700mA LED driver, LM317 or FET ?

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Hello,

I want to make a constant current LED driver circuit that runs at 700mA with an input voltage of 12V - 13.5V

Is it alright to use the LM317 to build this? (got this from this thread http://forum.allaboutcircuits.com/showthread.php?t=57820)



That would require the value of R1 to be 1.78 Ohms. The LM317 also needs a 3V drop to run.

I found this other circuit (on Instructables) using a FET and a NPN BJT that looked pretty good and supposedly only needs a 0.5V drop to run.


Any input on which would be more efficient and require a smaller heat sink? Thanks.
 

Wendy

Joined Mar 24, 2008
23,415
Anything linear (which both of your designs are) is going to get hot, probably very hot. Look for a switching type regulator (often referred to as a buck puck). Where a linear regulator is a variable resistor, a switcher converts. So if you have 700ma with a Vf of 3.6 volts is around 2.5 watts. A linear regulator using 12V is going to use 8.4 watts, while a SMPS is going to user somewhere around 3.0 watts. Anything over the power used by the LED is going to be waste heat.

There have been several thread on AAC where a LED driver have been made. I have one kicking around waiting to be tested.
 

SgtWookie

Joined Jul 17, 2007
22,230
Actually, a halfway decent buck-type constant current regulator should be at least 80% efficient; many of them are exceeding 90% nowadays.

Switching regulators are initially more complex and expensive to build than linear regulators or simple resistors, but they will pay off in the long run due to far less energy being wasted.
 

Thread Starter

russpatterson

Joined Feb 1, 2010
353
Thanks for the info. Makes sense. @Bill, post your design. I'd be interested to check it out. Maybe even build it and test it out. I've been using these
http://www.ledsupply.com/wired-buckpuck.php

BuckPuck from LuxDrive. They seem pretty modern and claim no additional heat sink required. They claim 95% efficiency. They also source 5V to power your board which is handy. If I could get close to the efficiency and build it for half the price it might be worth it. These things add up fast when you're buying for "n" lights.
 

#12

Joined Nov 30, 2010
18,224
If you wanted to do linear, both your circuit ideas are good. In fact, using up most of the voltage by putting LEDs in series would go a long way toward wasting less energy as heat. I like the FET idea and the quality of your drawing. Thanks for providing good information with your question.
 

Wendy

Joined Mar 24, 2008
23,415
Actually it was Wookie's design. Since it is still untested (I really need to power it up and see if it smokes) I can't guarantee it works, but the probability is high. I think Tom66 had a design in the works too. I'll email Tom and refer him to this thread (which is how it should be done), and research to find my old thread.

When I have mine I'll post it.
 

#12

Joined Nov 30, 2010
18,224
I've used that resistor from base to emitter as a current limiter for decades. That part works. The only doubt I have is that the gate of the MOSFET might need a cap to stop oscillations...or it might not.
 

#12

Joined Nov 30, 2010
18,224
I see post 163 where RB comes in with this circuit, and the cap is on the base of the bipolar transistor.
 
Top