Any form of regulation is better than no regulation at all. What some OEM manufacturers do is use a voltage regulator to set the voltage at 9 volts, which gives plenty of overhead for powering the LM7812, even when the system voltage is down to 11.7V. Current-limiting resistors use up the excess. The LM317T can be setup as both a voltage regulator or a current regulator.

If you look at the datasheets for the higher powered LEDs, including Superflux and the larger Luxeon style LEDs, the manufacturers almost always recommend using a current limiting device. Consistent brightness is acheived with current, not voltage since each LED will have a slightly different brightness at a given forward voltage. It is also for thermal management--the LED forward voltage varies with the current rating and will change with temperature. A typical 20-mA white LED has a forward voltage that varies between 2.6 and 4.2 V over the entire operating temperature range. Constant current also does a better job controlling LED color shift.

 

I think I get it now...this schematic helped me understand. He was building a battery charger, but the principle is the same:
So, if what I'm thinking is correct, if instead you used 3 of the LED's mentioned above instead of two, you could expect the output voltage measured at the adjustment terminal to increase by Vf but the current would remain constant at 62mA.

 

Exactly. You could add as many LEDs as you have voltage to supply and the current would stay at 62mA. With a fV of 3.3V each, you can only run 3 LEDs max and still have enough overhead to run the LM317, which I understand requires somewhere between 1.25V and 3V...although I think it's closer to 1.25V.
3.3 + 3.3 + 3.3 + 1.5V = 11.4 volts, and I have seen my battery at 11.8V with the engine off.

Anyways, yes, you've got the right idea.

 

The 317 has a dropout voltage of 3v between Vo and Vi, so coupled with the 1.25v across Rg you are talking about 4.5v of voltage required across the current source. There are other voltage regulators with lower dropouts, but since our input voltage ranges from ~10-14v they still don't quite allow you to run 3 leds per chain.

This is partyly what leads to the dismal efficiency of this circuit (~50%), but since it only draws an amp or two it's not a big deal.

- Matt

 

I am wondering if this varies with voltage. With no resistor on Vadj and a 9 volt battery connected to Vin, I measured 8.78V at Vin and 7.51V at Vout, almost exactly 1.25V.

 

There is a graph of dropout voltage vs. temperature for various load currents in the datasheets. With Iout=20mA and certain temps a 1.25v dropout is possible.

From the graph it looks like at 50C and .35A the dropout is roughly 1.75. The 3v dropout is their maximum or worse case rating.

http://www.national.com/pf/LM/LM317.html

The LM1086 has a 1.3v rated dropout which might be interesting, but 3 Luxeons in series still require a forward voltage of about 10.5v, so won't work if we assume a minimum 11v power source, bummer.

- Matt

 

What if you wanted to run 4 LED's off one LM317T? could you do it like the the picture below but by changing the value of the resistor R to:

1.25/R = 2 * .062A

R=10 ohms

 

No, you had it right a couple of posts up. A 20 ohm resistor will keep the circuit at 62 mA as long as you have enough voltage to drive the LEDs. You won't be able to four Superflux LEDs at 3.3V a peice because they would pull 13.2 volts (3.3 x 4) before you figure in the voltage drop of the LM317T itself.

 

But in the circuit I posted, you have 2 LED's in series, then in parallel. Since the sets of 2 led's are in parallel, and if I wanted 62mA through each, I would want 124mA total, which would require a 10 ohm resistor, correct? Since the voltage drop across one set of 2 led's is only 3.3x2, you're still ok.

 

Sorry, the diagram you posted did not show up at the time I responded. With 2 arrays, yes, a 10 ohm resistor will give you an output current of around 124 mA, divided into 2 arrays = 62 mA per array.

It is not recommended to use series-parallel arrays behind a constant-current source. This is taken straight from the data sheet:
"When controlling multiple LEDs from a single constant current source, the LEDs should be connected in series to ensure the current in each LED is identical. If LEDs are connected in parallel the varation in Vf will again lead to different currents and therefore different brightness in each branch." In addition, if one LED burns out, the full 124 mA will go to the other array and all of the LEDs will eventually fail. This is why I used one LM317 per array, and at 85 cents a piece, it seemed worthwhile.

I will refer you here for a full explanation.

Ken

 

Cool idea something else to add to the to do list...

 

I've been wanting to put lighting in the foot wells for awhile now but don't really know how i'm going to wire everything up. This looks really good and definitely bright!

 

if ur not sure how to work with leds that u have to solder urself, get some led bars or get some cathode light bars (not as efficient but work) and wire them up to the dome light. pretty simple if u have the soldering part done for ya...