assuming that 70% duty-cycle reduces the current by 70% as compared to the 100% duty-cycle current then yes, the brightness will be reduced by a roughly linear amount.

But realize that if the PWM controls voltage and the LED current is determined by a resistor in series with the LED than a 30% change in the PWM duty-cycle may cause more than a 30% change in current (as determined by ratio of the voltage drop across the resistor as compared to the drop across the LED).

 

Here are some of the image of the housing that i came up with.
I bought everything from Home-Depots (Water-pip section) for about 5 dollar.
The black housing is the water-prove part. I brought this thing down to 70 feet deep of water and no leak/damage.

The next step is to come up with a way to run copper heat pipe and wires from the housing. I have couple ideas in my mind, and will share it if i have a successful run.

 

Here are some of the image of the housing that i came up with.
I bought everything from Home-Depots (Water-pip section) for about 5 dollar.
The black housing is the water-prove part. I brought this thing down to 70 feet deep of water and no leak/damage.

The next step is to come up with a way to run copper heat pipe and wires from the housing. I have couple ideas in my mind, and will share it if i have a successful run.

It will most likely leak at pressure. Chances are the lens will crush or at least crack.


Before you spend a lot of money on the project you should first test the housing. Dive watches are rated at 200meter or more. This is because the rating is for static pressure.

But your dive light will be subject to dynamic pressure. As you move it about, the pressure will actually increase on the housing.