Hi All,

I have one EMI filter circuitry (not attached), it consists of standard sections as EMI filter does, for example Varistor, common mode filter, differential mode filter etc.

I have to perform the frequency response analysis using LTSPICE. I have the following questions:

1. It’s a 28 VDC powered filter, Do I suppose to have 28VDC source and add a 1V AC component in it to run the AC analysis?
2. Do I need to perform ac analysis for common mode and differential mode noise currents separately?
3. Should I sweep the frequency (range depends upon switching frequency on board and its harmonics) from output to input to simulate the conducted emission? If yes in that case what should I connect to my input? Input impedance (few ohms or 28VDC source)? or have LISN connected between supply and circuit?

Please bear with me if the above question(s) do(es)nt make sense. I basically want to analyse the EMI filter for all the possible scenarios I can
Or if you can suggest some other way it would be great help.

Thank you.
Sunney

 

hi sunney,
Do you have a LTS asc file you could post.?
Many AAC members use LTS.
E

 

The filter you are simulating may be rated for 28 Vdc, but that has nothing to do with its frequency performance. You do not have to have any DC power sources in the simulation.

All power line filters are characterized using LISN's, so you should use one in your simulation. Some filter catalogs have the LISN details, but for some you have to contact the manufacturer.

ak

 

hi sunney,
Do you have a LTS asc file you could post.?
Many AAC members use LTS.
E

Thanks E, but I cannot share the .asc file.
I got some luck while doing the simulation, kept the fixed AC source of 1Vac 225kHz (which is the switching freq of regulator present on board).
and then did the transient analyis and took the FFT of source impedance current.
Now the problem is as per standard, the x-axis should be dBuA and what I am getting is dB only in FFT.
Do you know what is the correct way?
Sunney

 

The filter you are simulating may be rated for 28 Vdc, but that has nothing to do with its frequency performance. You do not have to have any DC power sources in the simulation.

All power line filters are characterized using LISN's, so you should use one in your simulation. Some filter catalogs have the LISN details, but for some you have to contact the manufacturer.

ak

Thanks for the response AK, I have done the simulation using LISN and just 50 Ohm source impedance separately, I got some 20dB more attenuation by using LISN.

Sunney

 

Hi All,

I have one EMI filter circuitry (not attached), it consists of standard sections as EMI filter does, for example Varistor, common mode filter, differential mode filter etc.

I have to perform the frequency response analysis using LTSPICE. I have the following questions:

1. It’s a 28 VDC powered filter, Do I suppose to have 28VDC source and add a 1V AC component in it to run the AC analysis?
2. Do I need to perform ac analysis for common mode and differential mode noise currents separately?
3. Should I sweep the frequency (range depends upon switching frequency on board and its harmonics) from output to input to simulate the conducted emission? If yes in that case what should I connect to my input? Input impedance (few ohms or 28VDC source)? or have LISN connected between supply and circuit?

Please bear with me if the above question(s) do(es)nt make sense. I basically want to analyse the EMI filter for all the possible scenarios I can
Or if you can suggest some other way it would be great help.

Thank you.
Sunney

It might need a simulated noise source and probably a digital filter to cherry pick the bits of spectrum your EMI filter needs to deal with.

It isn't hard to get out of your depth with EMI filters - any spurious resonances can peak some band of noise rather than suppressing it.

I think the ferrites used are typically lossy - its better to turn noise into temperature rise than leave the possibility of it being reflected so it can cause trouble elsewhere.