Confusion regarding ground loops.

Thread Starter

vas_dub

Joined Feb 1, 2020
3
With regards to Single point ground and Multipoint (multi path) ground. The erroneous current (let's limit the discussion to dc) will flow from a circuit common to the ground.
Other circuits that are referred to this ground may also have such erroneous current being "sunk" into the ground. My question is-
Does any current flow into any of the circuit commons from the ground? Is this erroneous current predictable/computable? What governs the potential (positive or negative) of these circuit commons relative to the ground?
 

KeithWalker

Joined Jul 10, 2017
1,855
Do not confuse earth ground with circuit common. "ground" is used loosely to refer to either. I have illustrated the multipoint (daisy chain) and single point (star) ground connections below. As you can see, in the upper diagram, the voltage at the ground terminal of load 3 will vary by the voltage dropped across the ground wires of loads 1 and 2. If the currents are high or voltage spikes are present, it can be enough to interfere with the operation of the circuits.
Using the multipoint or daisy chain ground separates the ground return paths an eliminates this possibility.
The amount of voltage change at the ground terminals of the loads will vary with the amount of current flowing in the loads and the resistance of the ground return paths. In the daisy chain ground, load 3 ground voltage will vary by the amount of current through loads 1, 2 and 3.
Regards,
Keith
Grounds.jpg
 
Last edited:

BobaMosfet

Joined Jul 1, 2009
1,823
'ground' is nothing more than a common reference point from which to measure all other voltage drops in the circuit. Without this common, agreed-upon reference point, any other voltage value could be anywhere in the negative or positive realm with relation to any other voltage value. A perfect example is making a 5V digital circuit to control a 12V motor. Unless you connect the grounds of the two together- the controller circuit being 5V and the motor circuit being 12V, you don't have a common reference between them. Ground is created by the physics of 'neutralization' of charge. If it's connect together, with nothing in between, it's voltage value will be equal across it. in this case, we simply say that ground is zero volts, and measure everything else in the circuit relative to that zero volts.

Ground loops confuse a lot of people because they are generally not well explained. In truth, ground loops are really simple in concept. One or more ground-loops may be created if one part of a circuit has a different ground-value (0 volts) than another part of the circuit. A way this can happen is if say one circuit connects directly to ground, but a second circuit has a resistor between ground and the circuit. Suddenly you have two circuits each with a different ground level- and they won't operate harmoniously together. Many things can cause this-- for example a bad breadboard. One part of the circuit may have a good ground reference, but another part might have a poor one, so its ground is not equal to the other ground- this is a defect in the breadboard and resistances/impedances of its conductive materials.
 

MaxHeadRoom

Joined Jul 18, 2013
23,371
A ground loop is where a differential exists between different common or earth grounded points in a collection of circuits or parts of an installation.
To offset the possibility of this happening star point ground and/or equi-potential bonding systems are a couple of practices in order to eliminate or reduce these possibilities.
Max.
 

BobaMosfet

Joined Jul 1, 2009
1,823
Do not confuse earth ground with circuit common. "ground" is used loosely to refer to either. I have illustrated the multipoint (daisy chain) and single point (star) ground connections below. As you can see, in the upper diagram, the voltage at the ground terminal of load 3 will vary by the voltage dropped across the ground wires of loads 1 and 2. If the currents are high or voltage spikes are present, it can be enough to interfere with the operation of the circuits.
Using the multipoint or daisy chain ground separates the ground return paths an eliminates this possibility.
The amount of voltage change at the ground terminals of the loads will vary with the amount of current flowing in the loads and the resistance of the ground return paths. In the daisy chain ground, load 3 ground voltage will vary by the amount of current through loads 1, 2 and 3.
Regards,
Keith
View attachment 198067
This doesn't seem right. The very fact that the grounds are connected in parallel in the daisychain example guarantees that the ground value is the same for all loads. The only actual way you determine a ground-loop exists is by measuring the voltage drop between grounds across different grounding points in your circuit.

1580570513648.png

Both your star configuration and your daisychain configuration have the ground conductor tied together- neutralization of charge guarantees it will be the same in both cases.

It is fundamental physics that the only way in which a ground-loop can exist is if one or more varying impedances exist in one or more ground paths compared to other ground paths in the same circuit, other than the load impedance.
 

ronsimpson

Joined Oct 7, 2019
1,328
We would not be having this discussion if "ground" had zero ohms. What ever we call ground has resistance and thus when current flows we see voltage inside ground.

I try to be smart about how ground is connected together or be strong.
Strong = reduce the resistance of ground until the voltages are so small it does not matter.
Smart = example; the picture in post #5.
 

KeithWalker

Joined Jul 10, 2017
1,855
My question is-
Does any current flow into any of the circuit commons from the ground? Is this erroneous current predictable/computable? What governs the potential (positive or negative) of these circuit commons relative to the ground?
Current will only flow through a complete circuit.
If a circuit common is connected to earth ground, no current will flow between common and ground unless the circuit is connected to another potential source which is referenced to ground.
 
That's why I like the terms the grounding conductor, protective ground and earth.

Ground can be a reference or a fault point. Earth is a ground rod.

In some distribution systems protective ground is kept separate from the grounding conductor. One is a reference an done carries fault currents. orange outlets in the US are "isolated ground" receptacles.

All AC outlets should be home run and of this type, but it's not gonna happen. No daisy chaining.
 

MaxHeadRoom

Joined Jul 18, 2013
23,371
In the jurisdiction I came from you ran a Earth wire/conductor, not a ground, Unfortunately the term ground has come to be used for all kinds of circuit commons, earthed or not!.
Actually I prefer to use COMMON instead of ground, then there is less confusion. also unfortunate is the Earth symbol itself is used for common with little regard as to whether it is actually Earth.
In all my schematics I use the power common signal, and only earth where it actually applies.
.Max.
 

Attachments

Ran into an issue of my own doing. I didn;t pay attention. We put together a cheap 6 channel 0-5v readout for engineering units for 6 mass flow controllers. I wanted to convert it to computer control. I made calibration easy and decimal point selection easy.

These MFC's things are located about 15' from the controller. They are powered by +-12V to a common on the card. +5 was used to calibrate and to set stuff via a potentiometer. You saw the value.

Now when you add a computer to the mess. Differential measurements were possible for the flow, but the setpoint depended on the thermal valve currents.

Usually, I would use a current output for the setpoint, but it wasn;t possible.

We purchased a 4-channel pre-made controller in the interest of time. I would have had to add a difference amplifier for each channel.

The purchased readout allowed you to "move" the common to the computer chassis.
 
The 0 V point really should not be current carrying. The key is you know this when you start the design. You keep all of the like commons separated. Analog ground is likely only a reference, There may be a high current ground associated with this common point too. You don't mix them until the one final connection where they tie together.

Digital ground is likely "dirty". Again it's not mixed.

Your not going to run a wire to a star point for every component.

I designed an audio power amp and there really was one star point. The inputs were "isolated" with a 2.2 ohm resistor to ground. The resistor could break the loop.

My preamp had both 0-100 kHz and 0.5-100 Khz outputs. The amp had filters at 0.5 and 40 kHz. Most consumer stuff is capacitively coupled. Pro-audio is differential.
 

WBahn

Joined Mar 31, 2012
26,398
With regards to Single point ground and Multipoint (multi path) ground. The erroneous current (let's limit the discussion to dc) will flow from a circuit common to the ground.
The key concept regarding a ground loop is the notion of a "loop" -- so it's not just that some current flows from point A to point B.

If there are multiple paths between any two points in a circuit (it does not have to be "ground", it's just that this is the most common situation since this node generally has the most paths connecting to it -- the power rail is also a common source of "ground" loops, but any node can potentially be a problem).

If multiple paths exist, then a loop is formed -- current can flow from point A to point B via one path and then back to point B via another. The result is a loop antenna that can either pick up or transmit electromagnetic energy and thus couple the circuit to other things (including other parts of the same circuit), a phenomenon typically known as 'crosstalk'.

In DC circuits, the finite resistances cause crosstalk because changing currents in one branch cause changing voltages along the power and ground rails that other branches are sensitive to. A single-point ground and power distribution strategy is generally preferred to mitigate this. In AC circuits (or circuits sensitive to AC signals), a multi-point ground strategy is generally preferred (the "ground plane" is the the ultimate extreme in this regard). The higher the frequency of the signals under consideration, the more important it is to have a strong multipoint ground since the efficiency of the small loop antennas at coupling energy into and out of the circuit goes up along with the frequency. In practice, you can't have either and, at the same time, you would like to have both (which are mutually exclusive).

A good compromise is to approach your design with the goal of providing a good single-point power and signal distribution layout for DC signals (including power) via star-like topologies and the liberal use of things like blocking capacitors and ferrite beads but then use a multipoint grounding approach for AC signals via stitching capacitors between all of the various power planes and being sure to provide a path for the mirror currents that stay close to the signal path that produces them, thus minimizing the area of the inevitable antenna loops.
 
Last edited:

WBahn

Joined Mar 31, 2012
26,398
We would not be having this discussion if "ground" had zero ohms. What ever we call ground has resistance and thus when current flows we see voltage inside ground.

I try to be smart about how ground is connected together or be strong.
Strong = reduce the resistance of ground until the voltages are so small it does not matter.
Smart = example; the picture in post #5.
Lowering the resistance of the ground and power feeds reduces the effects of power supply crosstalk, but it would actually make the impact of the inevitable loops worse because it allows for greater current in the loop for the same generating signal. This is one of the reasons that resistors are placed in series with signals and that ferrite beads are used -- the increase in resistance (at the frequency of the interfering signal) greatly reduces the resulting current.
 

nsaspook

Joined Aug 27, 2009
8,524
The 0 V point really should not be current carrying. The key is you know this when you start the design. You keep all of the like commons separated. Analog ground is likely only a reference, There may be a high current ground associated with this common point too. You don't mix them until the one final connection where they tie together.

Digital ground is likely "dirty". Again it's not mixed.

Your not going to run a wire to a star point for every component.

I designed an audio power amp and there really was one star point. The inputs were "isolated" with a 2.2 ohm resistor to ground. The resistor could break the loop.

My preamp had both 0-100 kHz and 0.5-100 Khz outputs. The amp had filters at 0.5 and 40 kHz. Most consumer stuff is capacitively coupled. Pro-audio is differential.
Mixed-signal system grounding can be tricky.

There are times when you need to tie analog and digital grounds together at a distant circuit (AGND and DGND pins in mixed-signal devices). (remote spi ADC/DAC chips in a signal-processing system as an example)
http://www.ti.com/lit/an/slyt499/slyt499.pdf



Here, on my Solar Monitor current and analog remote board, back to back Schottky diodes connect the analog and digital ground planes on the PCB.
 
Last edited:

crutschow

Joined Mar 14, 2008
27,445
I worked on a satellite system that had a ground loop problem from numerous preamps that sent their signal along with signal ground to the same processing module, which were also all connected to power ground and the chassis (the chassis had to be power ground).
The loop between the signal ground and the power ground created noise problems there were difficult to solve.

On a new design, I went to differential outputs and inputs for the preamp signals to the processing module, which eliminated the loop and thus it had no noise problem.
 

MisterBill2

Joined Jan 23, 2018
9,028
Ground is 0V (reference).
Really, ground is most accurately described as the power source return connection point. But in reality "ground" is a theoretical point with zero impedance to the supply return connection. Unfortunately all "ground" symbol connections include some resistance value between the point implied by the symbol and the actual supply common point. That resistance also includes other circuit points as indicated by drawing ground symbols, in a fairly random scheme, which results in voltages being coupled between those ground symbol points.
A "ground loop" is an invisible nemesis that couples various ground symbol points together with an unknown and often variable series resistance back to the power supply common return point.
 

ronsimpson

Joined Oct 7, 2019
1,328
Lowering the resistance of the ground and power feeds ……. actually make the impact of the inevitable loops worse because it allows for greater current in the loop
apples and oranges, we see this very differently. (picture of grounds in post #5)
I would never have thought that reducing ground and power resistance would increase ground current measurably. The current is a function of the load not a function of ground resistance. Even if the current went up by 0.1% the greatly reduced resistance would cause the induced voltage to be much smaller. Ohms law give me the idea that as resistance approaches zero the voltage approaches zero and thus less effect.

You must be talking AC not DC.
I have never added resistors and inductors to grounds. I have added inductors to power and data lines to isolate RF noise.
To try to support your point, I have used common mode chokes on power lines and long runs of cables, to keep RF off the wires. In these cases the "grounds" have a area or length greater than 1/4 wave length and I am not fighting ground resistance but ground inductance or impedance at 100s of MHz or ghz. Even in this case lowering ground resistance from 0.1 ohm to 0.01 ohm will not measurably change the current in grounds.
 
Top