Secondary of offline SMPS is earthed....worse common mode EMC?

[treez]

Hi,
Do you agree if the secondary of an offline SMPS is earthed, then common mode noise can be worse, because radiation can go into earth from the secondary side and thats what common mode noise is....ie, energy coupling away to earth rather than returning in either live or neutral?

 

[c_mitra]

because radiation can go into earth from the secondary side and thats what common mode noise is.

I am not so sure. But I am willing to learn.

My understanding is as follows:

Common mode noise is associated with a common mode voltage. This becomes important whenever you are using a differential source (for example an op-amp or a transformer). An ideal op-amp or a transformer uses only differential voltage as input and the individual voltages are ignored.

For a real device, the actual potentials are important and that is not because of noise. Internal structure requires that the common mode voltage must be kept within some reasonable limits. For a practical transformer, the core saturation can put a limit or for an air core transformer, the dielectric breakdown may be considered.

For a transformer the output is also differential. This is not the noise we are talking about.

Consider this gedanken: you are using a battery as the input to a DC-DC isolated supply. How does the earth come in the picture?

In my understanding, earth is acting just like a capacitor- it can source or sink some reasonable amount of charge.

Noise is basically a voltage source and is related to impedance. Unfortunately common mode potential can be noisy.

What happens if you remove the bypass Y capacitor (as far as the noise is considered)?

 

[Easy peasy]

you have earthed one side of the CM antenna - so yes CM radiation will be worse ....

 

[treez]

Thanks, yet strangely yesterday i had an offline SMPS whose secondary was not directly connected to earth. I then put a 2n2 MLCC cap from secondary 0V to earth (chassis), and the radiated scan improved.
i say "radiated scan"......our radiated scan is just a cable coming out of the enclosed metal panel containing the offline SMPS...and this cable then being wrapped through a near field B probe loop 7 times...and then running on to a load resistor.
(out setup is just to see what radiation comes off the cable).
(The cable carries 0V and vout of the offline smps.)
Admittedly the cable was screened and the screen was connected to chassis earth of the panel.

 

[c_mitra]

i had an offline SMPS whose secondary was not directly connected to earth.

Please pay attention to the following facts:

1. what you are trying to say that the secondary is NOT DC coupled to the local circuit ground. The transformer provides reasonable amount of AC coupling between the primary and secondary sides.

2. if the metal enclosure is not physically DC coupled to the local circuit ground, you cannot assume that the electric field inside the box is zero. In other words, the box is not acting like a cavity but the electric field inside is non-zero (and dissipative).

3. You should use a 100E resistor and next a 100K resistor and then a 1pF and also a 10nF capacitors and measure the noise using the setup. If possible, please obtain a noise spectrum V^2/Hz or V/Hz^(1/2).

4. The screen of the cable should be connected to the mesuring instrument side and NEVER to both sides...