Filter desing for EMI frequance

[rsklan]

Hi everyone,

I am conducting EMI/EMC investigations on my PCB. I observed the noise levels and noticed a few things. For example, EMI noise is being generated by the inductor in the converter structure. This occurs at around 900 kHz. I have detected this, and I want to design a filter for it. In such cases, how should a filter be designed? For example, I have several noise sources, and I know their frequency values. What should I consider when designing a filter in this scenario?

 

[barry]

Radiated? Conducted? What do you want to filter, input to an opamp? Output connector? Power supply rail? This is too general a question to answer.

 

[rsklan]

Here's the translation of what you wrote:

---

I am just looking for general guidance on how to approach this kind of problem. For example, I have a motor driver H-bridge circuit, and I am unable to read the current through the ADC. From an EMI perspective, I used a near-field probe and detected noise between 900kHz and 2MHz. Later, I performed an EMI test and observed peaks at these frequencies. This is just a scenario. In such cases, what general approach should I take to move forward?

 

[barry]

Again, you don't provide any useful information. All I can infer is:
1) You've got some radiated EMI.
2) You're ADC is not working.

I see no correlation. Are you seeing the radiated frequencies in your ADC readings? Just saying "I am unable to read the current" tells us nothing about your problem. This is like saying "There's a bright light outside my window. My bicycle doesn't work."

There are LOTS of ways of mitigating EMI problems: fixing the problem at the source; adding a filter on inputs; shielding. Without information it's impossible to answer.

 

[stenzer]

Does the ADC works as expected when the H-Bridge is turned of? You have for shure an ADC ipnut filter where you can remove the series resistor and apply an external known voltage e.g. an AA battery, to check the ADC readings.

BR

 

[cupoftea]

Please provide drawing of where is you current sensor, and how you get its signal back to the ADC?....if by PCB...then how is your PCB layout?...if by wire, then what type of wire.....shielded coax, or loose wires?
Use of a current sense transformer can be a good way to beat noise on current sense signals.
Do you have the wires too close to the switching nodes of the inverter for example?
Hall sensors can be good too in defeating certain noise scenarios.

How is the PCB layout of your inverter?....or is it an offThe Shelf module?

Are you using a sense resistor, and its in the ground return of the inverter...and shares this ground with the ADC, but the amount of current flowing between these grounds means they are not actually the same potential....or maybe you have not filtered the HF current pulses of the inverter well enough, and so you are getting ground bounce which is messing your adc operation.