Capacitive coupling between hot and cold ground

Question

Its understood from other Q&A regarding this topic, the link between rectified mains side and low voltage DC side of most smps are to provide some sort of grounding for very high frequency EMI noise so that it is not emitted through whatever load connected to the low voltage DC side.

What confuses me is, that capacitive link could worsen the EMI problem by coupling the high frequency noise into the household power lines and turn the mains wiring a larger antenna.

Why its not happening?

Answer 1

EMI is effectively separated into two categories; conducted and radiated. The separation between is a frequency. This frequency depends on your area.

Basically conducted emissions don't care about anything above the cutoff frequency and radiated emissions don't care about anything below the cutoff frequency.

If you shunt a large amount high frequency noise into the power lines, your device is probably already failing regulations for radiated emissions. If your shunting just a little high frequency noise into the power lines then it's not a problem.

Also antennas occur when the frequency is some fraction of the wavelength. 1/2 or 1/4 and so on. A full wavelength also works, but when you get into areas where many many wavelengths fit on a wire it does not make such a great antenna. Further it is subject to greater losses.

Answer 2

Think in loops!

The goal of any decoupling or filtering system is to provide an express "Return To Sender" function for the unwanted signal (i.e. the EMI). In this case, the secondary-side high frequency EMI components are a byproduct of the secondary-side switching (i.e. the diodes). Without the Y capacitor, in a Class II supply, they would otherwise have to take a circuitous path back to the primary side of the supply (which is where they came from initially!) via whatever leakages and connections they could find. The Y capacitor, however, provides them with a short path back to the primary-side switch which initially generated them, reducing the EMI output of the supply at the cost of some mains leakage current through the Y capacitor.

(If you care that much about the mains leakage BTW, just get a medical grade supply -- they can't afford the leakage through the Y cap, so they use other techniques to keep the secondary-side noise down to the point where no Y cap is needed to meet the EMI specifications.)

Answer 3

The application of Ycap shunting AC rectified, neutral side ground (or hot side) to DC gnd (cold side) must include coomon mode and differential line filters. These line filters serve several purposes.

• reduce egress
  • i.e. conducted and thus radiated line noise (unintended radiators) as define by FCC and CE compatibility limits for EMC

• reduce ingress

  • i.e. power line transients (PLT) up to 6kV possible but must pass 3kV no damage or safety risk above which, a protective fuse may be used to limit crowbar circuits or arc gaps or component failures.
• RF noise sink for crosstalk or internal radiated noise
  • since the power line output impedance is lowest available reference point, you do not need a safety ground to shunt RF and internal SMPS transient noise on the chassis which can cause egress to internal and external devices.
  • study my prior answer board layout on this to see the 2 stage CM chokes with differential X caps and CM Y caps to create a low pass filtered Hot ground for RF yet high impedance to AC. Thus two redundant 250V Y caps are more reliable to connect to cold ground with SMD parts give good RF impedance shunting with the chassis as the RF sink to stray noise and not the AC line,

Answer 4

What confuses me is, that capacitive link could worsen the EMI problem by coupling the high frequency noise into the household power lines and turn the mains wiring a larger antenna.

I assume that we are talking about an SMPS that doesn't have a "third" ground pin connector that can be relied upon for shunting away this output noise and therefore it uses a capacitor like the one marked "2.2 nF 250 VAC" below (top middle): -

_{(source: 320volt.com)}

So, in the absense of that ground pin you have to use what is available and if that is the rectified live wire then that's all you have. If the noise passed to that node is too high to meet conducted emissions (or there is too much radiated emissions from the AC wiring) then you will have to use in-line filtering on the AC side.

That's not as bad as it sounds because, for instance, on a small power supply, the noise passed from secondary to primary side circuits is usually lower than locally generated primary noise due to the primary being switched at the PWM frequency.

On larger power supplies there inevitably has to be chokes present to reduce primary switching noise to a legal level and this will also reduce the secondary transferred switching noise.