USB Noise / Ground Loop - Best Solution

[javier]

Hello,

I just bought some speakers (Adam Audio T5V) and I'm having some noise issues.

My current set up is:

• PC connected to Topping D10S (via USB)
• Topping D10S connected to Topping L30 (via RCA cables)
• Topping L30 connected to the speakers (via out RCA cables)
• Topping L30 connected to my headphone

I hear some crackle all the time with this connections and a huge amount of noise when the GPU is loaded.

Some things I've tried:

• Plugging everything to a single power strip - doesn't help
• Plugging everthing to different power connections - doesn't help
• Connecting Topping D10S to a laptot - Fixes the noises
• Connecting speakers directly to the Topping D10s - doesn't help

Note that the noises also occur when using the headphone output if the L30 is connected to the speakers. If I unplug both speakers from electricty or from L30 the noise is not present in the headphone output.

My guess is that there is a ground issue between the USB powered D10s and the speakers.

What would be the best solution for this:

1. Buying a USB isolation dongle with external power
2. Buying a DAC with separate power
3. Buying a DAC with balanced outputs
4. Buying a DAC with optical input

Is there any other option? Anything I could do without buying anything?

 

[staticV3]

A USB isolator should fix your issue. They're around $50 nowadays:
https://www.tpdz.net/productinfo/733226.html
https://hifimediy.com/product/hifime-high-speed-usb-isolator-v2/
https://aliexpress.com/item/1005001927948863.html

 

[anotherhobby]

I had the same issue and tried several that did not work, and then finally found success with the Topping HS01. Here is a thread on it: https://www.audiosciencereview.com/...opping-hs01-useful-device-or-snake-oil.33345/

Here it is on amazon: https://www.amazon.com/Topping-Isolator-Compatible-PCM32bit-withstands/dp/B09MF9VTQG

 

[javier]

A USB isolator should fix your issue. They're around $50 nowadays:
https://www.tpdz.net/productinfo/733226.html
https://hifimediy.com/product/hifime-high-speed-usb-isolator-v2/
https://aliexpress.com/item/1005001927948863.html

Thank you, I will check those

I had the same issue and tried several that did not work, and then finally found success with the Topping HS01. Here is a thread on it: https://www.audiosciencereview.com/...opping-hs01-useful-device-or-snake-oil.33345/

Here it is on amazon: https://www.amazon.com/Topping-Isolator-Compatible-PCM32bit-withstands/dp/B09MF9VTQG

Did you have to use an additional power to the HS01?

 

[anotherhobby]

Thank you, I will check those

Did you have to use an additional power to the HS01?

Yes

 

[sam_adams]

Most of the GPU noise is being directly injected into the ground plane. That makes it hard to get rid of short of galvanic isolation.

 

[javier]

I ended up getting a cheap ADUM3160-based USB isolator and it fixed the issue completely.

I'm restricted to USB Audio Class 1 now but not a big issue, thank you everyone!

 

[Dacsalamanca]

hello everyone: The solution is easier than all that.... First of all, my solution only works with DACs not powered by USB power. Tested with Volumio installed on PC. You just have to remove the power pins of the USB connector (right and left ends) with some strips of insulating tape. canceling the passage of unnecessary and noisy current. I recommend using 3m electrical tape to avoid problems with poor quality adhesive.
Good look

 

[somebodyelse]

That's a "you might be lucky" solution. Depending on how both ends handle the shield you may still have a route for ground related noise. Missing ground and power connections can cause problems with detection, mostly with ports supporting newer USB versions.

 

[Dacsalamanca]

I did not modify the shield. only the power pins, so the solution is not valid for devices that need power, desktop dac
In Volumio the device remains configured after the cancellation of the pins (gustard x26 pro)

 

[KSTR]

hello everyone: The solution is easier than all that.... First of all, my solution only works with DACs not powered by USB power. Tested with Volumio installed on PC. You just have to remove the power pins of the USB connector (right and left ends) with some strips of insulating tape. canceling the passage of unnecessary and noisy current. I recommend using 3m electrical tape to avoid problems with poor quality adhesive.
Good look

Not a recommended "fix" as this can easily fry the sensitive inputs, also for the initial communication the USB data lines needs a ground reference. It may or may not work and for unknown time until it breaks.
At least now you know that proper USB isolation with Topping HS2 or the like would correctly solve your problem.

Depending on how both ends handle the shield you may still have a route for ground related noise.

Yep. It appears like there's no clear standard and I've seen all four combinations of direct vs. R//C-bonded shields and upstream vs. downstream ports.

 

[Dacsalamanca]

Not a recommended "fix" as this can easily fry the sensitive inputs, also for the initial communication the USB data lines needs a ground reference. It may or may not work and for unknown time until it breaks.
At least now you know that proper USB isolation with Topping HS2 or the like would correctly solve your problem.


Yep. It appears like there's no clear standard and I've seen all four combinations of direct vs. R//C-bonded shields and upstream vs. downstream ports.

Can someone confirm, is there any reason why the DAC is damaged by canceling the voltage input pins? I don't understand electronics, but I don't see what can harm the cancellation of the voltage, if the sending of music from the PC to the DAC is correct in all formats and the sound is cleaner than ever (more than the spdif)

 

[somebodyelse]

Can someone confirm, is there any reason why the DAC is damaged by canceling the voltage input pins? I don't understand electronics, but I don't see what can harm the cancellation of the voltage, if the sending of music from the PC to the DAC is correct in all formats and the sound is cleaner than ever (more than the spdif)

The USB receiver will have a maximum voltage that the data lines can have relative to ground and 5V before damage occurs. If one or both of your ends are 'floating' they can end up with significantly different relative potential if they don't have something to keep them together (the 0V line you blocked). If the difference gets big enough then damage can occur. A similar failure mode happens with industrial serial links when people assume the ground wire isn't important because the signalling is differential. This is where we come back to how the shield is handled - it might provide the link to protect you from this, but in that case it may also provide the route for ground-related noise issues that you were trying to avoid in the first place.

Creative violation of the USB spec may work if you're lucky, or you know exactly what you're doing in a specific situation, but it can also lead to things failing in interesting ways. A properly designed USB isolator stays within the spec and should 'Just Work' - the problem so far being that the inexpensive ones so far only do 'Full Speed' 12Mbps. Isolators using the newer chipsets for 'HiSpeed' 480Mbps are starting to appear, and should become widely available before too long.

 

[Zapper]

The USB receiver will have a maximum voltage that the data lines can have relative to ground and 5V before damage occurs. If one or both of your ends are 'floating' they can end up with significantly different relative potential if they don't have something to keep them together (the 0V line you blocked). If the difference gets big enough then damage can occur. A similar failure mode happens with industrial serial links when people assume the ground wire isn't important because the signalling is differential. This is where we come back to how the shield is handled - it might provide the link to protect you from this, but in that case it may also provide the route for ground-related noise issues that you were trying to avoid in the first place.

Creative violation of the USB spec may work if you're lucky, or you know exactly what you're doing in a specific situation, but it can also lead to things failing in interesting ways. A properly designed USB isolator stays within the spec and should 'Just Work' - the problem so far being that the inexpensive ones so far only do 'Full Speed' 12Mbps. Isolators using the newer chipsets for 'HiSpeed' 480Mbps are starting to appear, and should become widely available before too long.

I second this - 100% true.

 

[Zapper]

What would be the best solution for this:

1. Buying a USB isolation dongle with external power
2. Buying a DAC with separate power
3. Buying a DAC with balanced outputs
4. Buying a DAC with optical input

The OP had success with #1. For the record, #4 is also a viable option to provide a galvanic barrier (and implies #2 as well). #2 by itself or #3 are less likely to be effective.

 

[somebodyelse]

#3 should work if DAC and downstream device are correctly engineered, and does in my limited experience. Unfortunately there are still devices that don't do it right, so it's not as guaranteed as #1 and #4. It's worth noting that coax digital outputs may be transformer isolated too - it's required in the pro standards, and some hifi manufacturers do it even though it's not required there.

 

[Hatto]

View attachment 356243
hello everyone: The solution is easier than all that.... First of all, my solution only works with DACs not powered by USB power. Tested with Volumio installed on PC. You just have to remove the power pins of the USB connector (right and left ends) with some strips of insulating tape. canceling the passage of unnecessary and noisy current. I recommend using 3m electrical tape to avoid problems with poor quality adhesive.
Good look

Came to say this, but someone beat me to to it. Kudos!

 

[MC_RME]

This is not a solution but complete bollocks. Read the posts above!

 

[widemediaphotography]

In the examination of points 1,2,3,4, optical USB cables were not considered... In theory they would not appear to be galvanically isolated, but in practice they seem to be effective, avoiding the limitations of Toslink standard. What about of this solution?

 

[MC_RME]

Mod to the rescue...