dc-dc converter design for LED driver

[Mery96]

Hello everyone, I'm working on led driver design. Circuit sizes should be as small as possible. I will work at a switching frequency of about 2 MHz. My LED specs are 2.82-3.15 V forward voltage at 1050 mA current, maximum power 10 watts. my input voltage is not clear yet but I guess I will use a li-ion battery. I am researching a dc-dc converter circuit topology where I can operate at high efficiency, high frequency and small dimensions. my consultant tells me to use an LLC converter but I'm not sure it makes sense. This will be my first application in power electronics, Your advice is very precious to me.

 

[Diver300]

Firstly, if your LED spec is 2.85 - 3.15 V at 1050 mA, the power is 3.3075 W. The maximum power may be only allowed for short durations, and the voltage will be a bit larger.

Can you share the LED specifications?

The LED will generate quite a lot of heat. The majority of the electrical power will end up as heat which has to be dissipated somehow. LEDs are a bit more efficient at lower powers, so it may be better to have more LEDs running at lower power, which will mean less overall power and heat to be dissipated.

An LLC converter is complicated, and while it may be slightly more efficient, it it probably much more difficult to design. I suggest that you have either a buck converter to reduce the voltage from the battery to the LED voltage, or you have two or more LEDs in series and have a boost converter to increase the voltage to the LEDs.

It's also worth considering a linear regulator. That will be far easier to design. If the LED voltage is 3 V, and you have a Li-Ion battery starting at 4.1 V and finishing at 3.4 V, the average voltage is 3.75 V, so the average efficiency will be 80%. I know that's not a really good efficiency, but a single 18650 battery will run the LED at 1 A for over 3 hours, and a really efficient converter would push that out to maybe 3 1/2 hours, so I don't quite see what you will usefully gain from a few percent in efficiency.

Also what is the reason for the high frequency requirement. The inductor will be physically smaller if the frequency is higher, but you will tend to get less efficiency.

 

[Mery96]

It's a nice analysis, thank you. I will use Cree Xlamp XP-L as LED. The area where the LED will be integrated is very narrow, so I have to use a single LED. The driver should be as small as possible, with little ripple and noise at the output, so I chose a switching frequency above 1MHz. Since I will use a Li-ion battery, I think it makes sense to use a buck boost or sepic converter, depending on the charge discharge voltages of the battery. The long burning time of the LED is important to me, so efficiency becomes important, even for half an hour.

 

[ronsimpson]

You have a " 2.82-3.15 V " LED so it is easy if your power is above 3.5 or below 2.8. (buck or boost) If you have to have the batter voltage about the same as the LED voltage then you need a supply that boosts & bucks. A little more complicated but not bad. That battery voltage to LED voltage decision deeds to be made early.

I had ti.com open so here is a example from TI. R1 is set for 1.5A but you can change it to get 1A or what ever you want. This is what you will find inside a high quality flashlight.

Another good place to look is diodes.com. I have used this part at 1A and just under 1A. The IC, coil and LED all will fit inside a flashlight bulb for size.

 

[Mery96]

Thank you, I'll check ti.com and diodes.com again.

 

[ronsimpson]

"LED driver" or "CC" power (constant current)
Analog Devices (the old Linear), many companies have LED power.

If you are looking for a "boost" power supply, then often are measured differently. In this example the LEDs see 100mA. The internal transistor sees many time that current. So a 700mA rating on the IC does not get you that much current. While you are boosting the voltage the current is dropping.