What is a ground loop?

[cupoftea]

Hi,
Wiki says a ground loop is...

in an electrical system, a ground loop or earth loop occurs when two points of a circuit are intended to have the same ground reference potential but instead have a different potential between them.

...Surely this is not central to what a ground loop is all about?
A ground loop is literally that....a loop of conductor that is all of the ground net.
The problem with it is that it unfortunately acts like a loop antenna and receives noise and transmits it out..

Surely a ground loop is what you get when you have an offline SMPS with an earth input..... The earth is also taken across the isolation barrier where it connects to secondary ground...then someone connects something else to the secondary side, which also is connected to mains earth...then you obviously get an enornous earth ground loop going back through the wall socket , and through the building earth cableing. This is surely the largest and most awful (for EMC) exampel of a ground loop.

The other example is when we had 34 Vicor SMPS modules in parallel.......it obviously wasnt possible to have a single huge ground plane under all 34 of them....so we ended up with enormous ground loops...and had to put common mode chokes in each module's output, so as to break up the ground loops.

 

[betwixt]

The term is more generic than your specific example. It is usually used when different pieces of equipment are linked together and some difference in their individual ground potentials is present. It causes currents to flow between each equipment's 0V reference points and hence modulate the apparent signal they see relative to each other.

What you are describing is a ground topology within a single piece of equipment and signals radiated because of the inductance and resistance (effectively the impedance) between internal ground points. The EMC problem occurs when the impedance of a path is high enough that a significant voltage at the unwanted frequency is dropped across it. The resulting magnetic field and to some degree electrical field spreads as interference.

To design against internal ground differences you need to understand the paths current flows and the impedances along the way. Simply adding capacitors and chokes everywhere isn't the right approach and often NOT breaking ground paths is a better option.

Brian.

 

[cupoftea]

Thanks, so the "loop antenna" theory doesnt apply with ground loops?

I mean, surely , loops of current, must always be minimised in area to reduce the "loop antenna" effect?

 

[betwixt]

A loop antenna is usually a resonant device, optimized to have best radiation efficiency. However, any connection between two points that carries current will radiate to some degree. The trick to minimizing EMC is to understand where the currents flow, not to place barriers everywhere to block them.

I appreciate you work with power supplies and those by nature work at low frequencies but I think you would benefit by studying RF techniques. The same effects occur, at higher frequencies the interconnection impedances are higher so signal losses along them increase, at the higher currents you encounter the same happens simply because of larger voltage drops. RF people have learned how to shield and mitigate, it would help you to see the methods they utilize.

Brian.

 

[cupoftea]

However, any connection between two points that carries current will radiate to some degree. The trick to minimizing EMC is to understand where the currents flow, not to place barriers everywhere to block them.

Thanks, but surely a "ground loop" can occur between two adjacent mains connected devices.....and the loop occurs in the earth ground...there is no circuit current flow in this ground, but surely that is the "ground loop" which the term refers to?...there will be of course, RF currents injected into this loop of current...and they may than also re-irradiate, and cause problems to circuitry.

However, any connection between two points that carries current will radiate to some degree.

Thanks, AYK, it is high di/dt which causes radiation of EM fields. AYK, Constant DC current wont radiate.

The trick to minimizing EMC is to understand where the currents flow, not to place barriers everywhere to block them.

Thanks, yes you are right, in an SMPS , its obvious where the loops are that carry the high switched current. Though literally any circuit conductor loop can receive noise (radiation) from some noise source and arise problems.......this is what "ground loops" are all about?

The term is more generic than your specific example. It is usually used when different pieces of equipment are linked together and some difference in their individual ground potentials is present.

Thanks, i appreciate thats a problem, but surely thats just called "differing ground potentials".......i cant find on the web any reference that the "differing ground potentials" problem is actually known as a "ground loop problem".

 

[SWRmeter]

Differing ground potentials - especially due to high(ish) currents in conductors normally considered as 'ground' and mistakenly being at exactly the same potential - are exactly what ground loops are all about. As betwixt has succinctly explained.

The EM effects and other artefacts arise due to this. The rest is just nomenclature.

 

[cupoftea]

Thanks, well, the following agrees with you....

Ground Loop

A group loop is an unintentionally induced feedback loop caused by two or more circuits sharing a common electrical ground. Click here to find out more

www.sunpower-uk.com

So please may i ask, what is a ground loop where there is no circuit current flowing in it?....(obviously it will have induced RF currents flowing in it, because its a conductive loop.....a "loop antenna" in effect)....its just a very wide area loop of ground conductor...may i ask what that is called?.....i take it that it is not called a "ground loop"?

Also, for such a large loop of ground net...would you say it is a problem?

Differing ground potentials - especially due to high(ish) currents in conductors normally considered as 'ground' and mistakenly being at exactly the same potential - are exactly what ground loops are all about. As betwixt has succinctly explained.

The EM effects and other artefacts arise due to this. The rest is just nomenclature.

Thanks, though AYK, if the currents that flow as a result of "differing ground potentials" are just pure DC currents, then there is no associated EM effects.....so what do you call the "ground loop" in that case.?

Also, may i ask, what is the simplest, real world example of a "ground loop"?

___---___---__
And the loop of current that i describe in the top post...what is that called?...presumably NOT a "ground loop".
Here it is again, if i may..
QUOTE>>>
The other example is when we had 34 Vicor SMPS modules in parallel.......it obviously wasnt possible to have a single huge ground plane under all 34 of them....so we ended up with enormous ground loops...and had to put common mode chokes in each module's output, so as to break up the ground loops.
>>>>>>>UNQUOTE
.....what is the ground loop called?...as you can appreciate, it may not involve any general potential differences throughout the loop.
___----____----___

So "when is a loop of ground conductor called a "ground loop", and when is it not?

I ask because when circuits are noisy and malfunctioning, someone always comes along, and says, have you got problems with "ground loops".

What do they mean?
_________-------_________
Also, is a ground loop more of a problem with EMC..?...or is it mainly a problem with circuit noise and resultant malfunction?

 

[SWRmeter]

Thanks, well, the following agrees with you....

Ground Loop

A group loop is an unintentionally induced feedback loop caused by two or more circuits sharing a common electrical ground. Click here to find out more

www.sunpower-uk.com

So please may i ask, what is a ground loop where there is no circuit current flowing in it?....(obviously it will have induced RF currents flowing in it, because its a conductive loop.....a "loop antenna" in effect)....its just a very wide area loop of ground conductor...may i ask what that is called?.....i take it that it is not called a "ground loop"?

That's right, its not called a ground loop because there's no current flowing in it causing any trouble.
As far as the RF currents are concerned, you'd call it an inductor or an antenna.

Also, for such a large loop of ground net...would you say it is a problem?

Only if it leads to undesirable results. In which case you need to make THAT "loop" as small as possible.

Thanks, though AYK, if the currents that flow as a result of "differing ground potentials" are just pure DC currents, then there is no associated EM effects.....so what do you call the "ground loop" in that case.?

That's still called a ground loop. It may, or may not, cause undesirable results. Depends on your circuit and desirable specs and functionality. There's very few cases I can thing of which only have DC currents in their entirety, and where potential differences did not matter

Also, may i ask, what is the simplest, real world example of a "ground loop"?

If the PS for an audio amp output stage is the same as that for an input pre-amp stage, then you definitely need to ensure all ground points and Vcc points are physically connected at the SAME SINGLE "point" - to avoid ground loops.

___---___---__
And the loop of current that i describe in the top post...what is that called?...presumably NOT a "ground loop".
Here it is again, if i may..
QUOTE>>>
The other example is when we had 34 Vicor SMPS modules in parallel.......it obviously wasnt possible to have a single huge ground plane under all 34 of them....so we ended up with enormous ground loops...and had to put common mode chokes in each module's output, so as to break up the ground loops.
>>>>>>>UNQUOTE
.....what is the ground loop called?...as you can appreciate, it may not involve any general potential differences throughout the loop.
___----____----___

So "when is a loop of ground conductor called a "ground loop", and when is it not?

I ask because when circuits are noisy and malfunctioning, someone always comes along, and says, have you got problems with "ground loops".

What do they mean?
_________-------_________
Also, is a ground loop more of a problem with EMC..?...or is it mainly a problem with circuit noise and resultant malfunction?

 

[danadakk]

Ref material, attached.


Regards, Dana.

 

[cupoftea]

Thanks Dana,
Thats a great doc, but does not specifically address the problem of ground loops.

It reinforces the main points of PCB layout , ie........

1…..Avoid routing Switching Power current loops (currents with high di/dt’s) along lengths of ground trace that would disturb analog control circuitry through “ground bounce”. (voltage ‘bouncing’ in the ground due to V=L.di/dt, where L is the trace stray inductance)
…this is one of the main principles, and as you know, switching current loops are unfortunately also found in the FET gate drive current loops, as well in the general power current loops (Rectifier Loop and FET loop)


I am sure you are aware of the other principles, eg,
a…. reduce the area inside current loops by keeping “go” and “return” close together as much as possible.
b…. try and keep high dv/dt traces, eg the switching node , away from sensitive things like the inputs to the PWM comparator insIde the PWM controller IC.
c…..be aware of high dv/dt nodes capacitively coupling noise into adjacent metal areas, eg, the enclosure etc.
d….Be aware of the track thickness’s needed to carry high currents without overheating the PCB and the components on it.
e…..Be aware, with high currents, that if you don’t have much room for decent track thickness, then the voltage drop on eg the ground track will be significant, which as you know , isn’t desirable.


...but none of this address's the specific situation of "ground loops"

The attached is a de facto reference on PCB layout for EMC, but does not address "ground loops"

 

[crutschow]

I think basically a ground loop is where the is more than one connection between to grounds, which can form a loop the picks up or radiates EM waves.
For example, if an audio setup between two devices has a ground connection between them from the audio cable, and also each has a ground connection to the main's safety ground, then you have a nice loop which will pick up the AC EM mains signal, generating a circulating current and thus a differential voltage between the two grounds, causing hum in the audio output.
They sell isolating audio transformers to connect in the audio line and break that loop.

 

[cupoftea]

I think basically a ground loop is where there is more than one connection between two grounds, which can form a loop that picks up or radiates EM waves.

Thankyou...this is indeed what i always thought a ground loop is.

For example, if an audio setup between two devices has a ground connection between them from the audio cable, and also each has a ground connection to the main's safety ground, then you have a nice loop which will pick up the AC EM mains signal, generating a circulating current and thus a differential voltage between the two grounds, causing hum in the audio output.
They sell isolating audio transformers to connect in the audio line and break that loop.

...Thankyou , this sounds like a great example of a ground loop...in this case, an "Earth ground loop".
Eg an electric guitar would have a cable to the amplifier, and both electric guitar and amplifier would have a mains cable connected with earth. The cable between guitar and amp would have earth ground as the ground...and so you have a large loop of earth ground.....a "ground loop".

 

[dick_freebird]

Current must flow around a loop. There will be ohmic voltage variation
around that loop. If that loop is considered to be (ideal) ground, that
considered opinion is false to some degree. When fundamental
assumptions (like your reference plane being worthy of the trust you
place in it) are wrong, much becomes unknown.

Engineers don't like unknown "termites in the carpentry".

 

[barry]

If there's no current flowing in it, RF or otherwise, it's not a problem, I don't care how big it is.

 

[KlausST]

Hi,

If there's no current flowing in it, RF or otherwise, it's not a problem, I don't care how big it is.

Yes, "if". But how likely is this?

No current in a ground loop?
How can it be?
--> A circuit without power supply in a faraday cage?

My idea: if there is a power supply, then there will be current, and it's rather likely that there is GND current.
And a loop acts as an antenna, so as soon as there are some magnetic fields (HF) around ... there will be loop current.

Klaus

 

[cupoftea]

Page 6 of the following seems to show what ground loops are....
file:///C:/SMPS%20Course_SMALL_H67/Course%20Folders_SMALL/comms/RS485%20elec%20sigs%20_slla272d.pdf

 

[FvM]

Links to your local disk can't be resolved...

You probably mean https://www.ti.com/lit/an/slla272d/slla272d.pdf

 

[cupoftea]

Woops, sorry ...Many Thanks FvM!

BTW, as per the question in #7 above, is there any such thing as a loop of conductor that is ground net...but it is NOT referred to as a "ground loop"....and why not?

 

[Calm down]

Hello, I think it may be a better choice to connect the ground wire with equal impedance at the same place.

 

[dick_freebird]

The "ground conductor" as you say, is a -portion- of the ground _current_ loop.