Ground loop on electronic audio systems

[cianfa72]

TL;DR Summary

Clarification about the physics of ground loop causing noise/hum in electronic audio systems.

Hello, I'd like to discuss what is called "ground loop".

Consider the following scenario and call the set "C1 + S + C2" as "the system" -- Ground loop.

As far as I can understand, the "noise/hum" at power grid frequency (50-60Hz) is actually due to electromagnetic processes external to "the system". For instance variable EM fields "external to system" result in a current flowing through the earth protection cable between the two outlets to which C1 and C2 are attached respectively giving rise to a voltage drop between them. Hence, by mean of the system cable connecting C1 and C2 chassis, a current will flow in the ground loop being created (the current labeled as I in the picture above).

The latter is basically the cause of the noise/hum involving the system (e.g. audio system).

Does it make sense ? Thanks.

 

[cianfa72]

To be more precise, I should also include in the picture variable EM fields due e.g. to transformers within C1 and C2 chassis/enclosures coupling with the ground loop.

 

[Rive]

Hello, I'd like to discuss what is called "ground loop".

Are you tracking an issue in a real setup, or are you trying to understand the principle?

For the latter, the description is correct.

For the former, in the particular case of line appliances usually (by my experience) it's not about any external EM field, but about something fishy around grounding and the common line-EMI filters which are providing high impedance connection to PE (chassis) at half-line voltage.

 

[cianfa72]

Are you tracking an issue in a real setup, or are you trying to understand the principle?

No, at the moment no.

For the former, in the particular case of line appliances usually (by my experience) it's not about any external EM field, but about something fishy around grounding and the common line-EMI filters which are providing high impedance connection to PE (chassis) at half-line voltage.

If I were to draw a schematic, does it means it should be some sort of "Thevenin equivalent" driving the PE (connected to the equipment chassis) at about half-line voltage (i.e. 230/2 = 115 Vac @50Hz in Europe) ?

 

[Rive]

Yep. Power supplies here usually starts with these:

Then, whether the PE is properly connected: the filter is properly implemented, in a really symmetrical way: the way the signal grounding is related to chassis - and all these on both sides...
Double insulated appliances, mixing of different type of appliances, devices connected to different line phases...
By my experience external EM field is rare. Sloppiness in implementation - quite frequent...

 

[cianfa72]

Then, whether the PE is properly connected: the filter is properly implemented, in a really symmetrical way: the way the signal grounding is related to chassis - and all these on both sides...

Sorry, can you kindly help me in understanding such a scheme ? As far as I can tell, both coil windings do not make/implement actually a "transformer", they are just elements of an EMI filter including the capacitors Cx and Cy. What do you mean with

whether the PE is properly connected ?

By my experience external EM field is rare. Sloppiness in implementation - quite frequent...

Ok, got it.

 

[Rive]

What do you mean with
whether the PE is properly connected ?

PE is supposed to be connected to N and/or ground (depends on regulation: point of connection depends on regulation/system) with specified (low) resistance.

 

[cianfa72]

PE is supposed to be connected to N and/or ground (depends on regulation: point of connection depends on regulation/system) with specified (low) resistance.

Just to be sure: PE is just the wire/cable that is supposed to be connected to the Neutral wire (N) and/or to the local/domestic earthing system (i.e. by leveraging on grounding rods). The "Earth resistance" should be smaller as much as possible.

In your schematic in post #5, could it happen that PE and PE' points/terminals do not share the same potential?

 

[Rive]

Yes, PE is that specific PE terminal in the (line) connector which is supposed to be ... well, supposed to be the up-to-local-code PE terminal.

On the schematics those points are declared to be on the same potential.

In reality, the PE might be disconnected, may be too high of resistance (either not up to code, but up to code but just still too high to sink all the noise), might be 'infected' with noise/signal from outside...

 

[cianfa72]

On the schematics those points are declared to be on the same potential.

Ok.

In reality, the PE might be disconnected, may be too high of resistance (either not up to code, but up to code but just still too high to sink all the noise), might be 'infected' with noise/signal from outside...

You mean the PE wire/cable could be disconnected from the local earthing system or could have high resistance w.r.t. local ground...

Finally, neglecting the coupling of EM fields with the ground loop, the problem w.r.t. the schematic in the OP, boils down to currents at power grid frequency flowing through the PE wire segment between the two outlets equipment C1 and C2 are attached to.

 

[cianfa72]

By the way, w.r.t. the diagram in the OP, suppose the coax cable's external shield between C1 and C2 equipments had zero resistance. Then I believe, no matter how the earthing system actually works, even a single-ended transmission system would work without any noise/hum since the voltage drop accross that shield connection would vanish, right?

 

[Baluncore]

There is no such thing as zero impedance, or a vanishing voltage. A conductor will always have some inductance, and there will always be some capacitance between devices, even if they are joined by a thick conductor.

To eliminate hum from a ground loop, you must break the loop. That will require a separation between power ground and signal ground. Those should only meet at one point in any device.

There needs to be differential inputs, or transformer coupling, to eliminate closed loops in signal grounds.

 

[Rive]

Finally, neglecting the coupling of EM fields with the ground loop

This is my experience. Should somebody live/work in a different environment - 50/60Hz transformer based equipment (retro or special) or industrial environment with lot of induction motors and such, for example - then the experience may be very different...

 

[cianfa72]

There is no such thing as zero impedance, or a vanishing voltage.

Yes, of course. That was just a theoretical assumption just to understand things...

A conductor will always have some inductance, and there will always be some capacitance between devices, even if they are joined by a thick conductor.

Sorry, if you were to draw a diagram, where would you place the capacitors representing the capacitance between connected devices/equipment C1 and C2?

There needs to be differential inputs, or transformer coupling, to eliminate closed loops in signal grounds.

Ok, you mean use for instance transformers to get galvanic isolation to eliminate closed loops in signal ground.

 

[DaveE]

Honestly, there are way too many reasons that currents may be flowing on "ground" wires/traces to have a simple answer. Yes there are common scenarios, but you are better off, IMO, to just understand that stray currents will make voltage drops in conductors. Then in a good analog design this is analyzed and taken into account.

I have many, many times designed instrumentation in relatively complex systems with many "ground" designations in my schematics. There may be a Chassis ground (PE), a Power Supply ground, a Digital ground, an Analog ground, Shield grounds etc. As @Baluncore said the connections between each of these must be handled with care, usually only at one point. The separate naming allows you to communicate to other engineers looking at the design as well as to control routing on PCBs, terminal blocks, etc. Then the connection between them is explicitly identified and controlled, like a jumper on a PCB.

Their usage also needs to be carefully considered. For example, things like a uC with ADCs inside require thought.

There is an identical concern with voltage sources/references. You simply must control your analog signal paths if you need accuracy or low noise.

 

[Averagesupernova]

Truth is that there normally can be ground currents on the shield of coax cables that go unnoticed. The nature of a concentric shield around a single conductor results in the magnetic field caused by 60/50 hertz currents in the shield effectively canceled with regard to the center conductor. It is not perfect. There is a reason why shielded balanced transmission lines exist.
-
It certainly is good practice in a design to control where all currents go. I have seen way too many circuit boards laid out where there are multiple paths. Parallel paths are often bad news. Especially when one of the paths involves metal parts of a chassis which I have seen many many times.

 

[DaveE]

Truth is that there normally can be ground currents on the shield of coax cables that go unnoticed. The nature of a concentric shield around a single conductor results in the magnetic field caused by 60/50 hertz currents in the shield effectively canceled with regard to the center conductor. It is not perfect. There is a reason why shielded balanced transmission lines exist.

Yes, 100% agree. The grounding of shields in cables can be really complicated in practice. In more complex cables, I have sometimes just designed to accommodate a few choices with an "I'll figure it out later in testing" approach. Sometimes there are conflicting issues that just aren't easily predicted.

 

[Baluncore]

Sorry, if you were to draw a diagram, where would you place the capacitors representing the capacitance between connected devices/equipment C1 and C2?

When you replace the conductors with inductor symbols, you also connect capacitors between the chassis of all devices and of all grounds.

 

[cianfa72]

When you replace the conductors with inductor symbols, you also connect capacitors between the chassis of all devices and of all grounds.

With reference to the diagram in the OP, I came with the following schematic

L1, L2, L3 and L4 coils account for the inductance of the conductors involved, while capacitors C1 and C2 account for the capacitance between equipment E1 and E2's chassis and ground. In the OP diagram there are two power outlets connected through PE conductors to the main ground/earthing system.

Do you think it makes sense ?

 

[Baluncore]

Do you think it makes sense ?

There should also be a capacitor between E1 and E2.

 

[cianfa72]

There should also be a capacitor between E1 and E2.

Added capacitor C3.

Now from this schematic, one should be able to analyze the impact of ground loop on the single-ended transmission system depicted.

 

[Baluncore]

Now from this schematic, one should be able to analyze the impact of ground loop on the single-ended transmission system depicted.

Yes, you can, if you place values on all the components.

Notice that there is actually no L1 in a coaxial cable. The transmission line is supported between the centre conductor and the inside of the braid. The current that flows on the outside of the braid is independent of the internal signal.

 

[cianfa72]

Notice that there is actually no L1 in a coaxial cable. The transmission line is supported between the centre conductor and the inside of the braid.

Basically your point is as follows: L1 accounts for the self-inductance of the coax cable's center conductor, i.e. it accounts for the self-induced emf. However the external braid acts as a shield so there is no coupling of the centre conductor's self-generated EM field with itself, hence L1 inductance can be actually neglected.

Does it make sense?

P.s. assuming the two equipment 1 mt apart and 5 mt the length of each outlet's PE conductor to the common ground connection, which could be plausible values for L2, L3,L4 and C1,C2,C3 ?

 

[Rive]

Do you think it makes sense ?

I would add two caps between the equipments and the related outlet 'L' point (these may not be at the same phase!)

As it is now your schematics has no 'entry' point of noise.

Ps.: if you have proper equipment you may try to actually measure some details. Without line connection, of course.

 

[cianfa72]

I would add two caps between the equipments and the related outlet 'L' point (these may not be at the same phase!)

Sorry, do you mean power outlet's Line (L) points could represent different phases of the three-phase electrical power distribution system the outlets are attached to ?

As it is now your schematics has no 'entry' point of noise.

Oh yes, we can add some voltage or current source to model that....

 

[Rive]

Sorry, do you mean power outlet's Line (L) points could represent different phases of the three-phase electrical power distribution system the outlets are attached to ?

Yes.

 

[sophiecentaur]

There needs to be differential inputs, or transformer coupling, to eliminate closed loops in signal grounds.

It's like an adventure game. You can't be sure of what's going on without reliable tests - you often have just the results of making changes. It's often a choice between fighting induced E fields and H fields. Disconnecting the 'ground' connection from one end can sometimes work wonders.

An optical link can eliminate some earth loops too. Many devices are supplied with optical connectors; useful if you are using individual modules.

 

[Baluncore]

It's like an adventure game. You can't be sure of what's going on without reliable tests - you often have just the results of making changes.

This was not a true hum loop, but was a more worrying situation. I noticed a "hot to the touch" power cable on a newly installed faulty PC, and traced it to an unexpected and interesting loop.

The loop involved the Protective Earth in the wall outlet to the chassis of the monitor and the case of the PC and power supply. The video cable screen connected the monitor chassis to the PC chassis. During delivery, a standing axial resistor in the PC supply, had bent sideways to intermittently touch the internal chassis wall of the supply. That placed the +5 volt supply across the PE of the two power cables, and the screen of the video cable, hence the heat. There was less metal in the PE of the IEC power cables than in the screening of the video cable that remained cool. There was no low-ohms link in the PC supply between the 0 volt power ground and the PE chassis, that was established by the video cable screen, in what appeared at the time to be an attempt to avoid a video hum loop.

 

[sophiecentaur]

an attempt to avoid a video hum loop

At least that's not a modern problem; hum on a digital feed would need to be massive to affect the error rate.
Problems of synchronising video sources were very common in 'old TV' and were only sorted out with the advent of the analogue field store which at least would keep hum bars stationary