Questions About Analog? Step Right Up

If you're new to analog circuit design, you are going to come across terms in technical articles or from coworkers that may hit you somewhere between “I think I've heard of that” and “WTF is this guy talking about?” For the newbies this can be daunting. Note that newbies could mean that you're right out of college or that you've just been moved into new job responsibilities where the boss says you'll be doing some analog circuit design.

I want to make sure that you're not left feeling the way you did when you started a new engineering class and realized you were already two semesters behind. I won't go into a ton of detail on this blog, partially because this has been covered in detail before and partially because we will likely cover it in detail again later.

So let's just take a quick look at some items:

You'll see lots of discussion on ground, grounds, and grounding. “Ground” is a term used to refer to a common circuit connection or bus that runs to many circuit nodes (connections to multiple components). It properly means a connection tied to the earth; but in usage, only occasionally means that. It primarily means a circuit-common node at the lowest potential or closest to zero volts DC and AC. By extension, it means a node to which circulating currents flow on their way back to a voltage or current source. (See (Under)standing Your Ground.)

What makes this one tough to understand is that there are often multiple grounds or circuit commons in a system. There may be analog ground; so the voltage or current source mentioned in the paragraph above is probably an analog signal source or possibly the ±15 Vdc rails. There may be a digital ground and the associated logic signal sources. Parsing more finely, there could be a ground for the +5 Vdc rail, the 3.3 Vdc rail, the 1.8 Vdc, 0.9 Vdc, and so on. There are likely many bypass capacitors from many supply rails tied to these many grounds. There are, as a consequence, many paths for the resulting circulating currents. And the grounds all get tied together somewhere. The devil is in the details with that last bit.

Signals (both digital and analog) going from point A to point B (perhaps within the confines of a circuit board or perhaps traveling several feet or many miles) could be “single-ended.” Or they could be differential (balanced, balanced line, differential signaling; twisted pair). Differential signaling on a twisted pair of wires has the advantage of giving the circuitry at the receiving end the ability to ignore or reject electrical noise induced onto the pair.

The bypass capacitors mentioned above are another source of trouble waiting to happen. Besides the ground connection issues, there is the capacitor's none-ideal behavior. The dielectric material (the stuff between the two capacitor plates) is not perfect. It's supposed to insulate and help hold the charge, but in actually has some leakage resistance, and its dielectric characteristics change with the frequency of the applied voltage. And the plates themselves are not perfect conductors. Instead, they have some parasitic resistance and inductance. But otherwise, they work perfectly.

OK, so no easy answers on these issues just now — I just want to get folks thinking more about the problems. Then I'll tackle some of these issues in small, easy-to-handle bites in future blogs. Post comments or questions to help keep this discussion going.

5 comments on “Questions About Analog? Step Right Up”

1. amrutah
February 6, 2013

@Brad: I agree, “Differential signaling on a twisted pair of wires…”

this helps reduce the common mode noise riding on the signal, but this also helps us by removing the dependance on the Suppy or ground references, the destination module can have different ground plane compared to the source ground plane.

2. SunitaT
February 7, 2013

Brad, Thanks for the post.

In real world, primary concern is considering current flow and current paths. We need to have a different reference ground for high-speed clock signals than the reference ground for low-level audio signal, so that the transient current from switching does not affect the ground reference for tiny signals. The grounds are tied together at one point, so that there is no circuit path for current to travel between the paths, but they are at the same potential voltage.

Real ground paths often have resistance, inductance and capacitance and signals take time to travel from one part of the circuit to another. Different paths wouldn't be necessary if we had an ideal ground(zero volts).

3. jkvasan
February 8, 2013

amrutah,

I also can think of another situation where the ground can be different – the 4- 20 mA loop. When it is converted to voltage at the destination circuit, typically by a resistor of appropriate value, the lower (cold) end of the resistor would normally be the ground at the destination.

4. Brad Albing
February 12, 2013

It's OK (in general) if there are different ground potentials – until you have so much difference that you push the receiving end circuitry's input way beyond its acceptable range. Then it either stops working until the common mode voltage comes back into range and the input circuitry comes out of sturation; or it gets damaged and never recovers.

5. Brad Albing
March 28, 2013

And of course, real systems often must be configured in such a way that the single point star-ground cannnot be accomplished.

This site uses Akismet to reduce spam. Learn how your comment data is processed.