This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

CSD19538Q3A: Calculating Temperature when using pulsing

Part Number: CSD19538Q3A
Other Parts Discussed in Thread: CSD19538Q2


I want to make a constant current LED driver with a MOSFET, shunt, opamp configuratie. 

The specs of the LED driver:

  • Frequency: ≤100 Hz
  • Forward current: 1 A
  • pulse duration: ≤ 300 micro seconds
  • Source: 48 V
  • Forward voltage LED's: 34.5-45 V
  • Sense Resistor: 1 ohm
  • Pmosfet: max 12.5 W during pulse
  • Ambient temperature: 60 degree C

I came up with the schematic underneath Vic is set by a analog output and Pulse sets the frequency and duration of the pulse.

Now my question how can I find a suitable MOSFET that can handle the given power dissipation without the requirement of adding a heat sink.

  • Dylan,

    Thanks for considering using our MOSFETs.

    As for the application conditions, I estimate the losses will be <100mW in the MOSFET.

    The CSD19538Q3A will easily handle this type of power loss on a typical pcb design, I would simply suggest you make sure you follow the footprint guidelines in the datasheet. Some other pcb guidelines for these types of packages can be found here:-

    If you are looking for a smaller device, another alternative would be the CSD19538Q2, this is the same silicon but in a SON2x2 vs a SON3x3 package (~60% smaller). This could also easily handle 100mW of power dissipation.

  • Dear Chris,

    Thank you for your answer but I calculated that it will be 12.5 Watt during the pulse.
    Is this correct or do I forget something?

  • Dylan,

    How are you calculating losses??
    I did a quick calculation of losses based on DC , this would be I2R , @ 1A resistance of ~50mOhm the losses are 50mW, you are not DC and would have significantly less and the switching losses @ 100kHz would not double this, as such safely assumed <100mW losses in the FET.
    What is your gate drive voltage? The only way you could get 12.5W would be to use the FET in a linear fashion and operate just above threshold and have greatly increased resistance of the FET, for 12.5W this would mean a resistance of ~12.5Ohm for the FET given a 1A current.
  • Dylan,
    Thanks for reaching out. I reviewed the circuit and power loss calculation. The FET is operating in the saturation region and regulating 1A constant current for the LEDs when the FET is on. 12.5W is the instantaneous power for 300us when the FET is on:

    Pfet_inst = (Psupply – Pled – Psense) x IF = (48V – 34.5V – 1V) x 1A = 12.5W -> abs max power rating @ TC = 25C is 23W

    We can multiply this by the duty cycle (3% for 300us @ 100Hz) to determine the average power dissipation:

    Pfet_avg = Pfet_inst x 300us x 100Hz = 12.5W x 300us x 100Hz = 0.375W.

    Using Rtheta(j-a) = 55C/W, I calculate the temperature rise = 20.6C (TJ ~ 80C).

    I also looked at SOA using 1ms pulse (conservative) with VDS = 12.5V and it’s about 7A. If I derate for 60C ambient I get about 5A. You should be OK using the FET in this application with the requirements you provided.
  • Dear John,

    I thoughts I had to use his continuous duty cycle and use the K/W value from figure 1 of the datasheet.
    So use Pfet_inst = (Vsupply – Vled – Vsense) x IF = (48V – 34.5V – 1V) x 1A = 12.5W
    12.5W * 0.2 * 55C = 137.5C
    137.5C + 60C = 197.5C at the end of the pulse.

  • Hi Dylan,
    Thanks for the follow up. Your approach to the temperature rise calculation is correct. However, figure 1 is using junction-to-case thermal impedance, not junction-to-ambient. We can use this to calculate the junction temperature rise above the case temperature: 12.5W x 0.2 x 5.5C/W = 13.75degC. We measure case temperature at the drain pad on the bottom of the package. You can also measure the temperature on the PCB next to the device or use an IR camera to measure the top of the package. Most of the heat is dissipated thru the bottom of the package into the PCB so the top of the case will be a few degrees C below the junction temperature. We can only guarantee junction-to-case thermal impedance since junction-to-ambient (or PCB-to-ambient) is very dependent on PCB layout, copper weight, number of layers and etc. Using the transient thermal impedance, you can determine the acceptable case temperature in order to maintain Tj < Tjmax (with some derating).

    For additional information, please visit TI's MOSFET blog series: Part 6 discusses thermal impedance and has a link to TI's thermal analysis landing page: Part 4 discusses MOSFET pulsed current ratings and transient thermal impedance.