A case of switching power supply conduction and radiation processing -----Layout loop

This is a power supply with a wide input voltage of 120-277V 60HZ and an output of 48V and 273mA. It uses Maxic MT7933 and adopts Buck topology. 

Note: In the initial design, reserve inductors L1, L2, CBB capacitors C1, C2 as conduction test components, and reserve magnetic beads FB1, ceramic chip capacitor C9, chip resistors R14, R15 as radiation test components;

Conduction test:

1. Short circuit L2, L1=4.7mH, C1=0.1uf, C2=0.1uf,

120V voltage input, L line conduction image:

277V voltage input, L line conduction image:

Result: Input 277V, the margin is less than 3db after reading the frequency of nearly 150K

Correction method: increase C2 to 0.22uf, and test the image again as follows:

Result: Manually read the point, the margin is 7.19db, after verifying the N line, it passes without pressure.

Radiation test:

1. In the case of not adding magnetic bead FB1, not adding loop capacitance, and transformer not covered with copper skin, the radiation data seriously exceeds the standard;

2. Rectification plan: Strengthen the following rectification positions, namely: increase the magnetic bead FB1 (100M 60ohm), and the loop capacitance C9=1nf.

　　Result: horizontal test, the margin is close to the limit line; vertical test, the radiation data at 31MHZ, 41MHZ, 53MHZ exceeds the standard by 5-10db;

3. Rectification plan: Based on the second step, wrap the transformer with copper skin for external coils, and increase C9 to 2.2nf;

　　Result: no improvement in the situation;

4. Rectification plan: Based on the third step, a 60pf capacitor is connected in parallel to the D and S poles of the MOS tube Q2;

　　Result: no improvement in the situation;

The image is as follows:

5. Rectification plan: Focus on the high-frequency switching point (as shown in the red circle below), and adjust the Layout wiring design:

In the original wiring scheme, pay attention to the high-frequency switch part (the black line with "X"), and find that the high-frequency trace is too long and the loop area is too large:

     

After re-layout and Layout:

      

 result:

Design parameters without adding copper skin to the transformer, loop capacitance C9=2.2nf, magnetic bead FB1 (100MHZ 60ohm):

120V horizontal and vertical test pattern

   

277V 60HZ horizontal and vertical test images: