Grounding and EMI

One of the most common direct coupled noise sources is when the ground which is being used for reference or return is not referenced to earth as expected. This is especially prevalent in sensitive high-gain circuits.

Grounding and Shielding Existing Equipment – How to effectively minimize EMI issues when best practices are not available is a good paper mentioned by NASA Tech Briefs and Automation Weekly.

It tells that undesired signals couple into signals via four coupling factors: magnetic (inductive) coupling, capacitive (high speed voltage changes (dV/dt)) coupling, direct coupling, and radiative (Rf) coupling. Numerous books, articles and white papers are written all the time about this (many books on Electromagnetic compatibility cover those topics well). The material you will find include information on the coupling mechanisms and the sound mathematics of measuring inductance, capacitance, mutual inductance, resistance, and field intensity.

Most of the time the shield can be viewed as a band aid on a wound. Although necessary, one should not ignore the bleeding below. It is important to understand what is causing the noise and whether it can be resolved.

In summary, for protection against low-frequency (<1 MHz), electric-field interference, grounding the shield at one end is acceptable. For high-frequency interference (>1 MHz), the preferred method is grounding the shield at both ends, using 360° circumferential bonds between the shield and the connector, and maintaining metal-to-metal continuity between the connectors and the enclosure. Also safety considerations may require that the remote end of the shield also be grounded.

However in practice, there is often a caveat involved with directly grounding the shield at both ends: it creates a low frequency ground loop. In some cases (usually with balanced signals and differential receivers) the receiving end can be grounded with a low inductance ceramic capacitor (0.01 µF to 0.1 µF), still providing high frequency grounding. When you need to ground a shielded twisted pair cable from one end only, the ideal situation is to ground the shield at the driving end and allow the shield to float at the differential receiver.

If the receiver is a single-ended type, so there is no choice but to ground the coaxial cable shield at both ends.

For video applications where grounding at both ends leads to problems there are few tricks worth to try: Using a humbugging transformer on one end of the cable can reduce the problems considerably while still having both cable ends grounded. A normal single-ended video input can be made to work as differential input when you add video isolator to it. For single-ended audio applications audio isolation transformers are worth to try.


  1. Tomi Engdahl says:

    Return path discontinuities and EMI: Understand the relationship
    Minoru Ishikawa -June 11, 2015–Understand-the-relationship?_mc=NL_EDN_EDT_EDN_today_20151221&cid=NL_EDN_EDT_EDN_today_20151221&elq=f16edb6937274a16b479901f7f319cdb&elqCampaignId=26236&elqaid=29978&elqat=1&elqTrackId=287db830eedd4d7d8fab8b58a7ee48fb

    It’s conventional wisdom that a solid, continuous return path provides a better result in electromagnetic compatibility (EMC). This article discusses the relationship between return path discontinuities and EMC.

    A quality signal channel has a nice, uniform trace and a continuous return path from driver to receiver. Disruption to the return path introduces noise, and is typically caused by:

    Changing the reference plane(s) along the signal path
    Discontinuities within the reference plane

    There are two modes of high-frequency current flow

    Normal mode: This is the simpler mode. Current goes along a closed circuit loop, so the total current along the loop becomes zero. The loop is small/narrow enough, so the radiation from the incident current is canceled by the return current.

    Common mode: Noise power goes through both of the traces and, lacking an appropriate, closely spaced plane, something like the enclosure can become the return path. The noise induced by the currents on the signal traces is not canceled by a nearby return current, so strong radiation could occur. This physically larger circuit can act as antenna, so it may cause EMI as well as an EMS (electromagnetic suseptability) issue. The common-mode noise source could be the reference plane discontinuity mentioned in Normal mode.

  2. Tomi Engdahl says:

    Well-grounded serial networking

    Without proper grounding in a RS-422/485 installation, the user becomes vulnerable to common mode voltage transients that can compromise accurate data transfer and sensor measurements and damage equipment.

    The terms “2-wire” and “4-wire” are often used when discussing RS-422/485 installations. These terms can be misleading because they suggest that there is no need for an additional ground wire. That isn’t the case. Without proper grounding, the user becomes vulnerable to common mode voltage transients that can compromise accurate data transfer and sensor measurements and damage equipment.

    RS-422/485 systems can sometimes communicate successfully without the signal ground. This can happen when the nodes are located in very close proximity and the local ground is at the same potential, as in a controlled lab environment. This method is not recommended because if any significant distance separates nodes—and there is no signal ground—lightning strikes and other electrical noise can cause the common mode voltage to rise to levels that can prohibit communications and result in serious damage.

    Connecting signal grounds on both ends while keeping them separated from the earth ground is not sufficient to prevent issues in long RS-422/485 runs, even if you have a good physical earth connection and external surge protection.

  3. Tomi Engdahl says:

    Cable shields

    Of the different choices that are available for grounding a shield braid that encloses a differential pair of signal wires, please consider that the shield braid be grounded only at the signal source, at the input end, and not at the output end.

    As a first thought, and as something that is often advocated, grounding a shield at both ends may result in severe ground loop currents which could adversely impact EMI and isolation properties.

    As a second thought, with the shield grounded only at the output end
    the interground interference signal, Enoise, can induce a differential noise signal between the two outputs E1 and E2 that feed the differential amplifier

    As a third thought, grounding the shield only at the input end averts both the ground loop problem and the time constant mismatch problem.

    Enoise as a common mode signal so that no differential voltage is created between E1 and E2. The A2 differential amplifier is thereby protected from Enoise.


Leave a Comment

Your email address will not be published. Required fields are marked *