All trademarks are the property of their respective owners.
The term “ground” in the context of electrical engineering took shape with the invention of modern utility electricity and telegraph systems, where it was discovered that the Earth ground could be used to carry the return current.
In modern-day utility electricity standards, an uninterrupted Earth ground wire is required that runs from the source to the loads. This ground is also referred to as equipment ground. The purpose of the equipment ground is to prevent metal parts from being energized in situations such as broken insulation of the main conductors. Sometimes people also refer to the grounded neutral as “ground”. Strictly speaking, a distinction should be made between the two. Neutral completes the loop with “hot” or “line” and is the return path for the load current. Although it is connected to Earth ground at the main distribution panel, it is prohibited to do so at the point of load such as an outlet.
Figure 1-1 is a one-line wiring diagram for a typical household electrical system in the US. It shows the connection from the breaker panel to various loads, as well as how grounding should be connected for the loads. A load may be any common household appliance, as well as electrical outlet.
In the semiconductor industry, the term “ground” is used even more loosely. Ground often simply refers to a common connection without any physical connection to Earth ground whatsoever. In other words, it has nothing to do with Earth ground. After all, electrical standards generally do not require Earth ground for low-voltage DC systems, a domain which a majority of semiconductor circuits fall into.
Ungrounded DC circuits are used on a daily basis. For example, a battery-powered system works just fine without any of the battery terminals being grounded. Simply put, the Earth ground is not needed for a DC circuit to function. However, safety is a concern for high-voltage systems (> 50 V).
In a grounded high-voltage DC system, if a person comes into contact with the power rail, electric shock can occur. While in an isolated DC system, such electrical shock should not occur. However, in reality either side of the power supply may become grounded due to a random first fault. A person will suffer electrical shock when coming into contact with the other side of the power supply in a second fault condition. Because such ground fault cannot be prevented in an isolated system, electrical standards such as NEC require high-voltage DC systems to be installed with proper Earth grounding. Ground fault, overcurrent, and overvoltage protection devices are then installed accordingly to ensure safety.
Figure 2-1 shows one common grounding scheme of a grid-tied photovoltaic (PV) system which typically falls into the high-voltage category. The red lines connecting different components represent current-carrying conductors; the black lines represent uninterrupted grounding conductor or equipment ground conductor. In this grounding scheme, the DC grounding electrode is combined with the AC grounding electrode.
Ground loops form when multiple grounding electrodes exist, and there is a voltage potential between any two. Figure 2-2 shows ground loops can form when multiple grounding electrodes are provided. To prevent ground loops, use a single grounding electrode wherever possible.
While an Earth ground is not absolutely needed, a “ground” or “common” is. An exception are isolated circuits, such as those enabled by transformers and galvanic isolation barriers, where there could be two or more grounds defined by different voltage potentials. However, within each isolated domain, a single common ground still provides reference to all components. Such an isolated system is shown in Figure 2-3, where the primary side ground is separated from the secondary side ground. The two grounds can be defined by different potentials.
Multiple grounds are often defined even in a non-isolated circuit. An example is a typical mixed-signal system where analog ground and digital ground may be defined. To make matters more confusing, multiple grounds are often found for the seemingly identical ground. This type of ground partitioning is often found in applications where the exact same circuitry is cloned multiple times as shown in Figure 2-4. Schematic-wise, each clone might be put on a uniquely named ground which is typically designed to be one or multiple inter-connected ground planes in the PCB.
However it is accomplished, the goal of ground partitioning is to minimize interference and keep noisy circuitry away from the sensitive one. Furthermore, regardless how the ground is partitioned, all grounds will eventually be electrically connected to a single common point. In essence, each of the grounds constitutes an island on which a subsystem operates. It is sometimes not possible to contain all recirculating current within the island. In these situations, the ground plane and traces must be routed such that the current path does not pose interference to other sensitive parts of the system.
Three technologies stand out as integrated, isolated current sensors. They are shunt-based isolation amplifiers or modulators, fluxgate sensors, and in-package Hall-effect sensors.
Shunt-based isolation current sensing employs isolated amplifiers or modulators. Figure 3-1 illustrates an example. The AMC1300 input is similar to a non-isolated current-sense amplifier in that the small differential voltage, which rides on top of a large common-mode voltage, is extracted and amplified. The output is separated from the input circuitry by an isolation barrier that is highly resistant to magnetic interference. The input and output operate in different power domains, each with its own power supply and ground. From the point of view of downstream measuring circuit, the load current is completely isolated, and no return path is needed for the input bias current.
The isolation between the high and low sides are evident in the physical layout. The two sides are separated by an area void of any conductive material. There is no common ground connection between the two sides.
Magnetic sensors work without making physical contact between the sensor IC and the current it is measuring, thanks to their inherent isolation through magnetic fields. A galvanically-isolated barrier is possible that can withstand very high common-mode voltages.
An example of in-package Hall sensor is the TMCS1100 family, shown in Figure 3-2. Within the device, the high-voltage side load current passes through the low-ohmic lead frame path. No external components, isolated supplies, or control signals are required on the high-voltage side. At the low-voltage side, the magnetic field generated by the input current is sensed by a Hall sensor and amplified by a precision signal chain.
Also shown in Figure 3-2 is a recommended layout of the TMCS1100. The layout is optimized for thermal performance while at the same time minimizes stray magnetic field interference. A large creepage area is visible between the high side and low side. There is no common connection between the two sides and they are physically isolated.