The myth: Digital (high-frequency spectra) circuit boards can be isolated from chassis structures.
The reality: Digital (and all high-speed, high-frequency spectra) circuit boards are always coupled to chassis!
Of all the “myths” and “realities” in the description of systems and system-product implementation, possibly the most controversial (and least understood) is founded on the topic of the isolation versus grounding of circuit boards with respect to conductive chassis structures. Once the topic is opened, the controversy quickly moves from grounding as single-point versus multi-point. Since the concept originates with the field-transfer (coupling) relationships between circuit board and chassis, that is an appropriate place to start examining the controversies and their related processes.
When a circuit board is positioned above a conductive chassis structure, an immediate form of coupling occurs: distributed capacitance. The magnitude of the capacitance is determined by the surface area of the circuit board over the chassis plane and the distance of separation between the two structures. Since distributed capacitance is simply the dimensional relationships between these structures (board and chassis), one fact is inescapable: the boards are coupled to the chassis! This observation moves the discussion quickly from the fact of coupling to the magnitude of coupling and the performance significance of that magnitude. Due at the minimum to distributed capacitance, another inescapable fact appears: at higher frequency spectra, there is no such thing as a circuit board that is truly isolated from conductive chassis structures! Note that distributed capacitance values as small as 10 pico farads will yield coupling transfers in the region of a few tens of ohms in the spectra from approximately 300 MHz and higher.
Viewing the details of circuit board construction, a sequence of patterned layout inductance is established by layout details, including routing patterns and “Swiss-cheese” through the ground and power planes. The routing patterns and “Swiss-cheese” effect set up a form of distributed inductance. When common-mode electrical currents are impressed across these holes and patterns, electrical potentials occur as a sequence of losses. These potentials are the beginning of electromagnetic waves and their related impedances. In effect, a distributed transmission line process is immediately formed with inductance patterned in the circuit board and distributed capacitance between the board and any conductive chassis plane. All transmission lines, distributed or intentional, are characterized with some value of impedance.
If electrical potentials are formed across the board and if these ‘find’ the impedance of distributed capacitance, electrical currents will be displaced through the impedance set up by distributed capacitance. Note that the currents formed are developed across relatively low impedance values of distributed capacitance. Since the impedance of any electromagnetic wave at any point is equal to the value of the E-field intensity (in volts/meter) divided by the current of H-field intensity (in amperes/meter), the coupling between the circuit board and the chassis immediately becomes a spectral electromagnetic field by nature and structure.
When viewed as an impedance formed through a distributed transmission line coupled process or considered as an electromagnetic wave transfer function, circuit boards are indeed coupled to chassis planes. In terms of approximate magnitude for consideration, note that near fields (close to sources at, for example, 0.5 centimeters) with magnetic field dominance (described by sources that have low impedance and high current) will propagate in impedances below 40 ohms at frequencies below 1 GHz. These same conditions will exhibit electromagnetic field transfers below 20 ohms at frequencies below approximately 500 MHz. In light of this observation, the discussion can expand beyond the fact of coupling to focus instead on the design consequences of that coupling.
