EMI design for ACDC power supply

How to design a EMI filter for an ACDC power supply?

I don't want any ready made filter. How do I know if a filter is enough or do i need to a change?(I don't have any now)

Thanks

If it is offline and sub 75W...then your electrolytic capacitor after the bridge will pass most of it for you. Then if its a flyback you can just do the rest by putting your spec into power integrations PI Expert software.
Then when you get the scans done.....you maybe see a fail by say 10dB....then you can do an extra bit of filter which gives you that extra 10db of attenuation at the offending frequency.

The EMC receivers are extremely sensitive...and they kind of do the sensing in frequency windows of i think 9khz or so.....so its difficult for you to calculate it from scratch.
Especially if you have one of those switchers which modulates the switching frequency to reduce emi.

How's this for an active PFC filter?



- both ends of the jumper connect together just after the ATX supply fuse and the loop is clipped under a plastic tab next to the shielded ferrite core and wrapped under transformer tape.

There are no other winding's visible.

- it looks like an LC series notch filter for the PFC switcher to prevent conducted noise.


Any other hunches?

Tony

Except for the missing winding (!), it looks like a series choke as is has been used in first "PFC" PC power supplies. It could scarcely achieve the harmonic current specification for certain load conditions.

so what is the equiv circuit Frank?

The inductor core in the photo seems like a fake device - no useful equivalent circuit.

The custom plastic moulding is designed to fit the loop with vertical exit, so the loop couples the grounded magnetic element like an RF current transformer by capturing the stray field just a a shorted scope probe does for impulse noise.

So I see a low impedance coupling at high frequency above 10MHz. But it only ties to the active line and CM or DM but rather line to GND. An impedance transfer function could be done with a pulse or square wave about 3kV <1us rise time using a series R to capture the response when connected. Testing may be a challenge. I see the coupling capacitance in the >10pF range being in the 1kohm reactance range and the series inductance of the loop being about 100nH or a notch filter at 100MHz, rated for 6kV of insulation, which doesn't seem useful.

SunnySkyguy said:

How's this for an active PFC filter?

- both ends of the jumper connect together just after the ATX supply fuse and the loop is clipped under a plastic tab next to the shielded ferrite core and wrapped under transformer tape.


There are no other winding's visible.

- it looks like an LC series notch filter for the PFC switcher to prevent conducted noise.


Any other hunches?

Tony

Click to expand...