USB ground loop isolation?

Audio Gear Talk Raspberry Pi
1

First of all, please excuse my ignorance here – I’m no electronics or even electrics expert, so some of my questions are probably really dumb.

I run a Pi 3B+ using WiFi and Ropieee with USB out to a Mytek Liberty. I’ve been reading about USB ground loops, thinking I might put a ground loop isolator between my Pi’s USB output and my DAC’s USB input. However, after reading up on USB ground loops, I think this is unnecessary for three reasons:

  1. The DAC should be doing electrical isolation here. I suspect it shares the USB isolation properties of its bigger siblings.

  2. Both the Pi and the DAC are plugged into the same powerstrip, so they share the same ground, so the ground loop is pretty dang small.

  3. I hear no hum playing silence over the circuit with my amp at max volume.

Any experts care to comment?

2

If it ain’t broke. (Not an audio expert but advice that seems to be universally applicable)

1 Like
3

I won’t claim to be an expert in this field - because, well, I’m not :). Electronics inside computers are not my thing.
But I do know my way around electronics in other applications.

First advice was a fine one. Do you have a problem ? Then go and fix it. You don’t have a problem ? Well, then you don’t have anything to fix, do you ?

By the way : the fact that both devices are plugged into the same power strips, tells you nothing about the odds on a groundloop.
Check your Pi for instance : I’m willing to bet that it doesn’t even touch the ground on your powerstrip. (Correct ?)

I don’t know this for sure, but it would surprise me, if the data stream and audio stream share a common ground.
In other words : a ground loop between USB input and ‘something else’, should not necessarily mean that the analog audio is affected.

Somehow, I have the feeling that I’m answering a question that you have already answered for yourself :slight_smile: .

1 Like
4

Yes, I looked at the power supply (a standard RPi power supply) and it only has two prongs, so no ground.

I think that’s what I meant (though somewhat fuzzy about this in my own mind) when I talked about the Mytek DAC’s isolation capabilities.

5

One relatively cheap way to try to see if breaking the ground will help is to get an IFI Idefender. This will clean up any grunge on the USB hub as well and if you attach a good clean 5V via the mini USB plug in the side you’ll get rid of any noise from the USB port.

6

Thanks, Mark. That’s what I was originally thinking. But I don’t think there really is any noise over USB – it’s all digital, so analog noise isn’t really an issue. What is an issue is ground loops, and Marco has convinced me that they really aren’t an issue either (in this specific case). The iDefender would make more sense to me if the computer the data was coming from had its own ground plug.

7

Hi, I have become a bit wiser the last couple of years and the problems with interference is far worse than all digital guys claim. Pls. consider below as very well tested, however never measured statements. It have not been able to measure since we have no rfi free room, where we can perform validation tests.

  1. The Raspberry Pi is highly sensitive to rfi noise itself. The many housing in acrylic is of no other use that to keep things together and tidy. If you want your RPi clean, you must prevent airborne rfi and prevent the shields of any cable to dump shield and adjacent device noise in to the RPi.
  2. It is highly responsive to clean power. I suggest a linear, very stable, psu, but the Ifi iPower is not all that bad.
  3. If the DAC you are using relies on the USB 5V/DC, you should break the circuit and inject a far better 5V. It should be done very close to the DAC, not at the USB output on the RPi.
  4. RPi use a single bus for both USB and network port. The shield noise is highly contagious, since they are all soldered to a commen ground on the board.
  5. I recommend using a good enterprice switch, fed by 230 Volt AC mains with protective earth and a very short ethernet patch cable to RPi. The port shields must be low impedance grounded in the switch, max 2-4 ohm. You will then drain shield noise in the switch rather than having it glittering in the RPi. This noise spreads inside by stray capacitance and is almost fluent in behavior at 2,4 to 5 GHz. Shield and multiple grounding points, not letting your DAC be the first,

A few but important lessons I have learned from this.

2 Likes
8

Stefan, thanks for your input!

There are some measurements here you might be interested in:

  1. The Raspberry Pi is highly sensitive to rfi noise itself. The many housing in acrylic is of no other use that to keep things together and tidy. If you want your RPi clean, you must prevent airborne rfi and prevent the shields of any cable to dump shield and adjacent device noise in to the RPi.

As long as you’re only using the Pi for digital processing (no DAC hat), this should be of minimal influence. There is a rather nice Adafruit case which should form a good RFI shield, once properly grounded:

  1. If the DAC you are using relies on the USB 5V/DC, you should break the circuit and inject a far better 5V. It should be done very close to the DAC, not at the USB output on the RPi.

Indeed, particularly for older DACs. Again, there are some measurements:

  1. RPi use a single bus for both USB and network port. The shield noise is highly contagious, since they are all soldered to a commen ground on the board.

Yes, a real problem with RPi 3, but it seems to be fixed in the RPi 4.

  1. I recommend using a good enterprice switch, fed by 230 Volt AC mains with protective earth and a very short ethernet patch cable to RPi. The port shields must be low impedance grounded in the switch, max 2-4 ohm. You will then drain shield noise in the switch rather than having it glittering in the RPi. This noise spreads inside by stray capacitance and is almost fluent in behavior at 2,4 to 5 GHz. Shield and multiple grounding points, not letting your DAC be the first,

Good to know.

1 Like
9

I did many test in the past, trying to be as objective as possible. Since so many people where talking about computer noise I did some test. Play a CD while my audiocomputer is completely turned off and all cables disconnected. After that, play the same CD with the computer turned on, no difference. Now connect all cables, no difference. Stress the computer to 100%, no difference.

Now playback on the computer with only playback software running and minimized services. After that same songs but with all services on again and stress the PC to 100%, no difference, none at all.

Pretty solid statement when you have never actually measured anything!

2 Likes
10

Don’t blaim Bills quote of my statement:

Agreed, so that comes from in depth comparisons of mutual experience with my mentor and a hifi builder since at least 50 years and discussions with my dear friend Mattias Stridbeck, the founder of Harmony Design as well as discussions with Ben Duncan Audio Research in UK. They are all together with me shit-for-brains-snake-oil-preachers, so just discard from the statement.

Ben, btw, measured a self damping property of Supras LoRad mains cables from noise. He actually has a rfi sealed lab… and is no friend of computer playback unless substantial measures are taken to limit the ground noise as well as voltage ripple from the computer.

11

The Audio Injector Isolated DAC & ADC just dropped on Kickstarter.
It may be of interest to people who are using the Pi and have noise issues.

Check the project here :