2
\$\begingroup\$

enter image description here

enter image description here

My circuit and its conducted emission test result are in the attachment. The circuit is smps circuit for three phase energy meter. As you see it fails conducted emission test. What would you suggest to improve circuit to pass the test. Notice that I cannot add any more component because its pcb is manufactured. I can change only values of components.

\$\endgroup\$
3
  • 2
    \$\begingroup\$ No line filter with CM choke and Pi cap filter , tisk tisk. Quasi peak is 10dB > avg.at 200 kHz so LC filter must attenuate >10 dB more by increasing LC product by 10 \$\endgroup\$ Jul 20, 2018 at 13:22
  • 1
    \$\begingroup\$ I'd be surprised if it meets power factor regulations too. This type of bridge design is notorious at failing 3 phase power factor regulations. \$\endgroup\$
    – Andy aka
    Jul 20, 2018 at 14:53
  • 1
    \$\begingroup\$ You skimped with single sided L 1mH and have on Y caps with no X caps . X caps limited by gnd current but not Y caps which attenuate quasi-peak and average conducted noise /2 but are not current but need current limiting \$\endgroup\$ Jul 20, 2018 at 15:12

1 Answer 1

2
\$\begingroup\$

As said by Tony in his comment, you should increase the attenuation of the LC filter, however in the direction from the MOSFET to the AC mains: this means that \$L_2\$ and \$C_{64}\$ should be risen as much as possible. This measure could be sufficient, since the conducted emission from switching power supply is mainly a differential mode emission (it appears as two signals in phase opposition on the Line and Neutral conductors). Do not worry for the diodes, as they are seen as short circuits by the EMI conducted noise. To explain more precisely what I mean it is convenient to do a rough copy of your circuit, as I have done below:

schematic

simulate this circuit – Schematic created using CircuitLab

where

  • LISN is the LISN network needed for the conducted emission test, shown with the spectrum analyzer (or better the RF receiver) connected.
  • \$V_{AC}\$ is the voltage generator which models the power grid.

Modeling the \$Q_2\$ block as a square wave current generator and remembering that the diodes of a power supply bridge behave as low value resistors respect to the signals generated by the downstream circuitry (see Gray and Searle "Electronic Principles" Wiley, 1969), we have the following equivalent circuit:

schematic

simulate this circuit

where

  • \$Z_{Power Grid}\$ is the equivalent impedance of the power grid as seen fro the circuit input,
  • \$R_D\$ is the conducting diodes equivalent resistance

As you see, you have a circuit topology that resembles that of a \$\pi\$-filter: the only choice you have to make your circuit fit the conducting emission limits is to make this topology work for you, therefore you should try to rise \$L_2\$ and rise \$C_{64}\$ as much as possible, as already said.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ How about C58 and C63? Should I increase their values? \$\endgroup\$
    – mehmet
    Jul 23, 2018 at 6:09
  • 1
    \$\begingroup\$ Since the \$Q_2\$-\$TR_1\$ series is approximately modeled as a current source, rising the values of \$C_{58}\$ and \$C_{63}\$ should have a nice effect too on conducted emission, though probably not as good as rising \$L_2\$. \$\endgroup\$ Jul 23, 2018 at 7:46
  • \$\begingroup\$ @mehmet I am curious to know if you succeeded in solving this EMC problem: did you? \$\endgroup\$ May 20, 2019 at 18:56
  • 1
    \$\begingroup\$ Yes I solved the problem by rising C61, C62, C64, and L2. \$\endgroup\$
    – mehmet
    Jun 13, 2019 at 14:52
  • \$\begingroup\$ @mehmet, wonderful. So if you think that my answer has helped you, please consider accepting it. Best, \$\endgroup\$ Jun 13, 2019 at 14:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.