Audiophile PC ground loop issues?

[yavormoskov]

I am just building my audiophile PC and I have some serious issues and figured DYI guys would know better than most
I installed Burson Fun amplifier with Khadas tone board inside a PC case. Khadas is powered only by USB and Fun has its own power supply but inside the PC is powered by Molex connector from the PC power supply . My RCA cables are Canare 4e6s. When I open web browser or move the mouse there is a lot of added noise and distortion. And I mean a lot.
If I swap the cables to $5 generic ones the noise increases many times. So the better shielding helps a bit.
If I disconnect the Molex and connect the separate DC power supply, all the noise is gone and music plays normally despite the fact that physically both devices are still inside the same PC chassis .
Initially I thought that my computer supply is really bad but I decided to plug in the Molex power connector again but this time to connect the amp to my phone with a RCA to 3.5 adapter. No noise whatsoever. PC power supply, overclocked CPU, GPU, fans on max, crazy Asus Wifi adapter with 4 huge antennas and 0 noise. Total silence. I was shocked, what is going on?!
The nasty noise appears only when I connect the amp to the PC power supply and the Khadas DAC is powered by USB from the same PC.
The last test was to connect the Khadas DAC to my other computer and the Burson Fun amp is powered by MOLEX from the first computer and again there isn't any audible hiss or noise.
My conclusion is that when both draw power from the same computer noise issues appear. Do you think this is a case of nasty ground loop and how to solve it without the need one of them to be connected to external power supply? At the local Atlanta HiFi buys audiophile store they recommended me a company called Blue Jeans. The salesman told me Blue jeans sell Subwoofer isolation transformers and they could possibly solve my issues completely. I also found a product called iFi iDefender3.0 External USB Ground Loop Eliminator. Any other ideas? I would appreciate any help. Thank you.

 

[KSTR]

Yes this is a classic "ground loop" example.
The root problem is that the ground voltages at the DAC and at the amp are not the same (the difference is the noise) but are connected together tightly in the PC's power supply. This voltage difference is applied to your unbalanced interconnect cable where it forces a heavy current through the shield and that's where the noise is actually created as an audio signal. With better cables (low shield resistance) the noise is less, but won't go away.

You have to "break the loop" literally, unless all the supply voltages and the ground they are referenced to come from one single point.
This can be done in several ways:
1) Insert audio isolation transformers in the connection from DAC to Amp. Might be not an audiophile solution unless expensive transformers are used (Lundahl, Jensen, CineMag, Sowter, ...) and choice of model is critical.
2) Insert an USB Isolator that has enough power to supply the Khadas. Intona's USB Isolator might work well and is reasonably priced though not exactly cheap (I use it on a daily basis but haven't tried bus-powering anything from it, specs say it is good for at least 300mA).

3) Somewhat cheating, but still might work good enough at very low cost:
- Use the best and shortest(!) audio cables between DAC and amp, and add an additional thick(!) ground balancing wire that attaches to the (audio) grounds of these two devices. We want to make the resistance of this ground path as low as possible.
- Use the longest and cheapest USB cable in an effort to make this (parasitic) ground path high resistance (relatively speaking) so that it does not dominate. And it is a very good idea to wind up the cable as a coil, preferably on a large ferrite toroid core, or on a laminated core salvaged from an old toroid power transformer.
We want the noise voltage to be dropped along this path, not along the audio signal path.

 

[Veri]

Good tips from @KSTR. Tone board is perfect to go direct to a headamp, but apparently its grounding lay-out can be troublesome in some situations: https://blog.jdslabs.com/2018/11/jds-labs-atom-amp-official-release-benchmarks/#comment-22949 looks like you're running into such a situation.

 

[yavormoskov]

Yes this is a classic "ground loop" example.
The root problem is that the ground voltages at the DAC and at the amp are not the same (the difference is the noise) but are connected together tightly in the PC's power supply. This voltage difference is applied to your unbalanced interconnect cable where it forces a heavy current through the shield and that's where the noise is actually created as an audio signal. With better cables (low shield resistance) the noise is less, but won't go away.

You have to "break the loop" literally, unless all the supply voltages and the ground they are referenced to come from one single point.
This can be done in several ways:
1) Insert audio isolation transformers in the connection from DAC to Amp. Might be not an audiophile solution unless expensive transformers are used (Lundahl, Jensen, CineMag, Sowter, ...) and choice of model is critical.
2) Insert an USB Isolator that has enough power to supply the Khadas. Intona's USB Isolator might work well and is reasonably priced though not exactly cheap (I use it on a daily basis but haven't tried bus-powering anything from it, specs say it is good for at least 300mA).

3) Somewhat cheating, but still might work good enough at very low cost:
- Use the best and shortest(!) audio cables between DAC and amp, and add an additional thick(!) ground balancing wire that attaches to the (audio) grounds of these two devices. We want to make the resistance of this ground path as low as possible.
- Use the longest and cheapest USB cable in an effort to make this (parasitic) ground path high resistance (relatively speaking) so that it does not dominate. And it is a very good idea to wind up the cable as a coil, preferably on a large ferrite toroid core, or on a laminated core salvaged from an old toroid power transformer.
We want the noise voltage to be dropped along this path, not along the audio signal path.

Thank you, thank you a lot. Your vast knowledge on the subject really opened my eyes. The USB isolators or the Intona might be the ultimate solution. And Intona is much cheaper than iFi Nano iGalvanic3.0. When I saw the price of $379, I almost wanted to cry. I still have some time to think about. I still have other technical issues to solve. I have to create a PC back case bracket (with symmetrical holes) for female to female RCAs (https://www.parts-express.com/parts...F5U8bvEeh9jnLggegEihmArBR_njCZJRoCGSoQAvD_BwE ) so a male to male cable from the amp can plug in to the female RCA ends from the inside the PC chassis and from the other side powered speakers can be plugged in, I ordered low profile USB 3.0 header cable because it is blocking the amp output. Also, have to solder new PCB mount dual RCAs because the ones that came with the amp are trash and already corroded. And tomorrow will receive PC case front cable extension cables because the original ones are also blocking how deep the amp can go inside the 5.25 slot and do not want to damage the original ones. I will remove part of their plastic housing of the extensions and will carefully bend them out of the way so the amp can slide easily inside.

This is my first PC build, first soldering and first audio project. Wish me luck.

 

[yavormoskov]

Good tips from @KSTR. Tone board is perfect to go direct to a headamp, but apparently its grounding lay-out can be troublesome in some situations: https://blog.jdslabs.com/2018/11/jds-labs-atom-amp-official-release-benchmarks/#comment-22949 looks like you're running into such a situation.

Apparently all of the grounding issues manifested in my case but the Topping D50S did even worse. Just out of curiosity tried the same things and the result was the same.

 

[yavormoskov]

2) Insert an USB Isolator that has enough power to supply the Khadas. Intona's USB Isolator might work well and is reasonably priced though not exactly cheap (I use it on a daily basis but haven't tried bus-powering anything from it, specs say it is good for at least 300mA).

Apparently Khadas requires 5v/500mA which possibly could not be enough. Intona says maximum 500mA but this is the upper limit. Will email them.

 

[Veri]

Apparently Khadas requires 500mA which possibly could not be enough. Intona says 500mA but this is the upper limit. Will email them.

View attachment 37141

I have an Industrial ver. Intona (USB2 not newer more expensive USB3) but don't own the KTB to test. In limited testing, dongles at max output start to lose connection/drop out though, so I'm not sure it will do for a DAC at max volume if it needs to full USB power. Hmm.

 

[Wombat]

FWIW< https://www.soundonsound.com/techniques/solving-computer-audio-problems

 

[yavormoskov]

I also found this very interesting device ALLDAQ ADQ-USB 3.0-ISO-PS / 3.0 Isolator. Made and engineered in Germany. And also a video review about it by Hans Beekhuyzen. The only difference is that the review was done three years ago and back then this isolator was USB 2.0 and now they sell only the upgraded USB 3.0

 

[Severian]

FWIW the iFi iDefender3.0 solved my USB ground loop issues. It's $50, so far less than the full galvanic isolators. I previously tried external powered hubs and all manner of separate grounding wires spanning the various pieces of hardware and none of it worked.

Using optical for the DAC connection also works by severing the galvanic connection, but obviously your DAC needs optical input and that's not always practical.

 

[Veri]

Perhaps a cheap optical>coaxical converter box with optical source from desktop could fix the noise problem for a low price?
In case the KTB works over coaxical in your set-up.

 

[yavormoskov]

FWIW the iFi iDefender3.0 solved my USB ground loop issues. It's $50, so far less than the full galvanic isolators. I previously tried external powered hubs and all manner of separate grounding wires spanning the various pieces of hardware and none of it worked.

Using optical for the DAC connection also works by severing the galvanic connection, but obviously your DAC needs optical input and that's not always practical.

Yes, it could but iFi replied to me that it could not sustain the 500mA power draw requirement on its own so an external Micro USB power adapter is required. And I don't want to deal with external power. If I deal with external power it is easier for me just to route he amplifier DC power supply inside the PC chassis. I want completely contained, closed system inside the PC case.

 

[yavormoskov]

Perhaps a cheap optical>coaxical converter box with optical source from desktop could fix the noise problem for a low price?
In case the KTB works over coaxical in your set-up.

Tried that already, it made no difference at all. After all, both the amplifier and the KTB are powered from the same computer and connected to each other with RCAs.

 

[Severian]

Yes, it could but iFi replied to me that it could not sustain the 500mA power draw requirement on its own so an external Micro USB power adapter is required. And I don't want to deal with external power. If I deal with external power it is easier for me just to route he amplifier DC power supply inside the PC chassis. I want completely contained, closed system inside the PC case.

Ah. It is true that in my case the iDefender requires external power, even when the audio device itself isn't bus-powered.

 

[Veri]

Tried that already, it made no difference at all. After all, both the amplifier and the KTB are powered from the same computer and connected to each other with RCAs.

Huh. I thought optical was supposed to be free of noise pollution since it uses light, it should be galvanically isolated by default.

 

[graz_lag]

I would not hesitate to buy this ground separator from Supra, Sweden, top top quality @ abt. 100 EUR in Europe ...
http://www.jenving.com/products/view/ags-10k-audio-4xrca-1021000045

 

[Veri]

I would not hesitate to buy this ground separator from Supra, Sweden, top top quality @ abt. 100 EUR in Europe ...
http://www.jenving.com/products/view/ags-10k-audio-4xrca-1021000045

So, a 1:1 transformer? Wouldn't this affect the THD by a lot?

 

[KSTR]

They don't spec any max level or distortion figure, but I would assume it can handle line-level signals (2Vrms) good enough. Measurable distortion, sure, at low frequencies, but is it audible? Probably not...

 

[graz_lag]

So, a 1:1 transformer? Wouldn't this affect the THD by a lot?

You're right, it might - theoretically, compromise the SQ, however a friend of mine has it, we did several listening comparisons with different genres of music, no ABX testing though, we heard no difference in the SQ / soundstage with or without the Supra in the chain, downstream his Topping D30 DAC.
The hiss and noise were totally gone, instead, so he then decided to keep it.

 

[Veri]

You're right, it might - theoretically, compromise the SQ, however a friend of mine has it, we did several listening comparisons with different genres of music, no ABX testing though, we heard no difference in the SQ / soundstage with or without the Supra in the chain, downstream his Topping D30 DAC.
The hiss and noise were totally gone, instead, so he then decided to keep it.

Well that's good to hear