Ground loop elimination and PC usb noise

[solderdude]

Yes, you still can get a groundloop because the data needs a ground which will always be connected.
In theory, when a groundloop is present you could disconnect the ground so would only have the 2 data wires connected. There are some gizmo's doing that like the ifi I believe. Not all DAC's like that though.
The +5V wire is not the issue in most cases, the ground connection is.

 

[Rantenti]

Yes, you still can get a groundloop because the data needs a ground which will always be connected.
In theory, when a groundloop is present you could disconnect the ground so would only have the 2 data wires connected. There are some gizmo's doing that like the ifi I believe. Not all DAC's like that though.
The +5V wire is not the issue in most cases, the ground connection is.

The Matrix element H USB card has a switch to cut power going into the USB. So in theory neither this card nor taping up the power pins of the USB connectors would disconnect the ground?

I have an ifi iDefender which I think it disconnects the ground regardless of power supply. However, since I swiched to balanced interconnects, I dont hear noises any more, so there is no audible difference with or without the iDefender now. I'm curious though, whether there are still some ground noise that is beyond my audibility without the iDefender and it is better to leave it in the signal chain.

 

[KSTR]

@solderdude ,
5V_bus and GND are tightly coupled together by a regulator and bypass cap at the source and bypass cap at the sink (the device). As long as either GND or 5V_bus are connected to the device there is a low impedance AC path for the balancing current.

Disconnecting 5V_bus and GND at the device does works sometimes yet other times it does not. For USB enumeration a reference to GND is required as the signaling is not the current-source differential scheme used for the actual USB packets later. I seem to remember @mansr has a gadget which opens the GND/Supply path after enumeration and refreshes the data line signals.

And the current sources are lousy, just a few kOhms of output impedance, still no brake for balancing current coming from a high impedance high voltage potential like capacitive coupling in a class-II "isolated" mains PSU. But even if the output impedance were very high, the signal pins still do only have limited voltage range where they can freely spin so whenever the voltage signal seen on the line approaches or tries to exceed the supply rails the communication breaks. The connection is potentially unsafe and still doesn't block the leakage currents, that is. Nothing gained.

The impedance and voltage compliance range is large enough, though, when the balancing current is from a PE-to-PE connection (all devices class-I), the classic "ground loop". In this case even the enumeration often works without a direct GND as the remote GND path is low enough impedance and voltage differentials are in bounds.

 

[Krusty09]

Here it is


View attachment 43977

Is the computer monitor hooked into this power strip as well and everything else that is connected to the PC because remember that anything that is connected to that PC is a possible ground loop problem especially if plugged into a different outlet.

 

[Rantenti]

Is the computer monitor hooked into this power strip as well and everything else that is connected to the PC because remember that anything that is connected to that PC is a possible ground loop problem especially if plugged into a different outlet.

My old computer + audio equipment were all into one wall plug and there was quite a lot of noise (ground loop and EMI). The ground loop culprit was the router (ground loop noise disappears when I touch the router). EMI was from the GPU. I was using RCA at that time. I switched to XLR and all noises were gone.

I bought a new computer recently. Now the audio rack and the computer are using two wall outlets in the same ring mains circuit. No noises that I can hear. So I'm tempted to remove the iDefender. Just curious whether there is actually still ground loop noises which is soft beyond my hearing though

 

[solderdude]

@solderdude ,
5V_bus and GND are tightly coupled together by a regulator and bypass cap at the source and bypass cap at the sink (the device). As long as either GND or 5V_bus are connected to the device there is a low impedance AC path for the balancing current.

Disconnecting 5V_bus and GND at the device does works sometimes yet other times it does not. For USB enumeration a reference to GND is required as the signaling is not the current-source differential scheme used for the actual USB packets later. I seem to remember @mansr has a gadget which opens the GND/Supply path after enumeration and refreshes the data line signals.

Yes, that's what I said.
Even when the 5V is disconnected but the ground connection is not the +5V does not matter. For HF garbage they have 'the same potential'. So feeding with a 'clean' 5V while coupling the grounds won't make any difference unless you have a really crappy DAC or the 5V coming from the USB bus is well below 5V.

When a ground loop is present the ground is present on both sides (assuming V+ is disconnected as well) via the ground coming from both sides.
It could well be there is a voltage exceeding a required voltage difference in which case disconnecting the ground does not work.
Could also be well within just a few volt difference. Also depends on how the USB side is implemented in the DAC. Sometimes ground is needed to make 'first contact' and afterwards is of less importance. This all depends on the actual implementation of the USB receiver in the DAC.

For that reason, if one has an audible ground loop then the most obvious action is to see where it occurs and then take the best action.
Could be changing of equipment, could even be interlink cables, could be going balanced (if possible) or using isolation devices in the audio path (line-level audio transformers), using optical inputs (if possible and circumstances allowing) or SPDIF out via an data isolation transformer (when it isn't present) or when using USB by using a real USB isolation device. For the latter one should choose the one you need.
In some cases things like the ifi or similar working devices could reduce issues.

 

[solderdude]

Just curious whether there is actually still ground loop noises which is soft beyond my hearing though

Just remove the device. When you hear noises that should not be there simply put it back. When you hear no noises without the defender but get occasional 'ticks' when switching off devices somewhere in the house put it back.
When something is inaudible it may still be there but doesn't matter.

 

[Lambda]

Here it is

If you use RCA to XLR make sure to use this kind of cable !

Is all noise gone if you unplug the USB cables?
If so get a USB Isolator.
yes the limit it to USB1 speed but that's normally good enough.

You still have a ground loop between the Speakers PE Ground and the RCA gnd but at least your uncoupled from the noisy PC ground.
To minimize this loop area rout the speaker power cables tidy to the signal cables back to the DAC
hav a extra power strip at the DAC for the speaks.

The best way is to get DAC with Balanced outputs and Optical input.

 

[Rantenti]

If you use RCA to XLR make sure to use this kind of cable !

Is all noise gone if you unplug the USB cables?
If so get a USB Isolator.
yes the limit it to USB1 speed but that's normally good enough.

You still have a ground loop between the Speakers PE Ground and the RCA gnd but at least your uncoupled from the noisy PC ground.
To minimize this loop area rout the speaker power cables tidy to the signal cables back to the DAC
hav a extra power strip at the DAC for the speaks.

The best way is to get DAC with Balanced outputs and Optical input.

I now got a mains power conditioner supplying a fully balanced DAC plugged in and use XLR to XLR cables exclusively from DAC to Preamp to active speakers. The noise is gone. Though some say RCA sound better in some high-end equipment (I've never experienced it), for me, XLR is the quick solution to go.

 

[Lambda]

Though some say RCA sound better in some high-end equipment

"Some say" and "in some high-end equipment"
If its done right balanced is better but i'm sure there is some equipment where it is not done right.
And "Some" might even like the extra noise and distortion.

Its Almost like they keep this Myth alive to sell solutions to problem otherwise not existent like "power conditioner"

 

[Rantenti]

"Some say" and "in some high-end equipment"
If its done right balanced is better but i'm sure there is some equipment where it is not done right.
And "Some" might even like the extra noise and distortion.

Its Almost like they keep this Myth alive to sell solutions to problem otherwise not existent like "power conditioner"

Yes, balanced should be theoretically better I think from what little I've read before. There are certain enthusiasts who love vintage sound and swear by ancient telephone cables. I do agree it's the distortion that gives certain "vintage tone and character" that some like. Not my cup of tea though

My experience with the Topping A90 was noise on the RCA and clean sound on the XLR. So I'm happy to go with balanced. I do have a "Balanced Isolation Power Conditioner", the Keces BP-1200. Not something "fancy", bought second-hand cheaply when I was still using RCA, in an attempt to remove noise, anyway... I now treat it as a more reliable power distributor with some protective features.

 

[mansr]

Disconnecting 5V_bus and GND at the device does works sometimes yet other times it does not. For USB enumeration a reference to GND is required as the signaling is not the current-source differential scheme used for the actual USB packets later. I seem to remember @mansr has a gadget which opens the GND/Supply path after enumeration and refreshes the data line signals.

This one: https://www.audiosciencereview.com/forum/index.php?threads/the-usb-troll.23054/

 

[spiritofjerry]

Yes, that's what I said.
Even when the 5V is disconnected but the ground connection is not the +5V does not matter. For HF garbage they have 'the same potential'. So feeding with a 'clean' 5V while coupling the grounds won't make any difference unless you have a really crappy DAC or the 5V coming from the USB bus is well below 5V.

When a ground loop is present the ground is present on both sides (assuming V+ is disconnected as well) via the ground coming from both sides.
It could well be there is a voltage exceeding a required voltage difference in which case disconnecting the ground does not work.
Could also be well within just a few volt difference. Also depends on how the USB side is implemented in the DAC. Sometimes ground is needed to make 'first contact' and afterwards is of less importance. This all depends on the actual implementation of the USB receiver in the DAC.

For that reason, if one has an audible ground loop then the most obvious action is to see where it occurs and then take the best action.
Could be changing of equipment, could even be interlink cables, could be going balanced (if possible) or using isolation devices in the audio path (line-level audio transformers), using optical inputs (if possible and circumstances allowing) or SPDIF out via an data isolation transformer (when it isn't present) or when using USB by using a real USB isolation device. For the latter one should choose the one you need.
In some cases things like the ifi or similar working devices could reduce issues.

So I ended up solving ground loop noise today by simply kapton taping the 5v and ground pins on my USB cable into my DAC, which has it's own power. No more high frequency noise. Is this safe to do? It works perfectly, but I'm concerned I am at risk in some way

 

[solderdude]

When a ground loop is present and it works then it is fine.
No guarantees that the USB input from the DAC won't be damaged when for some reason the ground loop has gotten a voltage on it exceeding that of the components in the USB receiver.

Now that you've proven the culprit is the USB connection perhaps invest in a galvanic USB isolator ?

 

[spiritofjerry]

Now that you've proven the culprit is the USB connection perhaps invest in a galvanic USB isolator ?

This would be good, but damn most of them are expensive. Do you think something like the ifi defender could work, is it a true isolator?

 

[Lambda]

Do you think something like the ifi defender could work, is it a true isolator?

no its not a real isolator.

Is 24/96 not good enough?
because the low speed usb isolators cost like 10$

Have you tried using a usb hub?

Is this safe to do? It works perfectly, but I'm concerned I am at risk in some way

it can maybe damage the USB port on the PC or DAC since now all the ground loop curren/voltages goes directly into the data pins.
USB
If you want to make a hack like this do it between 2 usb hubs so in case something happens you blow up just a 2-5$ hub and not your Mainboard or DAC

 

[solderdude]

the ifi defender is an automated version of what you did and determines if it is safe to cut the power and ground.
It is not a true isolator but most likely safer to use than just cutting V+ and gnd

 

[pma]

If you use RCA to XLR make sure to use this kind of cable !

If you want for this to work properly and get reasonable CMR, you MUST insert a resistor into the "RCA Ground" branch and the value is same as Zout of the preamp or DAC used. Remember you often have Zout of 100 ohm and more and amp input impedance from about 2kohm (Topping ....) to some 100kohm. This makes huge CMR difference if the balancing resistor is not used. See below. You always need to get deep enough into the issue that is to be fixed.

 

[Lambda]

If you want for this to work properly and get reasonable CMR, you MUST insert a resistor into the "RCA Ground" branch

This is good practice and further improves CMR.

Some Include a resistor or resistor with capacitor parllel on the shield at the RCA side.
To my understanding this Increases Common mode noise at the signal wires but reduces common mode current on the shield.
Also reducing current on the RCA and therefore maybe reducing noise generated over its contact resistance?
Not sure if it would do any good, maybe in some cases?

Also the effectiveness of this method is very Debated here an it works manly for HF currents.
But Ferrite common mode chokes.
In my testing 2 25cent feritte clamps added about 0.5Ohms @ 10khz
If this is significant depends on your noise current's and the loop impedance

 

[Gradius]

TRUE galvanic isolator, EMI, etc, looks like this.

Yes, its high-speed compatible (480Mbit/s).