PC - DAC ground loop (GPU culprit)

[Dexter_prog]

I noticed a ground loop between my PC and my DAC. It's pretty much exactly the same described in this video (skip to 11:28). It also responds to locking fps and changes pitch accordingly.

This has been reported extensively across various forums. I'll just link to one nvidia thread, but I also saw it for AMD cards: {{MetaTags.og.title}}

I bought an IFI idefender, but not only did it not break the loop, but it also introduced a new hissing sound, so it's a no-go for me. I tried disconnecting everything except the PC, the DAC and the amp, and the ground loop persists. It even happens with the DAC disconnected from the current. Connecting the DAC to a laptop (and using the laptop's battery) doesn't produce the ground hum, so the DAC is not defective.

The only thing that breaks the loop is lifting the ground from either the PC or the power conditioner where all my audio stuff is connected to (I did try other extension cords with same results). I did try different power sockets in the room, same result.


Other than permanently lifting the ground from my audio gear, do you guys have any other suggestions?

Setup:
- PC: i5 8600k + asus z370-a prime + nvidia EVGA 1070 + 2x8 gb ddr3000
- Audio setup:
-- Audiowalle TP1000 power conditioner (AKA fancy power extension cord)
-- Vincent sv237-mk
-- DAC: SMSL m500
-- KEF R300

 

[Racheski]

Is the PC connected to the TP1000 as well?
What did you change that caused the ground loop to appear all of a sudden?

 

[amirm]

Use a Toslink optical if you have it instead of USB.

 

[levimax]

The only fool proof thing that always works for me when mixing computers with multiple other devices is an isolation transformer....not cheap but does the job...much safer for people and equiptment than lifting a ground //www.jensen-transformers.com/industry_type/audiophile/

 

[Vasr]

The isolation transformer as suggested above on the PC power line will work.

The EBtech one is not very expensive. But note 6A max draw on this one.
https://www.zenproaudio.com/ebtech-hum-x

There seems to be another brand also
https://www.sweetwater.com/store/detail/MHumX--morley-hum-exterminator-ground-loop-hum-exterminator

I have the same problem with my HTPC and it developed suddenly one day. Not sure why. After much experimentation, I realized the problem isn't created by the video card but when a ground loop already exists, the use of the video card makes it worse and introduces additional noise that may then become very noticeable.

To do an uncompressed multi-channel audio out, there is no reasonable connector with galvanic isolation.

I fixed it by lifting the PC ground and connecting the PC chassis to the amp chassis with a ground wire so it is safe. But this fixed it for one sound card but not another on the same PC! My USB out never showed this problem. Only HDMI and Line outs.

What it affects seems to depend on how the ground loop is manifesting itself on the PC. It could be injecting a signal into the ground which affects audio pins or the shield of the connector or the power supply through the USB that the DAC uses or all of them.

 

[dougi]

You can buy a few different USB 3.0 galvanic isolators, but they are not cheap. eg iFi nano iGalvanic 3.0, Intona that has been reviewed on this forum by Amir https://www.audiosciencereview.com/forum/index.php?threads/intona-usb-3-0-isolator-review.11700/

 

[Dexter_prog]

Is the PC connected to the TP1000 as well?
What did you change that caused the ground loop to appear all of a sudden?

What did you change that caused the ground loop to appear all of a sudden?[/QUOTE]
no, pc is connected to one socket, TP1000 to another.

I don't know what I changed. I think I started noticing it when I bought the SMSL m500 (previously I had a nuprime ida-8) but I may have just not heard it. I also don't know why it went away and now it's back when all I did was change cables. I did try the previous cables and the hum is still present.

Tried connecting the pc to the TP1000 (that is, all audio, pc and screen to the TP1000 and that to the power socket)... same ground loop

 

[Dexter_prog]

Use a Toslink optical if you have it instead of USB.

it defies the purpose of the DAC, not only do I want to bypass the PCs soundcard, I also want it to decode MQA

 

[Dexter_prog]

The isolation transformer as suggested above on the PC power line will work.

The EBtech one is not very expensive. But note 6A max draw on this one.
https://www.zenproaudio.com/ebtech-hum-x

There seems to be another brand also
https://www.sweetwater.com/store/detail/MHumX--morley-hum-exterminator-ground-loop-hum-exterminator

I have the same problem with my HTPC and it developed suddenly one day. Not sure why. After much experimentation, I realized the problem isn't created by the video card but when a ground loop already exists, the use of the video card makes it worse and introduces additional noise that may then become very noticeable.

To do an uncompressed multi-channel audio out, there is no reasonable connector with galvanic isolation.

I fixed it by lifting the PC ground and connecting the PC chassis to the amp chassis with a ground wire so it is safe. But this fixed it for one sound card but not another on the same PC! My USB out never showed this problem. Only HDMI and Line outs.

What it affects seems to depend on how the ground loop is manifesting itself on the PC. It could be injecting a signal into the ground which affects audio pins or the shield of the connector or the power supply through the USB that the DAC uses or all of them.

unfortunately, I'm in Argentina (50 hz, 220 v here). Those aren't supported here.

 

[Vasr]

Tried connecting the pc to the TP1000 (that is, all audio, pc and screen to the TP1000 and that to the power socket)... same ground loop

unfortunately, I'm in Argentina (50 hz, 220 v here). Those aren't supported here.

Electrically the TP1000 is no different from an extension strip and will have the same problem.

You should then try what I am doing which is to lift the ground for the PC which solves the problem as you reported and then run an insulated electrical wire from the PC to the amp. You can use eyelet or spade connectors at both ends and slip it into one of the back screws on both units. Some of the old-style amps used to come with a ground connector for this purpose! This will keep the PC grounded but not have multiple paths for the ground loop in the signal.

If you are worried about warranty in loosening the screw, you can connect an unused RCA or any line out/in connector. Verify its shield is connected to the chassis ground and connect the wire to the shield of a dummy cable coming out of that connector.

 

[AnalogSteph]

it defies the purpose of the DAC, not only do I want to bypass the PCs soundcard, I also want it to decode MQA

You do realize you are trying to shoot yourself in the foot pretty hard, do you?

As far as I'm aware MQA is pretty much a non-feature - it can be decoded in software perfectly fine, and nobody really needs it anyway. (Same goes for 384 kHz, but I digress. Anything beyond 24/96 is completely academic for audio playback, and arguably 24/48 or 24/44 depending on DAC.)

The main issue is the ground loop between PC + DAC and the Vincent amplifier, on which I spy a 3-prong IEC power connector, which smells of IEC Class I device. Not a good match with unbalanced connections. Traditional hi-fi is generally Class II (double insulated) for good reasons.

I suspect getting a Behringer HD400 with associated cabling would give performance no better than the DAC already built into the Vincent, complete with a Toslink output. Or the 500 when using Toslink input. A USB isolator may be another option, but these may be restricted to USB 1.1 data rates.

Plan B, replace the Vincent with either something sporting balanced inputs, or a traditional hi-fi integrated amp from large-scale manufacturers like Yamaha, Denon, Pioneer etc.

 

[ernestcarl]

The isolation transformer as suggested above on the PC power line will work.

The EBtech one is not very expensive. But note 6A max draw on this one.
https://www.zenproaudio.com/ebtech-hum-x

There seems to be another brand also
https://www.sweetwater.com/store/detail/MHumX--morley-hum-exterminator-ground-loop-hum-exterminator

I have the same problem with my HTPC and it developed suddenly one day. Not sure why. After much experimentation, I realized the problem isn't created by the video card but when a ground loop already exists, the use of the video card makes it worse and introduces additional noise that may then become very noticeable.

To do an uncompressed multi-channel audio out, there is no reasonable connector with galvanic isolation.

I fixed it by lifting the PC ground and connecting the PC chassis to the amp chassis with a ground wire so it is safe. But this fixed it for one sound card but not another on the same PC! My USB out never showed this problem. Only HDMI and Line outs.

What it affects seems to depend on how the ground loop is manifesting itself on the PC. It could be injecting a signal into the ground which affects audio pins or the shield of the connector or the power supply through the USB that the DAC uses or all of them.

The ebtech plugs don’t always work. In such cases I use the xlr/trs “transformer” boxes. I have three and they all work great and convert unbalanced DAC RCA outputs to balanced outs... expensive, though — for whatever reason the cheaper HE-2 ([TRS-only] also have a pair) doesn’t seem to work as well.

 

[rkbates]

Electrically the TP1000 is no different from an extension strip and will have the same problem.

You should then try what I am doing which is to lift the ground for the PC which solves the problem as you reported and then run an insulated electrical wire from the PC to the amp. You can use eyelet or spade connectors at both ends and slip it into one of the back screws on both units. Some of the old-style amps used to come with a ground connector for this purpose! This will keep the PC grounded but not have multiple paths for the ground loop in the signal.

If you are worried about warranty in loosening the screw, you can connect an unused RCA or any line out/in connector. Verify its shield is connected to the chassis ground and connect the wire to the shield of a dummy cable coming out of that connector.

Not trying to be a spoil sport but PEOPLE HAVE DIED when they muck around with protective grounds and use isolation transformers incorrectly. Definitely not recommended unless trained and licensed appropriately.

 

[Vasr]

Not trying to be a spoil sport but PEOPLE HAVE DIED when they muck around with protective grounds and use isolation transformers incorrectly. Definitely not recommended unless trained and licensed appropriately.

Not a spoil sport (caution is always a good thing to bring up) but unduly alarmist I think. Far more PEOPLE HAVE DIED likely from using extension cords or multiple outlet strips.

The EBtech hum eliminators are very commonly used in music stage and studios (which is why you find these in all the music stores). Their diodes are designed to fail to a short creating a straight ground connection. Many audio units use the same principle as part of their protection mechanism against shorts.

As for as connecting chassis is concerned, it is just a ground shunt to have a single path to the electrical ground by connecting multiple chassis together. You are not totally removing the ground connection from any of the units by doing this. This is safer than the ungrounded plugs still used in so many homes around the world for audio equipment. I don't think people are dying regularly from them. You don't need an electrician's license for this. You are not mucking around with any internal or live electrical wiring. The ground is a safety pre-caution if there happens to be a short in the unit, so it can flow through the ground (blow out fuses or trip relays) and not make the chassis live. That safety is still preserved with a ground shunt.

 

[rkbates]

Not a spoil sport (caution is always a good thing to bring up) but unduly alarmist I think. Far more PEOPLE HAVE DIED likely from using extension cords or multiple outlet strips.

The EBtech hum eliminators are very commonly used in music stage and studios (which is why you find these in all the music stores). Their diodes are designed to fail to a short creating a straight ground connection. Many audio units use the same principle as part of their protection mechanism against shorts.

As for as connecting chassis is concerned, it is just a ground shunt to have a single path to the electrical ground by connecting multiple chassis together. You are not totally removing the ground connection from any of the units by doing this. This is safer than the ungrounded plugs still used in so many homes around the world for audio equipment. I don't think people are dying regularly from them. You don't need an electrician's license for this. You are not mucking around with any internal or live electrical wiring. The ground is a safety pre-caution if there happens to be a short in the unit, so it can flow through the ground (blow out fuses or trip relays) and not make the chassis live. That safety is still preserved with a ground shunt.

All good when you know what you're doing, not so if you don't know what you don't know - so sticking to decent commercial products and using them in the intended manner will be a good outcome.

 

[KSTR]

The only thing that breaks the loop is lifting the ground from either the PC or the power conditioner where all my audio stuff is connected to (I did try other extension cords with same results). I did try different power sockets in the room, same result.

Other than permanently lifting the ground from my audio gear, do you guys have any other suggestions?

Hi,
this seems to suggest that you have the very classic ground-loop problem induced by earth-ground balancing currents between gear on the shields of your RCA interconnects which generate noise voltage.

IMHO, like @amirm said, try to break the earth ground connection at the best possible and most effective location, and that is the digital data feed connection to the DAC. This is the best location as (partial) lifting the safety-gounds on any piece of equipment is a clear no-go, and breaking the loop right at the RCA interconnects is also not the best place because the typical solution, good audio transformers, isn't exactly cheap and of course this is the most critical solution, being directly in the analog audio path. One could also use XLR cables with an external, powered balanced-to-unbalanced audio receiver module right at the amplifier input ... but again this seems awkward.

Alternatively, one could try "beef up" the problematic audio ground connection between DAC and amp and re-organize all the power cabling (so that it has high common-mode impedance, ideally reaching that of power isolation transfomers). This is not trivial either and can lead to endless frustration the next time you change anything significant as it does not really address the root cause but sure is worth a try. In this case, connect Pin 1 of both DAC XLR outputs to thick "GND" wires which go to all availabe RCA shell "ground-points" at the amp (analog I/O's only). And use the shortest possibly interconnects which should have very low shield resistance, good quality braided-shield cables. On the power side, try plug everthing into the same power strip first, notably the PC and the amp. Again a thick buss-bar type of connection between their earth-grounds might be benefical (think rack-mounted devices).

Therefore, I would suggest using an isolated input at the DAC for the simplest and cleanest 100% fix. Either use the optical in (you might need a small USB-->TOSLINK interface for this) or USB with an inserted true USB isolator device, like the Intona USB2.0 etc. True USB isolators are rather complex products so they tend to be more expensive than the USB-->TOSLINK devices but of course they are much more flexible and useful. I've truly learned to love USB galavanic isolation for audio and other fields...

To see if that really will cure the root problem please check if the noise is completely gone when the digital input cable (USB) is detached from the DAC but otherwise all is running. Maybe you also have a CD player with a TOSLINK output to test-feed the DAC's input.

General cabling tip, not closely related : use clamp-on ferrites generously on all cables except the RCA-interconnects. IME it always pays off trying to reduce RFI levels the gear must deal with, notably with earth-gounded PCs etc.

 

[Pluto]

These are highly effective in this situation but the sampling rate fed to the DAC may be no greater than 96kHz. Are we concerned about that? No, of course not.

But were this my site, I would instantly delete any posting that even discusses the 'virtues' of removing a protective mains earth.

 

[ernestcarl]

These are highly effective in this situation but the sampling rate fed to the DAC may be no greater than 96kHz. Are we concerned about that? No, of course not.

But were this my site, I would instantly delete any posting that even discusses the 'virtues' of removing a protective mains earth.

Cool device! I like to get one just to test — perhaps the ebtech I’m using is not even necessary.

Although... one would be hard-pressed to find anybody who actually believes removing grounding is a ‘virtue’. LOL

 

[Dexter_prog]

Thanks for all the replies. I have some questions:

1) Before spending money in the intona or some other similar device (take their price tag and multiply it by 1.5 or 2, and that is what it costs to import it to Argentina...), is there any way I can find out if such a device would work (other than in theory)? Because I spent money in the iDefender, which in theory should've worked, and it didn't...

2) If I lift the ground of the wall socket where the TP is plugged in (that is to say, all audio gear), but my DAC is connected to the PC by the USB cable, wouldn't that connection work as a shunt to the grounded wall socket where the pc is connected?

3) In Argentina, all buildings have circuit breakers installed. I'm not an electrician, but in the event of a non-grounded device malfunctioning and letting the current go through the chassis, wouldn't this circuit breaker cut all power?

 

[Dexter_prog]

Alternatively, one could try "beef up" the problematic audio ground connection between DAC and amp and re-organize all the power cabling (so that it has high common-mode impedance, ideally reaching that of power isolation transfomers). This is not trivial either and can lead to endless frustration the next time you change anything significant as it does not really address the root cause but sure is worth a try. In this case, connect Pin 1 of both DAC XLR outputs to thick "GND" wires which go to all availabe RCA shell "ground-points" at the amp (analog I/O's only). And use the shortest possibly interconnects which should have very low shield resistance, good quality braided-shield cables. On the power side, try plug everthing into the same power strip first, notably the PC and the amp. Again a thick buss-bar type of connection between their earth-grounds might be benefical (think rack-mounted devices).

I got a bit lost here. Could you please clarify?

To see if that really will cure the root problem please check if the noise is completely gone when the digital input cable (USB) is detached from the DAC but otherwise all is running. Maybe you also have a CD player with a TOSLINK output to test-feed the DAC's input.

Yes, I had tested this. When the usb is disconnected from either end, noise disappears. I do have a CD player with a toslink output, but what did you mean for me to try with it and the DAC?

---------


On the other hand, when I tried connecting the PC to the TP1000, I had to use a cable with an European schuko plug. When I did this, ground loop disappeared. Then I tried the same connection, but with a US plug (the one with 2 parallel connectors and a third rounded ground one), and this introduced the ground hum again. I don't quite understand how schuko plugs handle earth, but could this "solution" have taken place because either that scuko plug removed the ground by not having a third pin, or because there's something to do with the way to plug it (I noticed you haven plug it both ways, meaning you can flip it 180 degrees and it'll plug)