Telecom Lines Conducted emissions noise travel

Question

I am seeking an explanation of how conducted emissions on telecom lines are measured. The frequencies of interest are at 150kHz to roughly 850kHz. The size of the unit is 30cm by 30cm by 15cm. The enclosure of the EUT is metallic.

Specifically, I would like to understand the following:

1. How noise travels within the system, obviously through the wires?

2. How this noise is conducted to the Impedance Stabilization Network (ISN). Assuming the system has a chassis ground connected to the same reference ground plane as the ISN.

Here is what the system looks like. The chassis is a metallic enclosure that is grounded to the RGP (Reference Ground Plane) using a ground strap. The ISN is connected to the same RGP. Since we are measuring the LAN port, i.e. Asymmetric port I am expecting that we measure a Common-mode noise.

When performing the measurements, we do not see noise when the system is grounded, and we see noise when the system is floating. That seems to contradict our expectations since again we expect to see Common mode noise.

This is the setup where the unit is grounded.

This is the setup with the unit floating.

EDIT 2: The unit has two grounding points on the chassis. I followed Tim's advice and i shorted them both with a ground strap.

Here is the system floating.

This is with one of the grounding points connected to the RGP. Same thing happened when both grounding points were shorted to the RGP.

Answer 1

Presumably, no CM current in the Ethernet port means no ground current, and lifting it shouldn't make any difference.

This is an unverified assumption, however. A current clamp on the ground connection would prove it.

You don't say what schematic, layout or assembly is, so that leaves some possibilities still.

One possibility is the enclosure is multi-part, and a noise source is driving one part, while the ground acts as a filter ground reference point:

Presumably, this is easily solved by joining the pieces together, shorting out the offending source. Perhaps the enclosure itself is plastic, and it's a multi-board assembly, but in that case the PCB ground planes serve as enclosure panels, and need to be joined, to the same end. Grounded mounting holes and metal standoffs would be a typical method; a metal-lined enclosure is another (via self-adhesive foil or metal spray). Ferrite beads or CMCs on the connecting cables might also be useful.

Such a mechanism falls under the topic of enclosure port emissions, anything emitted from the EUT by internal radiation, or capacitive or inductive coupling. The coupling factor depends on the relative size of source and victim; a small EUT (few inches, say) might be constructed in such a way but not exceed limits, while a 0.5m+ box might handily do so towards the upper end of the band (10s MHz). Problems will be more apparent in radiated, where the panel or loop elements act as antennas, radiating energy.

More generally: a panel might act as a patch antenna or dipole, being "electrically short" at low frequencies, thus being in the asymptotic cutoff region where gain rises at 20dB/decade. At the low end, this can provide enough attenuation to pass, but at the high end [of the band], or for larger panels, emissions can be problematic. And similarly for magnetic sources (loops). Up into the radiated band, such antennas continue to rise in gain, until topping out with whatever resonances their structures produce.

---

EMC is a complex, interconnected topic, and EMC problems are holistic in nature. There is generally very little that can be said about a problem without also completely describing the setup and environment. Keep in mind that this answer only extends as far as it does: a possible explanation, for an incompletely described phenomenon. As one adds details to the question, the answer can change radically, as various other effects come to dominate. This answer is not a diagnosis, nor can a diagnosis be given based on such meager information.