Reticently i was working a little Solar lighting project, i need a really bright insane ~25W LED lighting. But issue was for 25W it a lots of power for LED and it require some skills to make such a high wattage LED driver. primary objective of this project is to get a descent knowledge of MPPT charge controller and MPPT charging algorithms. this project is part of the project , that i will post later.
First we need to chose which LED we are going to use , i really wanted to stuck with good manufacturer so Philips , Osram or perhaps everlight. with experience i had gone with Philips LUXEON Reble ES series leds, i.e. LXH7-PW40 or LXML-PWC2 . this series led are very easily available on mouser, digikey and farnell (element14 here). i have already used these LEDs in a very robust and can be pushed upto max 3W each so we need really few LEDs. although 3W is achieved with 1A at 3V. at 1A current efficiency of the led is little bit low as compare to maximum efficiency which is achieved at 700mA current with gives around 2W at 2.8V.
for this particular project my main aim is to learn about different software algorithms in MPPT chargers. so LED efficiency is not really a primary concern for me , at least for now. we can fix this later.
For the drive ,we are going to chose a simple switching drive, rather than a highly efficient but complex to design fly-back converter design.as designing fly back converter is itself a task. so first requirement itself is , that the circuit should be simple.
so here are the requirements, i need 360 degree light so we need LED pointing in minimum 3 directions 120 degree each . with capability to control any one of three 120 degree light individually. they should also be PWM dimmable. To have individually control each 120 degree segment , we either need some extra circuitry with just one driver , but i have decided to just make one PCB 9W capability with 3 LED, and will fit this PCB in triangle form to get a nice 360 degree shape.
the whole system will have large li-ion battery , actually two li-ion cells connected in series, which with 3.0V to 4.1V cell give , around 6V minimum and 8.2v maximum supply range. 3 led require around 9V to drive upto maximum current of 1A with will give 9W total.so we need a really powerful boost converter.
with these requirements i searched a lot but for quite some time , could not able to find some off the self part which can drive such a high wattage at such a low voltage.
but in the end if found LT3478 from linear technology. this IC has internal 4.5A switch, accept inputs as low as 2.8V and maximum upto 36V. we can program maximum led current and can also program over voltage protection for LED. adjustable operating frequency from 200Khz to 2.25Mhz depending on load , input voltage and inductor size. A another variant of LT3478 , LT3478-1 even has on chip current shunt, so less external components.
Let Look at The schematic.
schematic is fairly simple , just immediate copy of reference circuit which linear provide in the datasheet. there is actually schematic for exactly same family of led which we are using but the circuit ment for 15W with just single driver and 6LED. we going to have only 3 LED running a little higher current and slightly different input voltage.
and as we have only 3 led then we need to change over voltage protection resistor ladder and current setting ladder also. if you want to drive led at what ever the current you want just change the resistor accordingly . or if you change LED count then it is required to change over voltage protection ladder also.
Thermal consideration for PCB design
Philips LUXEON LED has a very specific land pattern , please refer to ANXXX for detailed description. in the github repo ,you will also find the Alitum library of Philips LED.
LED Running at full capacity , generate a descent amount of heat , due to unnecessary hassle of having drive on separate board and LED on another , i just totally avoided metal clad PCB , because metal clad is mostly single layer and dual layer MCPCB will be fairly costly. but i do opted for a very good quality FR4 metrical with better temperature grade.
after prolong operation at 700mA current the thermal design seems to be doing very fine for LEDs, though every thing worked fine as we were feeding fairly low minimum voltage ~6.0V driver is not very efficient at voltages less than 8.0V. more details about efficiency in available at the end of the article.
As i mentioned earlier that i need 360 degree cover with 120 degree LED , so i have made a PCB design such a way that PCB can interconnected to form triangle and cover whole 360 degree. reset is explained the PCB design. they suppose to be some what incontestable. although there are better way to do this with slots and holes. but i don’t want to waste my PCB real-estate in any fancy solution . So i opted for this solution.
Supply also chained between boards. what it means that you just have power one board , the all be connected electrically with the interconnection arrangement.though PWM dimming signal need to be supplied separately .
As i was already aware of poor efficiency at low input voltages i did some maths to get my self ready to have some arrangement for heat. as per theoretical calculations At 6V input efficiency is about 67.5 % only. so if LED running at 1A , they will consume 9W, and circuit will be eating around 14W ( without including power loss in inductor and other component ).by this we will be dissipating around 5 W in the driver IC only , this much power is too much for such a tiny IC , it is going to make drive too hot if heat is not spreaded across the board. as i intentionally kept the bottom the PCB totally clean , so i bought my self little heat sink , 20mm X 20mm X 10 mm and pasted on the back in whole length wither thermal tape, covering back of Driver thermal pad the LED also. with passive heat sink it worked just fine , driver was not getting too hot to fail. though IC has Program LED Current De-Rating vs Temperature. which means , user can program if they want current to lower down if IC get too hot. but i am not using that for now.
|Heatsink , size 20x20x10mm , could notable to fit to , cut into half 20x10x10mm
at 6.0V input , with 700mA led current , LED consume ~6 W , at 67,5% efficiency we will be feeding 8.88W ( without including indutor and other component power loss). by this we will be dissipating around 2.88 W in the driver IC only , this much power is too much , it is going to make drive too hot but i have tested it for days in relatively hot environment ~35 degree ambient , it managed just fine.
At 8V input efficiency is about 85% with 1A load, so power loss is about 1.5 Watt , which is fine.
as it turn out that this IC is not very efficient at low voltages. but it is fine for my purpose . i don’t care about the driver much , i am more concern about MPPT software .
Click here to view solar MPPT li-ion charger.