How does placing cap between signal GND and earth GND solve common mode noise?

Question

Thanks in advance for your time.

I am reading several articles about differential mode noise and common mode noise. Among which I found the below 2 pretty instructive:

https://www.murata.com/~/media/webrenewal/products/emc/emifil/knowhow/26to30.ashx https://micro.rohm.com/en/techweb/knowledge/emc/s-emc/01-s-emc/6899

Let me first talk about my understanding as a start point.

1. Differential mode noise is some noise generated with the propagation of power line, such as noise generated from some on-board chips' operation like source/sink power to GND. This will produce ripple on power line, which is noise. In this case, placing capacitor between power and GND line solves the problem, it conduct the noise current to GND and flow back to power source. This is illustrated in Figure marked 1 and 2.

2. Common mode noise is pretty confusing. It is generated through the coupling between your on board signal trace or plane and the Earth GND. The start point is, there is coupling between your on board trace or plane and the surrounding mechanical box which is connected to Earth Ground, so this coupling will generate some current through it. Then this coupling current will flow back to your power supply's Earth GND, which its surrounding mechanical box is connected to, then through its stray capacitance, the noise will flow to both the power and GND lines that the power supply feed to your board. This is the signal flow of common mode noise.

Is my understanding correct?

My questions are:

• In the first article, Suppression method of common mode noise (2), it says adding cap between signal GND and earth GND solve common mode noise, why? I understand method (1), using common mode choke, but why method (2) also works? Just because common mode noise is referenced to Earth GND, so placing cap between where common mode noise flow through and Earth GND, can filter it out?

• Why doesn't differential mode noise's current go with the common mode noise's path?

Answer 1

The commom mode noise exists between both the power supply lines and the reference GND. Without any measures, the commom mode noise is visible at the load over the stray capacity to GND. If you put caps from the supply lines to reference GND, the circuit of the noise source is shortend before reaching the load.

Answer 2

Ok, well this takes me back a couple of decades when I was in a discussion with an audio electronics engineer on how the power noise analysis can be explain in theory format to the audio techs. No one really came out with a good way of explaining it, but I will give you my perspective on it. I copied this from a pro audio forum, but since it was about this, this help you out. The only difference here is that in pro audio we have circuits with dual supply, and signals not always referenced to ground, which are considered balanced signals the could be in transmission lines or through specially designed circuits. So:

From here.

You see, when a power supply rectifies the signal to DC, the noise is superimposed on the rectified amplitudes on both the + out and and - out in phase. Now if we have two supplies in the circuit (like a +15 and -15 rails) we have one supply with the - out tied to a "grounding point" for the positive supply and the other supply's + out is tied to the "grounding point" to make the negative voltage supply. now what has happened is at the "grounding point" the noise from both supplies are added.

Now lets look at the states of noise that has developed with this dual supply with common DC ground. Which are defined like this:

Positive Supply Unbalanced Common Mode Noise: This is the noise measured and observed when referencing the Positive supply to the common DC ground.

Negative Supply Unbalanced Common Mode Noise: This is the noise measured and observed when referencing the Negative supply to the common DC ground.

Differential Supply Balanced Common Mode Noise: This is the noise measured and observed when referencing from the Positive Supply to the Negative Supply. Ideally, we would like to see a total cancellation here, and the ideal setting of equal and opposite supply voltages cancel the most noise deferentially.

Now to apply decoupling a certain way for suppression of diode junction noise, rfi, ect from the power supply depends on what noise source you want to suppress. Like the noise difference that develops between two grounds or two internal ac signal returns.

But you have to take in consideration: Where you are referencing the signal. Because an ac signal transmission line doesn't have a any dc ground reference, and internally, the ac signal in a circuit has several signal returns. They are the dc power ground, the positive dc supply, and the negative dc supply, and the inversion of the signal itself.

Answer 3

The truth is there is no such thing as differential mode noise and common mode noise. All radiation is dependent on the energy of the field between two points and every source of noise is dependent on a difference in electrical potential between two points.

Now with your differential signal there is a potential difference between the positive and negative terminal. Relative to earth ground the energy source, for instance a battery, has a higher potential as well. Even if earth ground and a battery are not connected there is a parasitic capacitance between them that depends on their distance and the geometry of the conducting plates. Usually this capacitance is in the range of picofarads.

There is always a current loop when there is a difference in potential, and E and M fields exist infinitely in space. The gradient of an E field is a potential difference and parasitic capacitances have a current known as displacement current. A very small current usually but it can be large given the right circumstances.

Therefore there is a path for all of your currents on your board that goes through ground and returns to your battery. Typically these are decomposed into a common mode current representation but it's equally fine to see them just as the capacitively coupled path to earth for this particular potential difference.

So what are we doing when we add a capacitor from our circuit to earth? We are doing the exact same thing when we add a decoupling capacitor in our differential line, changing the impedance of the path for different frequencies. If we have a 1m wire whose capacitive coupling is equal along the length we get a particular frequency response for the signal. If we add a capacitor the frequency response changes because we introduce at a particular point a lower impedance path for certain frequencies. This will depend on where along the length of the line we place the capacitor and the value of the capacitance. We influence both the loop area, and the impedance.

To most effectively do this, which is fundamentally just manipulating the cut off frequency for a low pass LC filter, we need to make the decoupling to earth equal for both lines. This minimizes the noise between a line and earth without adding differential noise.

In fact one can completely remove common mode signals in a circuit through a method known as impedance balancing, akin to an impedance balanced bridge.

If the impedance in a bridge is balanced the currents through the center path is nullified. If you consider the battery in this image to be earth injecting a common mode current into your circuit, take point 1 as V+ and point 2 as V- of a DC source, you can effectively use this to remove common mode currents from your circuit and hence common mode noise entirely (theoretically of course).

The above shows radiated emissions with both differential and common mode noise so it's not entirely clear how effective this is for the lower frequencies, but you can also get a greater reduction in common mode noise if you attempt to decouple to earth at the load.

Here the peak line is shown for radiated emissions of only the common mode noise. This method also decouples to earth at the load and is quite significant in its reduction.

Note: You may need to watch out for leakage current to earth through your capacitors depending on the voltage of your application in terms of safety standards.

References: Modelling and measurement of high frequency conducted electromagnetic interference in DC–DC converters - Grobler, Gitau

Modeling and Optimization of Impedance Balancing Technique for Common Mode Noise Attenuation in DC-DC Boost Converters - Zhang, Zhang, Lin, Takegami, Shoyama, Dousoky

A Study on Common-Mode Noise Generation in Switching Circuit due to Unbalanced Characteristic - Intachot, Klungwijit, Prempraneerach, S. Nitta