From Thomas, United States
Question:
My colleagues said that the out of spec common-mode emissions observed in system-level test in the 3 GHz range was from the high-data rate bus connecting two electronic boxes. The question we have was could the GHz emissions were the higher harmonics corresponding to high data rate transmission? The cable linking the two boxes were rather long, it seem to me that it cannot be an efficient antenna for such high frequency noise.
Answer:
Your colleagues might have a point, because all cables longer than 1/10th wavelength should be assumed to be relatively effective ‘accidental antennas’ (when they are not being very effective).
Because the measuring antenna is in the far field, it sees the aggregation (addition, combination) of all the little quarter-wave and/or half-wave near-field segments along a cable.
However, the emissions radiated from all these little near-field radiating segments all reach the antenna from differing distances, causing phase shifts that cause some degree of phase cancellation as they combine in the antenna.
Also, because real cables are not dead straight, and rarely perpendicular to the antenna, there are extra contributions to the phase shifts of the little near-field radiating segments.
The result is that – for a given surface (CM) current – a cable that is exactly as long as a quarter-wave (or a half-wave) tends to measure the highest far-fields at the measuring antenna. Longer cables don’t measure any higher, and often measure lower due to phase-cancellation of the multiple near-field radiating segments.
Complicating this further: all cables have a DM to CM conversion ratio, and all shielded cables have less-than-infinite shielding effectiveness. So, as a cable gets longer, it tends to create more CM surface currents, increasing its overall emissions – going some way to reversing the trend for phase-cancellation of multiple near-field segments to reduce the measured emissions. Possibly even causing overall emissions to increase.
Further complicating the situation, the boxes at the ends of the cables can behave as if they are extensions to the radiating ‘segments’ of the cable. This is because surface currents flowing on the outside of the shielded cable must flow on the outside of the shielded boxes the cable connects to.
That’s all I have for you now. I will also try to write some contents and do some tests on this subject later on. Leave theory aside, why not do some experiment to prove it?
Fischer has an RF current probe goes over 2 Ghz, https://www.fischercc.com/products/f-2000-12mm/?user_login=&_xoo_el_form=lostPassword&_wp_http_referer=%2Fproducts%2Ff-2000-12mm%2F . You can always measure the common mode current on the suspicious cable. Alternatively, disconnect the cable and test the system? Near field probing would also help you to distinguish the noise souce.