Attached is a circuit from the National data sheet. I want to drive a super-flux LED (well, 2 or 3) with a 317 wired for constant current. The data sheet was a little light on material, so hoping to see if I'm on the right track here.

The LEDs are 2.2 volts 70mA Max, amber (if anybody cares), power from NiCd battery pack, 4.8v, but can add more if needed.

They gave an equation for output current as: Iout=Vref/R1
I took Vref to mean 1.25 volts, from the beginning of the data sheet.
.07=1.25/R1 R1= about 18 ohms

Now, does mean I'll get 4.8v @ 70mA max output?
Since 2 LEDs in series will use 4.4 volts, will the remaining .4 volts be a concern?
Or do I loose 1.25 volts in the LM317, and only get 3.55 volts out?

Attached Thumbnails

 

 

The 317 won't give you 4.8V out if you put in 4.8V, it will drop a certain amount of voltage across the device - check the datasheet. As far as the series voltages of the LEDs are concerned, I'm not sure I understand why you are concerned aout the extra 0.4V. A constant current source's voltage is variable, but the current is constant.

__________________
"Everything that is done in the world is done by hope." -Martin Luther
"There are two ways to live your life. One is as though nothing is a miracle. The other is as though everything is a miracle."-Albert Einstein

 

If you want 4v out, give 7v in ... of course, the actual numbers vary a bit on things like current draw, but just off the cuff, I see the lm317 as "costing" 3v to operate ... since your source and load are very close in voltage, I don't see much need for an active regulator ... especially since batteries aren't known to suddenly change? a simple series resistor can eat those extra 400mV as heat, and will be more efficinet than the 317 in that case.

__________________
If you don't have a planet, what good are gold bars?

want to contact me directly? gmail gordonthree
check out my project website: http://projects.dimension-x.net
Favorite numbers:
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0

 

Yeah, I don't understand it either, that's why I posted it. I've burnt up a bunch of these LEDs using the standard series resistor method of limiting current. Not sure if its just cheap LEDs, or they need something different. These are the square, 4 lead super-flux LEDs. I've got some running on PWM (555 circuit), and get good brightness, and haven't smoked them yet. Just a lot of circuitry for driving 4 LEDs.

Best I can figure, the 317 uses 1.25 volts, the reference.

Oh, almost forgot... Should I expect needing a heatsink? I've gotten burned before, really.

 

The datasheet for the LM317 says its input voltage needs to be 2V more than its output voltage at your low current.
Its 18 ohm resistor needs an additional 1.25V.
Your LEDs might be 2.2V at 20mA or who knows what voltage at their max current of 70mA. Maybe 3V each.
You want to operate them at their max? Do they have proper cooling?

Therefore the input to the LM317 must be a minimum of 9.25V. The circuit is constant current so additional input voltage won't change the current, it will just create more heat in the LM317.

The heat is the voltage across it times the current in it. The datasheet says that the TO-220 package will get too hot with a dissipation of only 2W without a heatsink. If your input voltage is 9.6V then the LEDs will have about 6V across them, the 18 ohms will have 1.25V across it and the LM317 will have the remainder of 2.35V. At 70mA it will dissipate only 165mW of heat.

__________________
Uncle $crooge

 

I wouldn't worry about the constant current source, I'd just use a resistor to drop the 0.4V. This in't very reliable though as the LEDs Vf can vary significantly and it will dim pretty quickly as the battery voltage drops (though this isn't going to bt too bad with Nicads/NiMH batteries as the voltage stays the same through most of the discharge). Your best bet is to use a flyback transfromer switching regulator, this can give output voltages both higher and lower than the input voltage and can be converted into a constant current source, there also are ICs around that can do this.

__________________

I do not answer private messages asking for help because no one else can: benefit from advice I may give or correct me if I'm wrong.

Please ask on the open forum if you have a question and I'll be happy to help, if I know the answer.

 

Okay, I think I have enough to build something to test. Will go with the 6v Nicd battery pack (should charge okay from a 9v solar panel), and drive 2 LEDs.

I chose the LM317 mostly because I have atleast 20 of them sitting around, and have wondered about the constant current thing with driving LEDs. The driver modules they sell for the 3 watt LEDs are kind of expensive, and these LEDs aren't even close. If this works out okay, I'll make some 12 volt light strips.

I would like to learn more about buck/boost converters sometime, but it's a little more than I want to take on right now.

 

Check the data sheet with your LED's. 70 mA for a normal LED is a lot.
Normally 20 mA DC continuous.
You may possibly peak it to 70 mA as long there is a off period to cool the LED chip down.

If you drive it in a ratio 1 ON to 5 OFF at 50 Hz or more you probably get away with it.

FACTS :
2.2 Volts per LED each, so two in series is 4.4 volts.
4.8 Volts battery.
0.4 Volts at 20 mA to dissipate.

from U = I * R
0.4 = 0.02 * 20 therefore R = 20 Ohms.

from P = I²R then 0.02² * 20 = 8 miliWatt.

A 22 Ohms ¼ Watt resistor will suffice and the LED runs at 18 mA.

__________________
There are more ways to get to Rome.

Electricity, Electric clocks, Meters and Trains are great.

Please don't sent me private messages, I will not answer them.
The questions asked can be discussed in the open forums.
http://www.youtube.com/user/RODALCO2007 some interesting electrical stuff to watch.

 

Note that, that method will give you 91.2% efficiency which is better than most switching regulator ICs.

__________________

I do not answer private messages asking for help because no one else can: benefit from advice I may give or correct me if I'm wrong.

Please ask on the open forum if you have a question and I'll be happy to help, if I know the answer.

 

Rodalco beat me to it. The best way to use the 4.8 battery is to do what he said. The 22 ohm number is a standard value however, you could also use a trim pot to coax the max brightness out of your LEDs. I suspect you would be hard pressed to see much difference between 18 and 20 mA, though. The higher the resistor, value the longer the life of your battery.

If you use a constant current supply, I believe you don't need a resistor but considering the application, the single resistor is a far better solution - cheaper and better battery life. Even the switch mode solution would be hard pressed to beat the single resistor solution with a 4.8V battery (though not so for a 6V one).

 

Nobody guarantees the voltage of an LED. The voltage is a range of voltages because they can't make LEDs exactly the same. The voltage could be anywhere from about 2.0V to about 2.6V for a green LED. The voltage range is on the datasheet, but frequently it lists only the expected max voltage which is only for a few of them.

A 4-cell Ni-Cad or Ni-MH battery is not 4.8V. It is 5.6V right from the charger, 5.0V to 5.2V with only a small load current and 4.0V when it needs charging.

The range of voltages is so great that a higher voltage should be used across the current-limiting resistor to even it out.

__________________
Uncle $crooge

 

These LEDs aren't the normal 5mm bullet shaped parts!

They are the square, 4 lead, super-flux, Pirahna/Spider types.

Using a series resistor, I have a high failure rate, even at 40 mA. I did a light strip PCB for them, so heatsink for the LEDs should be okay (left a lot of copper for them). I'm experimenting with fewer LEDs right now, and just using a wire jumper to bypass the unused holes.

These types of LEDs come with the warning of being more sensitive, so I'm exploring other options beyond the typical series resistor. The LM317... I saw the circuit in the data sheet, have plenty of them on hand, and constant current supplies for higher brightness LEDs seems a must.

Just seems like a simplistic current controller, a little more effective than the series resistor. If the LEDs continue to burn, will try something else. Would like to get these cheaper LEDs under control before moving up to the more expensive 1 watt+ LEDs ($2.00 -$15.00 a pop). Paid 10 cent each for these, so not too upset over all the smoke, its a learning thing.

 

if the leds are burning out with a properly calculated series resistor, it is a thermal design flaw, and a constant current regulator will not help much.

__________________
If you don't have a planet, what good are gold bars?

want to contact me directly? gmail gordonthree
check out my project website: http://projects.dimension-x.net
Favorite numbers:
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0

 

Thanks, could have sworn NiCds were 1.2 volts. Will make the adjustments. The batteries still have a couple hours left on the charger.

 

The nominal voltage is 1.2V, check my discharge graphs at http://www.lpilsley.co.uk/viewbatt.htm to see what they actually perform like! - they are from two AA's in series feeding a small torch bulb, which is a fairly representative use.

__________________
PIC programmer software, and PIC Tutorials at:
http://www.winpicprog.co.uk