Driving 1-3W LEDs with lm3406

I would like to preface this post with stating that I am pretty new to the world of electronics. Therefore these may be stupid questions.

So I'm making an aquarium light with several 1-3W LEDs. Specifically I'm using Cree XT-E royal blues, XR-E Reds, XR-E greens, and XBD R4 cool whites.

The count is probably going to be something like 4 red, 4 green, 8 blue, and 4 white. I would also need to be able to dim each color individually.

So far my plan is to drive them with an lm3406 with the LEDs in series strings of 4, which would mean I would need 5 lm3406 ics. The lm3406 from what I understand would allow me to dim the LEDs hooked up to it with a pwm signal from the arduino; the voltage would be modified by the pwm signal and the current would be fixed. If the psu is at 12 volts, and the string doesn't dissipate the full 12 volts, would I need to add a resistor in line with the LEDs, or would throttling it down with pwm be enough protection?

Also, do I need any extra hardware to drive the LEDs with the lm3406? There's a circuit diagram in the datasheet that calls for a number of other components.

The LM3406 (please use capital letters) is a switch mode current regulator. The PWM signal turns it on and off.

No, do NOT add any resistors in series. Yes, you must use the other components in the circuit diagram.

Ok so I do need all the other components listed... Well... that drives up costs and complexity a good deal. The datasheet lists the part numbers, but I've noticed a lot of them aren't stocked or made anymore.

Are there complete boards of the circuit available at a reasonable price?

If not are there any other more cost effective solutions?

00101:
The datasheet lists the part numbers, but I've noticed a lot of them aren't stocked or made anymore.

Application notes for ICs refer to the "reference design" or the development boards they may have made to demonstrate the device when it was introduced.

Most electronic parts - ICs being the exception - are specified by characteristic, not a "part number" - a resistor has a resistance and a power dissipation, a capacitor a capacitance and working voltage (and a dielectric type specification) and so on.

Search on Google (or eBay!) for "dimmable LED driver", possibly using LM3406 as a search term but more likely including "3W LED".

This particular one did list part numbers, but as I said most are no longer stocked. However, I did find equivalents to everything to the tune of about $6 (including the LM3406) per chip.

Would I need to heatsink the LM3406's if they get an input voltage of say 12, but are only outputting like 4.5? Or does the switching regulation make it efficient enough that there won't be substantial heat?

To determine whether you will need a heatsink, calculate the temperature rise as shown on page 27 of the datasheet. If you get the right components, then it is unlikely that you will need a heatsink using those LEDs.

Pay special attention to selecting a suitable inductor, as shown on pages 24-25.

Hi there,
I wanted to have an LED illumination on my aquarium as well, but I wanted it low cost and simple. It works for 6 months now, so I think it is not all that bad.
I use 9 x warm white 3W, 6 of which with lenses, and 6 x 1W royal blue, all OSRAM Dragons. The simplicity is that there are always 3 LEDs in a row. That allows to use 12V almost without problems, especially since one can use PWM and never give 100% (they tolerate spikes up to 2A!). Anyway, I put a big 12 Ohms resistor in the white lines (actually, I am not sure about the value and cannot have a look right now, but there is a resistor!). The blue ones are fine without, since they would need 12.6V and are bright enough by far. The PWM pins switch the LEDs via MOSFETs.
For cooling the LEDs I mounted them on aluminum blocks and those onto a stainless steel tube. An extra pump starts after the sum of all PWM channels exceeds a value (because, if you already have PWM, you HAVE to have dawn and dusk, right?). So my LEDs are pretty much only a tiny weebit above room temperature (I monitor temperatures before and after the cooling tube and room temperature).
So if you don't want it fancy it is possible the simple way. But if you learned how to read a datasheet it was worth it already :smiley: Good luck!

So I'm about 95% sure that I'm going to be using the LM3406. I've been off and on reading the datasheet for the last week. Right now I'm going through design example 1 and following the instructions on how to calculate the parameters for each component. I have a few questions since reading it.

First of all, as I understand it, by changing the value of a particular resistor in the circuit, the output current is changed. My plans, since I don't know how bright I need them exactly was to put 3 resistors on the board that correspond to 350mA, 700mA, and 1500mA. I would use 3 small switches to then set the current. One question is, is that plan a good idea? Or would just driving them at 1.5A and then PWM dimming them provide enough of an output gradient? Or would there be other advantages to driving at lower currents for lower brightness? I want to have the option to drive them at capacity for adaptability of the lighting unit.

Secondly, for the calculation for RON, it calls for D, which I believe is duty cycle. Since I'd be PWM dimming them, the duty cycle would vary based on whatever brightness happens to work and it would change as they are fading on and off. What value should I use for D? Or is D not duty cycle?

The datasheet defines D at the top pf page 9. It is the switching regulator duty cycle, which depends on the ratio of input to output voltages (with allowances made for the switch and the Schottky diode), and is not related to PWM.

IMO using more than one resistor and a DIP switch is a good idea. But you can get 4 levels of current with just 2 switches, by having the switches connect extra resistors in parallel.

Hopefully, my last stupid question.

Since I want to be able to run these at different currents, and several of the calculations call for the current, should I run the numbers for each of the 3 currents I'm planning on, and then get a part that would satisfy the worst case scenario? Specifically for the inductor and capacitors.

00101:
Since I want to be able to run these at different currents, and several of the calculations call for the current, should I run the numbers for each of the 3 currents I'm planning on, and then get a part that would satisfy the worst case scenario? Specifically for the inductor and capacitors.

Yes. In particular, choose the highest inductance that the calculations give (you will probably find that lower current requires higher inductance).