"Ground" has become one of those terms that gets used so often that it means everything and nothing at the same time, but it's a very important issue whenever you have an AC line in the system -- and sometimes even when you don't.
"Ground" is one of those words that gets tossed around so much in our profession that it has an almost mystical-like aura. It seems that whatever your design topology, you need a good, solid ground to solve your problems -- except, that is, unless you really don’t want a ground, and instead you need isolation. Not so simple, is it?
By "ground" here, I am not referring to a signal or DC rail "common" connection (return path), which is often mistakenly called out as "ground." (See: Can we find common ground about "common" and "ground"?.) I am referring to a true ground that is associated with AC power lines with a direct connection to Earth. It functions as a near-infinite source and sink for excess electrons, which also can result in a near-zero voltage between chassis and another similar point -- more precisely called the "potential difference." Where would we be without ground?
The answer is: it depends. The dilemma is that sometimes you want to make sure everything is grounded, via solid, low-impedance connections to the Earth, for either user safety, system performance, or both. But then there are other times where that same Earth ground is something you actually don’t want, in order to achieve those same objectives.
Think about a basic power tool: Are you going for a product with two-wire AC-plug and using a double-insulated case, or one with a three-wire AC plug and a grounded metal case? (And where is the "ground" on that airplane, anyway?)
The ground issue becomes complicated when you are trying to isolate a circuit that eventually connects to an AC line and its associated ground. The easiest way to isolate your system from the line ground is via a transformer, which is a reliable, well-understood, and well-behaved component. But there are times when you need to instead isolate your AC-powered circuit in its DC and electronics subsection, and that brings new options and new issues. It also means you have to be careful to isolate not just power, but also any signals, as a non-isolated signal path can negate the benefits of an isolated AC line.
If you think the issue of grounds and grounding is straightforward, think again. There's no shortage of papers and guidelines on the subject, as well as the various electrical codes of different countries and even local municipalities to consider. The recent posting "When Good Grounds Turn Bad -- Isolate!" by Thomas Kugelstadt, a Senior Applications Engineer and prolific author at Texas Instruments, shows there is always room for more insight and understanding.
Back in the days when there were products that used only line-AC internally -- say, a heater connected to the power line -- it was easier to decide on your grounding approach. Now, even basic AC-connected products have internal electronics, AC/DC supplies, and even connectivity ports, so new challenges arise. Unless you design exclusively in the world of battery-powered, handheld, non-chargeable devices -- and yes, there are lots of those -- sooner or later, the issue of grounding will come up with questions such as "Do we ground?" "If so, where do we ground?" "How do we ground?" "Maybe really want isolation, instead?"
Grounds and grounding may seem like mundane topics for low-voltage designers, but eventually, many of these low-voltage rails work their way up to an AC line and the real world. It never hurts to understand how Earth grounds affect your product design, performance, and safety. Perhaps you need to take a break from the world of low-levels signals, fast slew rates, and other attributes of today's circuits, and review those basics -- and reexamine them in a new way.
What's your experience with grounds, grounding, product safety, and product performance? Have you ever been stung, both literally and figuratively, by AC-ground problems?