5.14 Grounding as Applied to Electronic Hardware

In electronic hardware, a conducting plane is often called a ground. This is not a power industry term. In some analog designs (below 100 kHz), there may be several different grounds or reference conductors. These grounds or commons may or may not be directly earthed. In general, these conductors are used in the processing of non-power-related signals. When these reference conductors are inside of hardware, they are not controlled by the National Electrical Code (NEC). In commercial products, this grounding is controlled by UL or its equivalent. The terms signal common or signal reference conductor are preferred over the term ground.

A cell phone may have a conducting plane used as a return path for forward transmission. Obviously, this conductor is not grounded or earthed in any way. This plane is certainly not controlled by the NEC. These “grounds” control the flow of field energy within the device. In some ways, it is unfortunate that we use the same word to describe all of these different conducting surfaces. The semantics problem gets even more complicated when we add adjectives such as signal ground, digital ground, power ground and output ground, and safety ground. These adjectives have little effect on making a facility electrically quiet.

The ground conductor used in a digital circuit is used as a current path (return path) for logic signals. Correctly stated, the common conductor is used to confine and route the fields used to carry logic information and energy to the various logic components. The fields are confined in the space between traces and conducting planes. In a digital circuit, the ground plane is one of the signal conductors. External to hardware, grounds should not be used to carry return signal current. In cases where analog signals are carried over long distances, shielded cables should be used. The signal return may be connected to the shield at the source end of the signal. It is important that interference currents flow in the shield and not in the grounded signal conductor located within the shield enclosure.

In analog designs below 100 kHz, several signal reference conductors are often used. In low frequency analog designs, there is no economic advantage in using a ground plane. Circuits can be made to work with or without this added plane. The reasons relate to the limited bandwidth, and the fact that there are very few analog signals processed at the same time in one circuit.

Analog circuits can sometimes be impacted by the presence of fields that are above the band of interest. In these cases, there may be signal rectification in the circuitry that results in in-band signals. Once the rectification takes place there is no way to eliminate the error. This means that passive filtering may be needed to keep out-of-band signals from reaching a sensitive input circuit.

In analog board layouts, the signal fields and the interfering fields can usually be kept apart without the need for conducting planes or even shielded conductors. In digital work, ground planes are the best way to tightly control the fields associated with logic and energy transport. Fortunately, logic signals have amplitudes greater than 1 V and a 0.1 V error is not usually an issue. In the analog world, it is often necessary to worry about microvolts of interference. This means that the two approaches to interference control are very different even though they both involve fields.

Equipment grounds of any sort should not be used as a signal return conductor. The reason is obvious. Conducting surfaces outside of the hardware must be power grounded (earthed) to satisfy the requirements of the NEC. These conductors are associated with power safety and therefore with power line filters. Since these conductors are exposed to the outside world, fields in the local environment reflect and couple to transmission lines formed by these conductors. These fields include switching transients as well as television, police radio, radar, and cell phone signals to mention a few sources. Reflections of waves at these conducting surfaces imply surface currents. Mixing these interference signals with signals of interest is not a good idea. The approach often used is to place all signal conductors that interconnect hardware inside of shielded cables. The shields carry the interference currents on their outer surface. These shields are often bonded (grounded) to the housings at both terminations. This shielding arrangement can be deficient in noisy environments. In a quiet environment, unshielded ribbon cable can be used. An example might be a connection when the cable is routed inside of a rack enclosure.

In a typical cable connection, some of the signal conductors inside the shield are connected to the circuit commons in both pieces of hardware. This means that there is a parallel path for interference currents. Some of the interference follows the shield and some of it follows the common conductors inside of the shield. If the cable shield is of adequate quality and if the shield terminations at the hardware are properly handled, most of the interference current will stay on the outside of the shield.

Another way to limit the flow of interference currents is to use optical couplers or coupling transformers. The latter solution requires balanced drivers, balanced lines, and balanced receivers. If optical couplers are inside of the receiving hardware then the entering cables can still couple interference into the hardware. Optical coupling must be placed at a shield boundary, if it is to be effective in limiting interference coupling.

It is poor practice to float any circuitry inside of hardware in an attempt to control interference coupling. The best approach is to use balanced or differential circuitry, so that any interference coupling is symmetrical. This way, balanced circuitry can be used to reject the interference as a common-mode signal. Fortunately, digital signals have high noise immunity. In very sensitive low frequency analog work, guard-shield techniques can be used. In this solution, the guard shield grounds at the signal source and does not connect to the amplifying hardware at low frequencies. The shield protects the signal along its entire route, including the path inside the receiving hardware. The input connections must be near the connector. Above 100 kHz, the guard shield is best connected to ground at both ends of its run. This can be handled by placing a series RC connection between the shield and signal common. A typical connection might be a series 100-ohm resistor and a 0.01-μF capacitor.

Conducting planes play the same role inside or outside of hardware. Inside of hardware, they are used to confine signal fields. Here, the plane can be the return path for signal current. Outside of hardware, the conducting plane cannot be used as the return path for signals, as this same conductor carries interference currents. The only role the external conducting surface can have is to limit the coupling of external fields to the cable shields routed on its surface.

Coupling is proportional to the loop area between the cables and the conducting plane. In most applications, openings in the shield coverage (lack of bonding) at the connector create the biggest problem. If the fields associated with shield currents penetrate the hardware at the connector then the presence of ground planes is of little value.

Some of the current on the outside surface of a braided shield can cross over the inside surface of the braid. This means that there are fields coupled to the inside of the cable. The ratio of external current to coupled signal voltage is called a transfer impedance. Fine braid is better than coarse braid. Twin braid is somewhat better and solid-wall shielding is the best. Manufacturers of cable can supply figures on transfer impedance for various cables.