• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Ground Loops 101

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,662
Likes
2,688
Location
Monument, CO
#1
< Author's note: Like many of these articles, this one is about ten years old and could probably be updated. Maybe when I retire... Until then, the basic information should still prove useful. I hope. - Don >

This article is to help show what causes those pesky ground loops many of us have had to deal with at one time or another. Essentially what happens is that a signal is created between two ground points in the system. The gain from source to speaker can be very high and amplifies the “ground” signal, leading to that familiar raspy buzzing sound.

Consider the system shown below. Please note this is not meant to be technically rigorous but to illustrate how ground loops can form. Red lines are positive signal wires, black lines are negative signal wires, and dashed green lines show the current flow. The source (CD player, BD player, turntable, cable box, etc.) is connected to earth ground but the preamp and power amp “float”. This shows a system where only the source has a three-prong plug. The signal current can only flow one way and terminates at the single (source) ground terminal. Ground noise affects only the source, but since all components reference the source, the noise is suppressed. Any ground voltage at the source moves its (+) and (-) outputs equally. With no other ground connections the preamp and power amp follow. No ground loop, no nasty ground loop hum.

1556412051689.png

Now consider the case when all components have three-way plugs or are otherwise grounded. Now there are three ground loops formed: source to preamp, preamp to amp, and source to amp. Any ground leakage or noise current in any component can circulate among these loops. The currents through the resistors cause voltage fluctuations that affect each component. Since some of the signal current now goes into the ground loops, instead of the straight signal path shown above, the voltage induced is seen as signal by the components. They cannot tell if it is “true” signal voltage or voltage caused by ground currents. Since the ground current is often 60 Hz leakage from transformers or unbalanced loads that is the source of the 60 Hz (and harmonic) buzz commonly associated with ground loops. Note that any ground noise can cause this effect, whether 60 Hz power line signals or EMI/RFI.

1556412061237.png

A simple simulation can illustrate the severity of the problem. Vsrc may be a turntable, CD player, cable box, etc. It is not used for these simulations thus its amplitude is zero. Notice the source ground, Gsrc, is connected to earth ground by another signal VGsrc. This represents the induced ground signal and is coupled to the source ground pin Gsrc through resistor R3. VGsrc is a 1 uV, 60 Hz signal representing induced ground noise. The preamp and amplifier (represented by ideal gain stages Epre and Eamp) are also connected to earth ground through resistors R4 and R5. Finally, signal grounds are connected through R1 (source to preamp) and R2 (preamp to power amplifier). The preamp has 60 dB of gain, the amplifier has 30 dB, and the 8-ohm load R6 represents a loudspeaker.

1556412069554.png

This example shows R1, R2, and R3 small (1 u-ohm), shorting the source, preamp, and amplifier grounds to the induced signal source VGsrc. They represent ideal cables and ideal source chassis ground. R4 and R5 are large (10^6 ohms) to float the preamp and power amp (e.g. two-prong power plugs). Because there is no low-impedance path to ground from the preamp and power amp, no ground loop is formed, as shown earlier. You can see this is the plot below; the signals are near 0 V at all points (f = femtovolt = 10^-15 V; u = microvolt = 10^-6 V). The top plot shows the source (blue), preamp (red), and amplifier (green) output signals. The bottom plot shows the ground signals at the source (blue), preamp (red), and amplifier (green). Even with only 1 micro-ohm there is some signal generated by the ground signal but it is very tiny (< 100 fV, or 0.000000000000001 V)

1556412090861.png

Now observe the results when R4 and R5 are also 1 micro-ohm, essentially shorting signal and earth grounds through tiny resistors. This represents the case when all components have three-prong plugs. The resistors act as voltage dividers to couple the signal grounds at each point in the chain. The 1 uVpk ground signal has been amplified to about 10 mVpk (about 6 uW) at the speaker. That would be about 36 dB SPL from a pair of 90 dB/W/m speakers 6’ (2 m) away. Still small, but what if the ground signal was 100 uV? That does not seem like much, but then the voltage across the speaker would be about 1 Vpk or about 62.5 mW. With a pair of 90 dB/W/m speakers the volume (SPL) six feet away would be about 75 dB, definitely audible (I would say loud!), and that is with no room gain. We have a ground loop; the signal ground goes one way, and chassis ground another, providing two places for the signal to flow. Now ground is not really ground so any ground signal is amplified, to our detriment.

1556412104087.png

Clearly small leakage or noise currents in the ground loops can cause very audible problems (as we know). Note cables typically have resistance of milli-ohms, not micro-ohms, and components may be plugged into different circuits so the ground path can be very long. Larger (parasitic) resistor values and additional ground signal sources will generate larger and potentially more complex noise as the signals interact among the components and with the primary signal from the source. Even low-level noise can often be noticed by its absence when the system is modified to eliminate ground loops.

One additional figure will help show why balanced connections are helpful. As you can see, with differential (balanced) circuits, the signal (+) and (-) lines and current flow are independent of the chassis or shield ground. Ground loops can still form but do not impact the signal. However, it is possible for ground currents to couple internal to the component through various “sneak” paths in the circuits. A commonly used technique is to lift (disconnect) the shield at one end of each interconnect, breaking the ground loops, but maintaining the desirable shield to block RFI/EMI.

1556412152409.png

HTH - Don

 
Last edited:

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,662
Likes
2,688
Location
Monument, CO
#2
< @amirm : Amir -- Was answering questions elsewhere and did not see it here. Please move to the technical area at your convenience. >
 

Tks

Active Member
Joined
Apr 1, 2019
Messages
193
Likes
99
Location
NYC
#3
Thanks again Don! Now I can say for sure I understand what a ground loop is with some conviction.

One thing that confused me was near the end:

However, it is possible for ground currents to couple internal to the component through various “sneak” paths in the circuits. A commonly used technique is to lift (disconnect) the shield at one end of each interconnect, breaking the ground loops, but maintaining the desirable shield to block RFI/EMI.
Not a clue what actually has to be done for example >_>
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,662
Likes
2,688
Location
Monument, CO
#4
Thanks again Don! Now I can say for sure I understand what a ground loop is with some conviction.

One thing that confused me was near the end:

Not a clue what actually has to be done for example >_>
For a balanced cable such as an XLR there are two internal signal wires and an outer shield. You can disconnect (lift) the shield connection at one end to prevent any current flow through the shield. It leaves the shield grounded at one end so it still blocks noise from reaching the pair of internal signal wires. Some cables make it easy to lift the ground, and most direct input (DI) boxes, some mixers, and other components include a little switch to allow you to lift the ground. I leave the ground connected at the source end (e.g. preamplifier or mixer) and lift at the load (e.g. power amplifier) end.

HTH - Don
 

Tks

Active Member
Joined
Apr 1, 2019
Messages
193
Likes
99
Location
NYC
#5
Aah now I understand, there are mechanisms to get this done. At first I thought there would be need to pry something >_<
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,206
Likes
1,867
Location
The Neverlands
#6


I would like to add the following:

One can easily substitute 'source' for PC and 'preamp' for USB DAC and power amp for headphone amp.
Or take out the source block and substitute the preamp for PC and the power amp as USB DAC with integrated amp.

PC power supplies and some switchers used for DAC's etc. can create some really annoying audible sounds but almost always it works pretty well when the used equipment is designed properly and proper cabling is used.
 
Last edited:

Music1969

Active Member
Joined
Feb 19, 2018
Messages
263
Likes
60
#7
Hi @DonH56 @solderdude @amirm

I've seen people reporting hearing hum through headphones with just a combination DAC/headphone amp connected to an SMPS (connected to AC power) and connected to headphones - even hearing hum through the headphones with no source connected to the DAC/Amp.

I saw an explanation that this can occur as the ground 'loop' can form through the body of the listener wearing the headphones - is this true?

So the leakage currents go from the SMPS and through the headphone cable and through the body of the listener, to ground?
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,206
Likes
1,867
Location
The Neverlands
#9
Hi @DonH56 @solderdude @amirm

I've seen people reporting hearing hum through headphones with just a combination DAC/headphone amp connected to an SMPS (connected to AC power) and connected to headphones - even hearing hum through the headphones with no source connected to the DAC/Amp.

I saw an explanation that this can occur as the ground 'loop' can form through the body of the listener wearing the headphones - is this true?

So the leakage currents go from the SMPS and through the headphone cable and through the body of the listener, to ground?
In this case, most likely, the DAC/headphone amp was connected via USB to the PC or via another source.
The groundloop is mains -> powersupply PC -> USB -> USB input DAC/Amp -> power supply DAC/amp -> mains.

The reason why some DAC designs can have 'USB related noises' in the analog part has to do with how the USB ground and analog ground are 'coupled' inside the DAC itself.
Most DAC designers use separate ground planes for the digital side and analog side. This is sensible but somewhere these ground planes are tied together somehow.
When this is not coupled 'correctly' or some wire routing after the DAC chip is not done correctly or the powersupply layout on the PCB is suboptimal a small resistance may be present between the digital ground and somewhere in the audio/power supply part (PCB layout, board connectors, wiring) and when a common mode current flows through that resistance a small 'signal' is added to the analog signal.

Breaking that groundloop (optical or inductive) or by using TOSLINK or AES or SPDIF via an input transformer prevents that common mode current from flowing and thus there is no audible garbage.
 
Last edited:

KSTR

Active Member
Joined
Sep 6, 2018
Messages
118
Likes
200
Location
Berlin, Germany
#10
Nice introductory article!

The source (CD player, BD player, turntable, cable box, etc.) is connected to earth ground but the preamp and power amp “float”
At at closer look they don't, and that's why even in this case quite a bit of "balancing" or "leakage" current can flow on the GND wire/shield of unbalanced interconnect. With non-grounded SMPS this current can actually be higher and more nasty than when the SMPS were grounded. The capacitive coupling is large enough to cause significant leakage current and the voltage it couples to isn't exactly mains but rather *rectified* mains (that's why it is much less benign than a dominantly 50/60Hz frequency component).

So we always have to look on the actual coupling mechanism between device reference voltage and mains, it doesn't matter much if there is a true galvanic connection to the mains PE conductor or not.

I don't like the term "ground loop" at all but I understand it is commonly used.

---------------:--------------

A commonly used technique is to lift (disconnect) the shield at one end of each interconnect, breaking the ground loops, but maintaining the desirable shield to block RFI/EMI.
Commonly used but actually not the best way because it taxes the common-mode rejection of the receiver. It is actually better to make the mains feed high-impedance by using mains-filters with significant common-mode inductance/losses including PE, that is, filters with PE chokes.

The point is that we want the signal connections to define/dominate the reference potentials, not the mains connections. Therefore, the mains connections should have much larger common-mode impedance that the signal connections. The signal connections, notably with unbalanced interconnects, can be helped with additional substantial connections between the reference "grounds" of the devices. It is a pity that most devices don't have an explicit "grounding post"....
 

Music1969

Active Member
Joined
Feb 19, 2018
Messages
263
Likes
60
#11
In this case, most likely, the DAC/headphone amp was connected via USB to the PC or via another source.
Nope, as mentioned, USB source not connected to the DAC/Amp... no source connected... as mentioned, only thing connected to DAC/Amp is SMPS (only 5Vdc) and headphones cable... and listener heard hum through headphones...

It was suggested by DAC designer that leakage currents are going through the headphones cable and through the listener's body and this is the cause of hearing hum through headphones... listener also felt little tingle when touching the DAC/Amp...

Plausible?
 
Last edited:

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,206
Likes
1,867
Location
The Neverlands
#12
I think he should use another adapter. The one used would have waaaaayyy too much mains leakage in such a case.
Connecting the USB to a PC that has a proper ground connection should cure the hum and tingling.

I have never seen any headphones make electrical contact with someones body anywhere.
The only possibility would be an all metal in-ear with one of the drivers making electrical contact of sorts.
 
Last edited:

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,662
Likes
2,688
Location
Monument, CO
#13
Hi @DonH56 @solderdude @amirm

I've seen people reporting hearing hum through headphones with just a combination DAC/headphone amp connected to an SMPS (connected to AC power) and connected to headphones - even hearing hum through the headphones with no source connected to the DAC/Amp.

I saw an explanation that this can occur as the ground 'loop' can form through the body of the listener wearing the headphones - is this true?

So the leakage currents go from the SMPS and through the headphone cable and through the body of the listener, to ground?
That is almost certainly noise coupled from the DAC/amp combo and not through the body, ground loop or otherwise. The body is not that great a conductor and most if not all headphones isolate the body from any direct electrical connection anyway.
 

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,662
Likes
2,688
Location
Monument, CO
#14
Nice introductory article!

At at closer look they don't, and that's why even in this case quite a bit of "balancing" or "leakage" current can flow on the GND wire/shield of unbalanced interconnect. With non-grounded SMPS this current can actually be higher and more nasty than when the SMPS were grounded. The capacitive coupling is large enough to cause significant leakage current and the voltage it couples to isn't exactly mains but rather *rectified* mains (that's why it is much less benign than a dominantly 50/60Hz frequency component).

So we always have to look on the actual coupling mechanism between device reference voltage and mains, it doesn't matter much if there is a true galvanic connection to the mains PE conductor or not.

I don't like the term "ground loop" at all but I understand it is commonly used.
It's an example, intended to help explain to folk who might not have seen a description of a ground loop, not intended to cover all the nuances. I intentionally did not include the myriad of other coupling sources that exist in the real world.

Commonly used but actually not the best way because it taxes the common-mode rejection of the receiver. It is actually better to make the mains feed high-impedance by using mains-filters with significant common-mode inductance/losses including PE, that is, filters with PE chokes.

The point is that we want the signal connections to define/dominate the reference potentials, not the mains connections. Therefore, the mains connections should have much larger common-mode impedance that the signal connections. The signal connections, notably with unbalanced interconnects, can be helped with additional substantial connections between the reference "grounds" of the devices. It is a pity that most devices don't have an explicit "grounding post"....
Lifting the shield has been used for decades in professional and consumer installations. The shield is still grounded at one end, provided some isolation but not a full control loop, so I am not sure how it would significantly increase CMRR requirements on the receiver. It is often impractical to add mains filtering and in some cases that hurts the performance in other ways by limiting available current. But again this is getting esoteric for the target audience. It'd be good to write an article that digs deeper for more technical folk, something maybe you could contribute? The loops I deal with, for GHz circuits, are usually much smaller than a typical audio installation... :)
 
Last edited:

DonH56

Major Contributor
Technical Expert
Patreon Donor
Joined
Mar 15, 2016
Messages
2,662
Likes
2,688
Location
Monument, CO
#15


I would like to add the following:

One can easily substitute 'source' for PC and 'preamp' for USB DAC and power amp for headphone amp.
Or take out the source block and substitute the preamp for PC and the power amp as USB DAC with integrated amp.

PC power supplies and some switchers used for DAC's etc. can create some really annoying audible sounds but almost always it works pretty well when the used equipment is designed properly and proper cabling is used.
Yah, I added a note (disclaimer) to the top of the first post. This is a copy of an older thread, where I tried to stay very general with the blocks, but using terms (component names) more meaningful is a good comment.

I should note most of these articles were (and are) written pretty much off-the-cuff. There are undoubtedly much better (and some worse) articles available on the web and elsewhere; these are intended to be an intro for people reading the forum who have some basic knowledge of terms. They tend to leave out a lot of detail and are not always completely rigorous. Hopefully a decent starting point, enough to get the basics across, but those who know this stuff already will be able to find all sorts of holes of omission and where I gloss over some of the nitty-gritty details. My only ask is that you keep that in mind before declaring me completely incompetent (fortunately my reputation in industry does not depend on my quickie internet posts).
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,206
Likes
1,867
Location
The Neverlands
#16
Nothing wrong with your article and still up to date.
Just thought I would add some info that addresses the most common ground loops these days, PC-DAC-Amp combos.
Your pics are quite illuminating in showing the issues.
 

Similar threads

Top Bottom