LT3952 to power 450mA 42V LEDs with 4000:1 PWM

Hi

I am trying to design a LED driver circuit to drive string of two LEDs at 450mA and 42V (constant current - voltage is maximum from LED specs, in practice it is around 39-40V). The voltage on input is provided by 12V 6A power supply. I have used DC2361A evaluation board schematic as a base but quickly run into problems with the rectifier burning off. The current schematic uses slightly amended circuit with better schottky and new inductor with larger current tolerances from CoilCraft.

My circuit is controlled by MCU and uses the full set of contrast ratio (4000:1) hence the choice of switching frequencies (MCU uses 100Hz at 1:4000):

I now run into a thermal management problem - the LT3952, inductor and shottky get pretty hot (84-85 degrees Celsius). Using single LED (450mA at 21V) keeps the LT3952 at 44-47 degrees Celsius. The board layout I am using follows the guidelines and switching node layout follows the same patterns as DC2361A but on 2 layers PCB. I am looking at some advice how to optimise the temperature - any advise on layout or changes to the circuit would be greatly appreciated.

Thanks

Alexey

Parents
  • Thanks. I have slightly redone the layout - moved R1 closer and parallel to the LT3952 (shortened the traces by 1/3) at 15mOhms that should help somewhat and more importantly added vias stitching all the way through the LT3952 thermal pad. Will try this layout in the next few days.

    I also have experimented with existing layout and different frequencies The figures are as follows (temperatures are all in Celsius, in all the cases after 30-35 minutes of running on always on):

    1) At 1 MHz switching, CoilCraft XEL5030-472 (4.7uH at 36mOhms DCR) the thermals are: LT3759 - 64 degrees, inductor and diode at 58-60 degrees

    2) At 1 MHz switching, CoilCraft XAL5050-103 (10uH at 41mOhms DCR) the thermals are: LT3759, inductor and diode all at stable 63 degrees

    3) At 400 KHz switching, CoilCraft XAL5050-103 (10uH at 41mOhms DCR) the thermals are: LT3759 and diode at stable 57 degrees, inductor - 64 degrees

    4) At 400 KHz switching, CoilCraft XAL5050-223 (22uH at 91mOhms DCR) the thermals are: LT3759 at 70 degrees, diode, inductor - 67 degrees

    Will also try the 8uH one at 400KHz to see if that can be improved. So far the (2) and (3) look the best. Tested PWM range at (2) and all 4000:1 dimming seems to work fine. At (3) range of useful PWM is reduced to 1000:1 (have not tried 2000:1 yet - may work as well).

Reply
  • Thanks. I have slightly redone the layout - moved R1 closer and parallel to the LT3952 (shortened the traces by 1/3) at 15mOhms that should help somewhat and more importantly added vias stitching all the way through the LT3952 thermal pad. Will try this layout in the next few days.

    I also have experimented with existing layout and different frequencies The figures are as follows (temperatures are all in Celsius, in all the cases after 30-35 minutes of running on always on):

    1) At 1 MHz switching, CoilCraft XEL5030-472 (4.7uH at 36mOhms DCR) the thermals are: LT3759 - 64 degrees, inductor and diode at 58-60 degrees

    2) At 1 MHz switching, CoilCraft XAL5050-103 (10uH at 41mOhms DCR) the thermals are: LT3759, inductor and diode all at stable 63 degrees

    3) At 400 KHz switching, CoilCraft XAL5050-103 (10uH at 41mOhms DCR) the thermals are: LT3759 and diode at stable 57 degrees, inductor - 64 degrees

    4) At 400 KHz switching, CoilCraft XAL5050-223 (22uH at 91mOhms DCR) the thermals are: LT3759 at 70 degrees, diode, inductor - 67 degrees

    Will also try the 8uH one at 400KHz to see if that can be improved. So far the (2) and (3) look the best. Tested PWM range at (2) and all 4000:1 dimming seems to work fine. At (3) range of useful PWM is reduced to 1000:1 (have not tried 2000:1 yet - may work as well).

Children
No Data