Question about EMI and Signal Analyzer

Abdelrahman Ali

Hello PSim Community, I hope all is well.

 

I am a MSc student and my project this semester involved investigating EMI related issues in GaN device power converters.  I am wondering whether the Signal Analyzer has a resolution bandwidth for calculating the EMI signal and whether I can change that. I am also sure my professor is going to ask me about that because as far as I know, EMI receivers have a resolution bandwidth and otherwise, we cant compare the measurements with the standards "CISPR25" for example.

NikosDimitrakopoulos

Dear Ali,

Thanks for reaching out!

The signal analyzer, coupled with the line impedance stabilization network (LISN), offers a straightforward means to analyze EMI for your circuit under test. Here are three important comments to consider for these kinds of applications:

1. PSIM make things easy for you by offering the LISN and Signal Analyzer block, as well as EMI filter blocks. However, it's crucial for designers to incorporate accurate parasitic paths, encompassing parasitic capacitances and inductances within the circuit and components (non-ideal components). Non-ideal switching and gate driving are also important aspect. Remember, there's no one-size-fits-all tool in EMI simulations, and the 'Garbage In Garbage Out' rule holds true.

2. The Signal Analyzer's EMI, DM, and CM noise outputs present Time Domain Results. The resolution of these results primarily depends on the Time Step. Furthermore, to analyze results in the frequency domain using PSIM's FFT tool (we'll get to this shortly), it's important to consider two main principles:

    a.  The FFT algorithm in PSIM adheres to Nyquist's criteria, which means that the maximum frequency of the FFT analysis is calculated using the following formula:

         freq_max = 1/(2*time_step)

 

         In other words, in order to get a higher max frequency you need to use a finer timestep for the simulation. 

 

      b. Please also keep in mind that the frequency incremental step of the FFT spectrum is:

 

           delt_freq = 1/total_sim_length

 

3. I'd also like to offer some insights regarding the FFT algorithm. It's important to note that while the FFT is a valuable tool, it may not be ideal for EMI analysis. This is not a PSIM issue but rather an FFT issue. It is just not the perfect tool for the job due to the following reasons:

• Spectral leakage due to non-smooth signals at the beginning and the end of the simulation (sudden onset and decay of signals)
• Spectral leakage due to the fact that FFT sampling frequency is not an integer multiplication of measured signal frequency
• Spectral leakage due to the fact that the FFT window doesn't contain an integer amount of frequency component periods
•  Different FFT results based on the chosen time window. 

This is the reason why windowed frequency analysis algorithms are preferred in literature and EMI detectors. A simple example of a windowed FFT is the STFT (Short-Time Fourier Transform). You can export the PSIM time domain results as txt and then leverage Altair Compose and Python to use a ready-to-go python STFT algorithm to delve deeper in those issues.

Hope this helps!

 

Best regards,

Nikos Dimitrakopoulos