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.

LMR38020: Extremely poor efficiency

Part Number: LMR38020


I am designing a power supply around the LMR38020. The specifications that led to the selection of this part are:

  • Input voltage range: 9VDC-36VDC (typical), 70VDC (absolute maximum peak during certain transient tests).
  • Output voltage: 3.3VDC.
  • Load: 200mA (typical), 300mA (used as worst-case during design calculations).

The design I arrived at is the following:

The switching inductor PN is TYA3015220M-10. Additionally there is an LC pi filter at the input:

I am obtaining the expected output voltage, the UVLO is working as intended, and noise measurements don't show anything concerning. However the efficiency of the regulator is extremely poor. The load at 3.3V was measured 192mA, with a 24V input measured 45mA -> 59% efficiency.

Attempting to verify the design against Webench, there are a couple odd behaviors of Webench that I'm fighting (trying to set the operating frequency to 300kHzm, every time I enter the value the site reverts it back to 200kHz,  but it appears that efficiency should be much better - in the mid-high 80's at least. The Webench report indicates this, although as noted I cannot set it to the 300kHz operating frequency that is designed.

The layout is on a six-layer PCB, with all components on the top side (L1), and a solid ground plane underneath (L2). The inductor and bootstrap cap are close to the switching node and routed with a wide pour, the input capacitors are placed as close as possible to the input pin, and in general there are no departures from the general recommended power supply layout rules:

Compared to general rules and also the suggestions in the datasheet:

Can you please help me identify the issues that are leading to such a significant loss of efficiency and steps to correct them - help is much appreciated.

  • Hi Matthew,

    Thank you for sharing your details. A couple of comments:

    1. Are you using the FPWM variant or the PFM variant?

    2. Using an input filter often will decrease efficiency.

    3. You can try to increase the value of R21 to increase efficiency, but it is recommended to keep this value under 1MOhm. 

    4. It is generally recommended to have the input capacitors as close as possible to the IC. Your input capacitors are close to the input pin, but the path to ground is relatively long.

    5. I believe your thermal pad is entirely disconnected from the top side ground. This needs to be connected on the top side ground for thermal performance. Additionally, some thermal reliefs on the top side plane from the individual pins as well as the components can negatively impact thermals and efficiency.

    6. Your ground connections also seem isolated from each other on the top plane (GND pin, VIN capacitor GND, thermal pad, other component GND pads). It would be recommended to connect these ground planes on the top plane to ensure good heat dissipation and power flow.

    Thank you,

    Joshua Austria

  • Hi Josh,

    1. We are using the PFM variant, full P/N is LMR38020SDDAR.
    2. Noted - we will investigate the impact of bypassing this filter. Besides minimizing the ESR of the series filtering element, are there other criteria to be aware of for their impact on efficiency? I'm familiar with TI's application note SNVA801 but efficiency is not discussed.
    3. Noted - will investigate.
    4. Noted - will investigate.

    5/6. Apologies for not making this clear, the top plane is a power pour, connected to the input 24VDC. The ground connections are made to the solid ground plane on L2, so the capacitors/thermal pad/ground connections are all tied together.

  • Hi Josh,

    We resolved this issue - the problem turned out to be a different component, and when we removed that and isolated the buck converter we are seeing ~83% efficiency.