The LM2576 is a 3 amp regulator. If you wanted the circuit to regulate at .64 amps then you would use a 1.87 ohm resistor instead of the variable resistor and R8. If you didn't need voltage regulation then simply remove R 5,6,7 and the switch.

No I do not need voltage regulation since it will only be powering the LED light. I will be using a LM2576 -5 for my Mcu. What would be ideal is if I can find a switcher that can give me 5V regulated out on one pin and 12v with regulated current on the other. Similar to what the current buckpuck does but I am guessing that is asking too much. I would be happy with limiting to ~640ma with 12V in and <= 12V out.

 

Most of these switchers use a current sense resistor of no more than 1/4 watt. The pt4115 does this. Pretty simple to use but I am guessing there are better chips out there now and that is why I am posting.

 

The wattage (2W) was for a previous application. At .64 amp a 1.87 ohm resistor would dissipate .77 watts. A 1 watt resistor would be fine.

It is efficient. The typical efficiency is >85%.

Most of these switchers use a current sense resistor of no more than 1/4 watt.

That is a different kind of current sense. It is sensing the saturation of the inductor, not the output current.

 

I would be happy with limiting to ~640ma with 12V in and <= 12V out.

OK, anything else??

 

OK, anything else??

Thanks for all of the help. Does TI have a designer app for the current mode? I can't even find the blasted voltage one now and I have used it a number of times before. I would just search and find it. Should have book marked it. .

 

Ah did bookmark it.

http://webench.ti.com/webench5/powe...=13&O1I=0.05&O1V=5&op_TA=30&application=POWER

Now I need to figure out if it does the current configuration. But it is late now and I still have work to do that earns me a paycheck.

 

Here's the LTspice simulation of a simple hysteretic LED current driver that uses a common LM339 or LM393 comparator as the control element, which you might want to play with.
Your board looks like it has groups of 3 LEDs in series so that's the load it has.
For the feedback values shown it regulates at about a 645mA average current with about 40mA of ripple (which would be invisible).
Replacing R3 with a 100kΩ pot would allow adjustment of the regulated current value.
The simulated efficiency was over 90%, not including any losses in the inductor.

 

Here's the LTspice simulation of a simple hysteretic LED current driver that uses a common LM339 or LM393 comparator as the control element, which you might want to play with.

Nice. Does that Si443DY require a heat sink?

Your board looks like it has groups of 3 LEDs in series so that's the load it has.

That is what I figured too. But there are 64 LEDs so it does not add up unless they have one hanging out there on it's own. And I can't figure out where I came up with 640ma. I just dragged out the old controller and looked at the sense resistor again. It is a .33 which makes that board regulate 333 ma if I am reading the datasheet and resistor correctly . And if you do the math on the array it comes to around 440ma which would be about right for original controller. I did this 6 years ago so who knows what I was thinking. Unless I intended to rebuild the light for 640 ma and bought the 700ma buckpuck to prepare and really have it set to 330 ma. Of course I don't have any notes. I guess I will have to measure it to be sure.

 

Sorry a bit late to this party. Just thought I'd give you another option..
Allegro A6211 has been used to make good LED drivers..
http://html.alldatasheet.com/html-pdf/476579/ALLEGRO/A6211/3868/13/A6211.html

DIY board w/ 8 DIY drivers:

 

Yeah thanks I am aware of those. My problem is space issue. I need to fit it in the same space as the current buckpuck. It can be a little bigger but not much.

 

Nice. Does that Si443DY require a heat sink?

That is what I figured too. But there are 64 LEDs so it does not add up unless they have one hanging out there on it's own. .......................

The simulation shows the MOSFET dissipating less than a half watt so should require no heat sink.
Note that you can use a different P-MOSFET than the Si4433DY I used but it should have a total gate charge of no more than about 10pC (as shown in its data sheet) for proper switching and minimum dissipation.

Incidentally, I count a lot more than 64 LEDs on that board.

 

Sorry 108. I got to bed about to last night and then maybe got 3 hours sleep. So that is 720ma. I was right!


I might go with the zxld1360 or the original pt4115 (I have some one hand if I can just find the sense resistor without paying 200x more for the shipping). Both are tiny and low part count. And they have PWM control which I need. Actually I just need to be able to turn it on and off.

I also found cheap flood lights on line. They have their own current regulation.

Then again I might take another look at your circuit.

 

There are no resistors. Current is regulated by the LED driver.

The driver in the talkingelectronics LED projects I mentioned a week or two back could probably be scaled for a higher voltage and bigger LED array.

The transistor has a current sensing resistor, so the bias control transistor switches it off as the current rises to a specific point. The LED is in series with the inductor, so it is driven while the current rises to the control point. When the transistor switches off; the catch diode routes the back emf through the LED, so it gets two bites at the cherry - after losses; the energy recovery can't be more than what was put into the inductor during the current controlled phase.