I was once asked by a friend/colleague if it is alright to pick any of the grounds in the circuit to make a measurement. I answered it depends on what you are trying to measure, and I reminded that what he calls ground is supposedly a "reference" to his circuit, and will be a reference to his measurement. There will be multiple "ground" points in the circuit, and not all of them are created equal.
What we call ground is ideally zero volts but this is often confusing, because there really can only be one point in the circuit that is truly zero volts. Given a ground plane or ground traces in our PC board, they can actually be at varying potentials due to errors introduced in the circuits by what we call ground loops.
There will be errors unique to each individual ground points. These errors may or may not affect your circuit's performance depending on the applications. We can perhaps get away with our little concept of ground being always zero if the errors don't matter in our circuit, but when it comes to low level signal and precision applications these small errors need to be accounted for.
Ground provides a common current return path to our sources and supplies in the circuits. What we really would like our ground to be is a low impedance node, same as how we would like our supplies to be. Having a low impedance means current may come and go our ground node without introducing an error, and thus maintaining its potential. The key to understanding ground errors and avoiding them is knowing where the current flows in the circuit. What comes out of one power source eventually has to come back to it. Kirchoff's Current Law at work here, and nothing violates it.
Picking a point of reference to your measurement, or to the another part of the circuit can make a big deal of a difference if that reference includes the errors produced by some other part of the circuit that is either returning a huge or dynamically changing current, i.e, an AC.
If all current are dc, the errors will be caused by trace resistances, or maybe contact resistances if present in our PC board. DC current takes the path of least resistance, therefore, the shortest possible route. An AC current in contrast takes the least impedance path. How significant the errors will be depends namely on the amount of current flow, frequency, and the conductor trace parasitics, which can be resistive or inductive.
Consider a 10 mA DC current flowing through a 5 centimeters long copper trace which will typically have a 0.1 ohm. This will create a 1mV voltage drop on the trace. For low level applications, 1 mV might already matter greatly. For a 12-bit Analog to Digital Converter (ADC) with as 5V full scale which has about 1.2mV resolution, that introduces an 83% error, and we haven't even talked about the errors that AC signals can cause because conductors are like small inductors and eats up its share of the voltage pie at high frequencies!
Part II of the series has already been posted on this site.
Recommended Readings and References:
An IC Amplifier's Guide to Coupling, Grounding...by Paul Brokaw
Staying Well Grounded by Hank Zumbahlen
What we call ground is ideally zero volts but this is often confusing, because there really can only be one point in the circuit that is truly zero volts. Given a ground plane or ground traces in our PC board, they can actually be at varying potentials due to errors introduced in the circuits by what we call ground loops.
 |
| Figure 1 Ground Loop |
There will be errors unique to each individual ground points. These errors may or may not affect your circuit's performance depending on the applications. We can perhaps get away with our little concept of ground being always zero if the errors don't matter in our circuit, but when it comes to low level signal and precision applications these small errors need to be accounted for.
Ground provides a common current return path to our sources and supplies in the circuits. What we really would like our ground to be is a low impedance node, same as how we would like our supplies to be. Having a low impedance means current may come and go our ground node without introducing an error, and thus maintaining its potential. The key to understanding ground errors and avoiding them is knowing where the current flows in the circuit. What comes out of one power source eventually has to come back to it. Kirchoff's Current Law at work here, and nothing violates it.
If all current are dc, the errors will be caused by trace resistances, or maybe contact resistances if present in our PC board. DC current takes the path of least resistance, therefore, the shortest possible route. An AC current in contrast takes the least impedance path. How significant the errors will be depends namely on the amount of current flow, frequency, and the conductor trace parasitics, which can be resistive or inductive.
Consider a 10 mA DC current flowing through a 5 centimeters long copper trace which will typically have a 0.1 ohm. This will create a 1mV voltage drop on the trace. For low level applications, 1 mV might already matter greatly. For a 12-bit Analog to Digital Converter (ADC) with as 5V full scale which has about 1.2mV resolution, that introduces an 83% error, and we haven't even talked about the errors that AC signals can cause because conductors are like small inductors and eats up its share of the voltage pie at high frequencies!
Part II of the series has already been posted on this site.
Recommended Readings and References:
An IC Amplifier's Guide to Coupling, Grounding...by Paul Brokaw
Staying Well Grounded by Hank Zumbahlen
Transformer Power Amplifiers Exporter
ReplyDelete