Why my LTC3533‘s efficiency is only 35!

Hello:

     When I use LTC3533, i refer to  some information form ADI,such as < PCB Layout Considerations for Non-Isolated Switching
Power Supplies
 > <999A>(LTC3533's demo),and <power master class>(power class  form ADI'wechat in china). but my LTC3533' efficiency is only 35%.

     The specific phenomenon:

      1.when I first power for my circuit board, it's quiescient current is 140uA! but i setting automatic burst mode. when i touch Rburst\Cburst,the quiescient current is normal:47uA.

      2.when input 270mA  at  3.66V,  my  electronic load is 3.5V at 100mA.the efficiency is only 35%! and this monent the SW waveform is so ugly,IC can heat,IC's input voltage is bring down.

     

    a.schematic:

      

   b.PCB

my LTC3533 circuit board is two layers,is different with 999A.

c.

  

Parents
  • 0
    •  Analog Employees 
    on Nov 13, 2020 7:32 PM 4 months ago

    Hi,

    What is the operating point for the application?  Did you check if the part is not in PWM mode. Also the with the input current value something in the circuit looks incorrect. For Buck mode the input current should be lower. I hope the measurements are taken with Kelvin connection.   

    I think both the phenomenon are related with board layout. Also try to solder the parts and components well.

    I do not see a wide VOUT plane. Also the VOUT trace runs through the bottom layer. So the current return path is obstructed and might lead to noise coupling and additional thermal stress . Have you checked the 999A demo board ? You can also refer the demo board design files for comparison.

    Thanks     

  • Hi,

    My application is to test the LTC3533's efficiency, because we are designing a portable instrument. I am interested in how to check if the part is in PWM mode. and the  measurements are not taken with  kelvin connection(i will try later on).

    My layout was refered to 999A, i think it is all same except 999A is four layer board,because my board is two layer.Also as you mentioned: my VOUT trace runs though the bottom layer and do not have a wide plane,but 999A is also! May be i should make VOUT trace wider. I remember continuous current can runs though the bottom from top, only pulse current should not to do so.

    Finally, one of ADI's engineers who in china told me : if I use ADI' switch power supply, if I want get over 90% efficency, then I must use four layer board. and he did not tell me why. But if I use four layer board, then the cost would be high……

    Very thankful~

  • 0
    •  Analog Employees 
    on Nov 17, 2020 5:00 PM 4 months ago in reply to 纪烈臣385

    Hi ,

    999A is 4 layer board so the VOUT trace is on the bottom layer and ground plane is not slotted. So if you are making two layer board it is not advisable to refer 4 layer design. Try to increase the VOUT plane area and signal  separate the signal and power ground. Efficiency can get reduced if the board has poor thermal management and the part gets heated.

    Thanks

Reply
  • 0
    •  Analog Employees 
    on Nov 17, 2020 5:00 PM 4 months ago in reply to 纪烈臣385

    Hi ,

    999A is 4 layer board so the VOUT trace is on the bottom layer and ground plane is not slotted. So if you are making two layer board it is not advisable to refer 4 layer design. Try to increase the VOUT plane area and signal  separate the signal and power ground. Efficiency can get reduced if the board has poor thermal management and the part gets heated.

    Thanks

Children
No Data