Ground loops - measurements and solutions

[pma]

Groundloops – measurements and solutions

When 2 audio components built in class I are interconnected with a single-ended link cable, troubles may be expected almost for sure and they are rising with length of interconnecting cable and distance of the components. The result is a hum and buzz from speakers.

How does it happen? The following image shows class I signal source and class I amplifier interconnected by a simple single-ended link cable. Both components are supplied from 2-wire mains network (L + PEN), which is the worst option. The groundloop is shown in red. Loop current flows in the groundloop and it is sourced by magnetic induction B (from EM field sources) and by a voltage drop across resistance of the PEN wire, due to supply currents. This creates a low impedance voltage source, which creates a voltage drop across Rsh cable shielding resistance and this voltage drop is added to the audio signal, resulting in hum and buzz.

We need to know how big is the ground interference voltage. For this purpose, we disconnect the loop at one point and use a battery operated instrument or the one supplied through an isolation transformer.

Next image shows the real measurement in the groundloop. Please note that the voltage swing is almost 1V. Loop current was about 63mA.

Next, we shall see how different signal component designs would affect the resulting hum and buzz. In all cases, the signal source will be in class I. On the receiving side will be a class I preamplifier with both single-ended and balanced inputs. Preamplifier output is set to drive the power amplifier to its full power, full swing. Al the spectra are ref 0.83V, 0dB = 0.83V.

First measurement is with source balanced output, 12m of balanced twisted pair cable and balanced input of the preamp. We can see that the preamp with balanced input almost completely eliminates any hum/buzz voltages.

Now the same measurement, but single-ended input of the same preamp is used. This is terrible, isn't it.

The last one is a balanced input again, but the XLR pin 1 is wired improperly. Not to chassis, but it goes through a piece of wire to PCB ground. We can see worse result than in the 1st case.

We might think that if the preamp on the right side would be in class II, the situation would be better. Yes, but not completely. Some capacitive current still flows in the loop and worsens S/N in the single-ended mode.

 

[pkane]

Groundloops – measurements and solutions

When 2 audio components built in class I are interconnected with a single-ended link cable, troubles may be expected almost for sure and they are rising with length of interconnecting cable and distance of the components. The result is a hum and buzz from speakers.

How does it happen? The following image shows class I signal source and class I amplifier interconnected by a simple single-ended link cable. Both components are supplied from 2-wire mains network (L + PEN), which is the worst option. The groundloop is shown in red. Loop current flows in the groundloop and it is sourced by magnetic induction B (from EM field sources) and by a voltage drop across resistance of the PEN wire, due to supply currents. This creates a low impedance voltage source, which creates a voltage drop across Rsh cable shielding resistance and this voltage drop is added to the audio signal, resulting in hum and buzz.
View attachment 46096
We need to know how big is the ground interference voltage. For this purpose, we disconnect the loop at one point and use a battery operated instrument or the one supplied through an isolation transformer.
View attachment 46097
Next image shows the real measurement in the groundloop. Please note that the voltage swing is almost 1V. Loop current was about 63mA.
View attachment 46098
Next, we shall see how different signal component designs would affect the resulting hum and buzz. In all cases, the signal source will be in class I. On the receiving side will be a class I preamplifier with both single-ended and balanced inputs. Preamplifier output is set to drive the power amplifier to its full power, full swing. Al the spectra are ref 0.83V, 0dB = 0.83V.

First measurement is with source balanced output, 12m of balanced twisted pair cable and balanced input of the preamp. We can see that the preamp with balanced input almost completely eliminates any hum/buzz voltages.
View attachment 46099
Now the same measurement, but single-ended input of the same preamp is used. This is terrible, isn't it.
View attachment 46100
The last one is a balanced input again, but the XLR pin 1 is wired improperly. Not to chassis, but it goes through a piece of wire to PCB ground. We can see worse result than in the 1st case.
View attachment 46101

We might think that if the preamp on the right side would be in class II, the situation would be better. Yes, but not completely. Some capacitive current still flows in the loop and worsens S/N in the single-ended mode.

That's helpful, thank you. So what are the solutions? Assuming I have a device, say a DAC or a preamp forming a ground loop with an amp, and I can only use single-ended interconnects, what are my options besides a full isolation transformer?

 

[pma]

That's helpful, thank you. So what are the solutions? Assuming I have a device, say a DAC or a preamp forming a ground loop with an amp, and I can only use single-ended interconnects, what are my options besides a full isolation transformer?

At least one of the instruments should be in class II, if there are only two used. If there are more than 2, like CD (PC), preamp and power amp, only one should be in class I. The other helpful cure is a power supply isolation transformer, safety built. This may be an excellent option and I always use it for oscilloscope measurements not to bring the ground into the measured setup.

 

[Tks]

power supply isolation transformer, safety built.

I'm guessing Amazon searching for one isn't what qualifies as "safety built"?

 

[amirm]

I'm guessing Amazon searching for one isn't what qualifies as "safety built"?

None of the ones they sell as isolation transformers are actually isolated! Safety ground is shared at both ends...

 

[amirm]

FYI the unbalanced inputs on my Audio Precision Analyzer by default are floating (not ground referenced). This helps a lot with ground loops although sometimes I get better results by connecting ground.

 

[pma]

I'm guessing Amazon searching for one isn't what qualifies as "safety built"?

Hi Tks, I am attaching a photo of my transformer and also a link to some documentation.

https://www.tme.eu/Document/874123f9c6e613741592032f04f67daf/PFS-EN.pdf

BREVE TUFVASSONS PFS250/230/230V

 

[Wombat]

Groundloops – measurements and solutions

When 2 audio components built in class I are interconnected with a single-ended link cable, troubles may be expected almost for sure and they are rising with length of interconnecting cable and distance of the components. The result is a hum and buzz from speakers.

How does it happen? The following image shows class I signal source and class I amplifier interconnected by a simple single-ended link cable. Both components are supplied from 2-wire mains network (L + PEN), which is the worst option. The groundloop is shown in red. Loop current flows in the groundloop and it is sourced by magnetic induction B (from EM field sources) and by a voltage drop across resistance of the PEN wire, due to supply currents. This creates a low impedance voltage source, which creates a voltage drop across Rsh cable shielding resistance and this voltage drop is added to the audio signal, resulting in hum and buzz.
View attachment 46096
We need to know how big is the ground interference voltage. For this purpose, we disconnect the loop at one point and use a battery operated instrument or the one supplied through an isolation transformer.
View attachment 46097
Next image shows the real measurement in the groundloop. Please note that the voltage swing is almost 1V. Loop current was about 63mA.
View attachment 46098
Next, we shall see how different signal component designs would affect the resulting hum and buzz. In all cases, the signal source will be in class I. On the receiving side will be a class I preamplifier with both single-ended and balanced inputs. Preamplifier output is set to drive the power amplifier to its full power, full swing. Al the spectra are ref 0.83V, 0dB = 0.83V.

First measurement is with source balanced output, 12m of balanced twisted pair cable and balanced input of the preamp. We can see that the preamp with balanced input almost completely eliminates any hum/buzz voltages.
View attachment 46099
Now the same measurement, but single-ended input of the same preamp is used. This is terrible, isn't it.
View attachment 46100
The last one is a balanced input again, but the XLR pin 1 is wired improperly. Not to chassis, but it goes through a piece of wire to PCB ground. We can see worse result than in the 1st case.
View attachment 46101

We might think that if the preamp on the right side would be in class II, the situation would be better. Yes, but not completely. Some capacitive current still flows in the loop and worsens S/N in the single-ended mode.

The secondary of the TX in the signal source is not 'floating'. Is any audio gear built that way?

 

[pma]

FYI the unbalanced inputs on my Audio Precision Analyzer by default are floating (not ground referenced). This helps a lot with ground loops although sometimes I get better results by connecting ground.

Yes, AP has very nice solution of balanced floating I/O, especially the older series. And also the floating transformer coupled outputs (27XY), I am not sure if 555 still has it, it was very useful.

 

[pma]

The transformer in the signal source is not isolating the secondary from earth. I wouldn't build like that.

The introductory image is not a lecture how it should be build or how you/I would build it, it is a worst case example that can be met in real world and it is an equivalent to desktop PC, e.g. I know so many people complaining if they connect a PC into their system, which often has a power amplifier in class I - and you have groundloop. Even many notebooks have the DC adapter connected to PE ant their ground, 0V, is then connected with PE which makes them class I signal sources. You will also find measurements in this forum that are clearly affected by groundloops. Ohmic loops, as a worst case, or capacitive, that are not so evident.

Yes signal ground is connected to chassis in many audio products and if it is class I, then to PE as well.

 

[Wombat]

The introductory image is not a lecture how it should be build or how you/I would build it, it is a worst case example that can be met in real world and it is an equivalent to desktop PC, e.g. I know so many people complaining if the connect a PC into their system, which often has a power amplifier in class I. Even many notebooks have the DC adapter connected to PE ant their ground, 0V, is then connected with PE which makes them class I signal sources. You will also find measurements in this forum that are clearly affected by groundloops. Ohmic loops, as a worst case, or capacitive, that are not so evident.

Yes, most of the issues posted involve computers and their ilk. The designers/assemblers are at fault for lacking understanding that there is more to it than just connecting modules together. Cable routing can be awful as well.

 

[pma]

I have replicated the measurements now, after 2 1/2 years. Same cables and basic arrangement between 2 rooms, as in the 1st image of post #1. But different audio components - I do not have the former ones used in the 1st test. The result is very similar, almost identical. The SE path adds hum/buzz noise (voltage across cable shielding/signal return) to the signal, in unacceptable level. Yes the distance between components is bigger than usual as well as the cable length (10m, 12m), but it makes the things clear. For a shorter distance and smaller loop, the problem is smaller, but not negligible. The balanced path behaves much better, according to its CMR. And the XLR pin 1 must be on chassis to lead the loop current to PE and not to a signal ground on a PCB. This is very important. Measurements are calibrated in dBV, receiving preamp (class I, signal gnd - pin1 - on chassis and thus on PE) volume is set to give about 1V max. output voltage, which would drive my power amp to its maximum swing.

 

[restorer-john]

Another issue with earthed components that is a tough one to solve, is monoblock amplifiers that are located a distance away from the source/preamplifier and/or use different power outlets. The earth potentials are different and current flows through the RCA or GND between the pre and power amps causing hums and buzzes.

 

[Wombat]

This can occur with separated power power outlets on the same circuit. Over time, the thermal expansion and contraction of screwed outlet terminal connections can lead to poor contact pressure and even contamination of the contact. This can be bad enough to cause house fires.

So cleaning and re-tensioning power outlet terminals, especially in old houses, is worth considering.

 

[pma]

Another issue with earthed components that is a tough one to solve, is monoblock amplifiers that are located a distance away from the source/preamplifier and/or use different power outlets. The earth potentials are different and current flows through the RCA or GND between the pre and power amps causing hums and buzzes.

Exactly! And many potential customers do not (and have no chance if they are not experts) realize this. The monoblocks are often located close to their speakers so there is a huge loop from preamp to left monoblock - PE wire - right monoblock and back to preamp. Again, properly designed balanced interconnection should be used. Or - a floating preamp output. This is very rare, but works well, if the preamp has output CMR capability. The circuit hint is in the DRV134 datasheet.

 

[KSTR]

You might want to have a look at this older post of mine where I showed the influence of even rather low balancing currents on unbalanced interconnections.

1) Whenever there is any other current on the cable shield than the audio signal return current you have a "ground loop", better called balancing current problem. No Earth Ground is needed for this, only two pieces of equipment that are at different reference potentials for whatever reason.

2) Best way to deal with it is eliminating the balancing current by isolation: digital data stream isolators, supply isolators (mains side via iso-transformer or DC output side of "wall-warts" via DC/DC isolator), audio signal isolators (transformers).

3) But at first, try to mininize and divert the balancing current with cheap/simple means! The points here is to make the connection of the audio grounds (RCA shells/shields) dominate the (hidden) power supply connection by giving the audio connection the lowest possible impedance whereas the power supply impedance (common-mode) should be as high as possible :
- use the shortest feasible interconnects.
- use interconnects with especially low shield resistance (quality braided shield, maybe even two layers).
- tie all additional GND points (notably RCA shells of unused analog ins/outs) together between gear with short thick separate cables, forming a GND grid, almost a plane, "everything connected with everything" with a short low-impedance path.
- use the longest/thinnest mains cables available and coil them up (you may use eg 30m/100ft and even longer cable drums). One of those "common mode coils" per device! Especially when they have audio ground at protective earth. And all going to the same power strip, of course. Note that when using very high current devices like huge class-A power amps you may have to uncoil the drum to prevent overheating and triggering the associated fuse in the drum (check before using!).

EDIT:
- use the highest possible send level at the interconnect's source end, then maybe add a passive attenuator to bring down the level, notably for power amps that don't have an input level pot.

 

[Wombat]

Don't coil up power cords that are in use as you get an inductive heating device. Many house fires are caused by this simple act. Use a cord of the right length. If you have an extension cord on a reel, unwind it for use.

 

[stereo coffee]

That's helpful, thank you. So what are the solutions? Assuming I have a device, say a DAC or a preamp forming a ground loop with an amp, and I can only use single-ended interconnects, what are my options besides a full isolation transformer?

A L pad comprised of 4x NSL32SR3 suitably powered

 

[KSTR]

@Wombat , at least here in Germany any industrial quality cable drums are additionally rated for coiled-up duty (with de-rated power specs, ususally 1/3rd of the un-coiled max power) and also don't have metal core. Plus they always have the resettable thermal fuse. Anything else would be a complete nightmare, for example for construction workers.

Don't use your grandfather's stuff you found in some dusty corner of your basement/garage!

 

[KSTR]

A L pad comprised of 4x NSL32SR3 suitably powered

How would any L-Pad reduce balancing current on the shield?