Stopped noise from nuc8i7, usb out, by disconnecting AC ground

I just finished putting together my first Roon/ ROCK installaton on a new NUC8i7 (in a Turing case) with ethernet connection to xfinity router. Equipment powered through an APC conditioner. Very noisy listening with two of the three usb DACs I tried wired to the nuc8, whether music playing or not. Noise increases with volume.

First dac, S.M.S.L M8 ver. 1 (external switching ps). Second was a Questyle dap in dac mode (battery power). Both had a high level of random noise, and a regular beep at about 1Hz (about 30 dB down) Very audible. The Questyle was only slighty better of the two.

But a Grace m903 was was immune, with no noise heard up to full volume. Prefer not to use this one for conversion though.

I tried the m8 powered from a wall socket without improvement, but it was an adapter to remove the ac ground connection on the nuc’s power cord that eliminated all audible noise.

So, happy listening now, but not sure this was ideal solution. Kind of like to figure out what’s going on. Maybe I’ll try cutting usb power circuit.

fyi- I have been searching this problem, but didn’t see much.

Have you verified the NUC supply does indeed have the earth side of the mains connected to the DC (-) side output?

Good question! So I just checked, and ac outlet polarity looks OK.

From a review of the Grace m903: The USB interface is also isolated from the M903’s audio ground to avoid ground loops with the noisy computer ground.

Sounds like a properly designed unit.

You may have something there. The m903 appears fo be a far more robust design in general.

Having said that, upon hearing the noise with the other DACs, it was hard not to believe the NUC was broken, as it sounded like it was talking to me, with all the beeping.

The M8 is not an overly expensive instrument. Perhaps the isolation design is just poorly done? Perhaps the designers were more familiar with the analog waveform creation part, which is after all the essential premise of a DAC.

As for the Questyle DAP (assuming the QP2R?), there’s a bit of chatter on the Web about USB mode, which seems to be an afterthought in the overall design. Like https://forum.headphones.com/t/questyle-qp2r-official-thread/2053/21. So the battery is still charging, apparently. I’m a little surprised you’re hearing things there, but the suggestion is to try different USB cables.

Perhaps if you listen to it long enough, you will begin to understand what it’s saying? :grinning:

Yes, it’s professional level gear from a respected shop, similar to the Mytek stuff. Also 10 times the price of the M8.

I had the same problem with a much older NUC into my Meridian 818v3. I just gave up using USB in that configuration.

My dCS Rossini is immune.

I appreciate all the feedback. Still don’t know why the noise showed up with the NUC. All DACs were noisless with an ancient windows7 laptop running Foobar2000 with various drivers for several years.

Maybe I’ll try modifying the usb cable to see what happens. I have tried a few others that I have around. I suppose I could be more selective choosing DACs, but then can I reliably predict which will work?

The AC ground to the NUC remains disconnected for the time being.

That is an excellent question. These guys who test DACs, like @Amir_Majidimehr and @Archimago, don’t seem to test the input signal conditioning. I think they use inputs that are as clean as possible, to focus on the D/A conversion part of the DAC, but that seems a bit one-sided.

Perhaps it’s difficult to cobble up a test input USB output with intentional noise on it, but I do wish they’d figure it out and start testing it.

And, while I’m thinking about it, it would be great to test the power supply isolation as well, by connecting a power supply that’s intentionally noisy.

Hmmm, sounds like an idiosyncratic issue with the NUC unfortunately especially if @denydog was not having issues with previous Win 7 laptop. Certainly points to issues with grounding given the cure was to use something like a “cheater plug”; clearly some deficiencies with specific DACs.

Actually, I do not use clean USB input for my testing :slight_smile:. For example this post from late 2018 was between an older i7 computer vs. Raspberry Pi for example to demonstrate that my circa 2013 TEAC UD-501 DAC did not have any significant issues:

Before this, back in August 2016, I compared the NUC, i7 HTPC, ODROID SBC, and Surface laptop even using the inexpensive Light Harmonic Geek Out V2 DAC powered off the USB port itself:

I do still have that NUC6i5 if anyone has a hypothesis about what the issues might be or with what kind of DAC! Not sure if we would even see such issues with the 6th Gen i5 though.

Yes, I’ve read that, and most everything else on your blog. Thanks! I love this latest round of amplifier testing.

But with the i7 test – you don’t know if that was a really “dirty” input, did you? It’s just a computer that a lot of people consider “dirty”, but maybe really isn’t. You say,

Surely there must be all kinds of electrical noise that gets generated and passed down across the USB cable to the DAC, right?

But you don’t actually measure that noise, do you? And, sure enough, the TEAC (a nice DAC) isn’t affected by the noise that may or may not be there. I’d bet the Grace m903 and my Mytek Liberty would also be unaffected. Because they’re well designed.

I’m looking for more of a controlled test with maybe even multiple levels of noise over the cables. Same for jitter – a sender with really bad jitter would be different from one with really low jitter. Stress the input isolation and see if it works.

DACs powered off the USB port itself, like the OP’s DAP, are especially problematic, and it’s particularly important to test the noise isolation on them.

All that being said, the OP’s M8 and QP2R might simply be susceptible to a bad port or cable. I’d try all the ports on the NUC and a couple of different USB cables.

First thing I did was try all the NUC’s USB ports. They all sound the same. Different cables = no change.

It’s not ideal that the Questyle DAP/DAC was exposed to USB power when I tried it. I think it would be more informative operating on battery only. I will modify a USB cable tomorrow. I’d rather cut the USB power to the DACs than the AC ground, if I can get the same result.

I’ve read something elsewhere that has me questioning the NUC’s external power brick. I’ll see if I have something around to to swap in for a check.

Yes, very true Bill, I have not specifically looked at the noise of the USB port from the computer itself. Will need to specifically “inject” that noise into the power line for example to get a controlled reading… Not sure how that would be done properly and safely except in a lab with specific equipment.

What might be more interesting would be for someone to have a look at the noise level of the NUC8 in question sort of like in this post:

Maybe the NUC would look horrible even compared to some of those!

As for jitter… Hmmm, unless there is loss of data integrity, I think with a decent modern asynchronous interface one would not see any significant effect. It just doesn’t work in such a fragile fashion as some in the audiophile world would want us to believe.

Me either. I thought one of you big brains could probably figure out something, though.

I believe this, too, but again, controlled testing would allow us to see if the engineering is up to snuff. If the device actually has a “decent modern asynchronous interface”.

@denydog, the only thing I can suggest at the moment is the standard consumer advice: buy equipment from established companies with good reputations, like Mytek or Grace Labs. To that I might add, look for companies with a relatively narrow focus on the kind of equipment you’re interested in purchasing.

Good advice Bill.

Just to update: I tried restoring the NUC8’s AC power ground connection, and instead disconnecting the USB cable’s power and ground connections. This also reduced the noise, but not to the same degree. I could hear noise from the speaker from a couple feet away, with the system volume control maxed (nothing playing of course).

I did notice that as I maxed out the volume, the noise got louder for a moment, and then settled to the lower level just described. This is with the SMSL DAC. The Grace m903 is still immune to the noise.

At this point, I only wish I had another NUC8 to try, or had a way to check that this one is not misbehaving. The old laptop didn’t cause the same noise in the SMSL, but then it had an ac adapter without a ground connection (and a battery).

I’m back to using the NUC power supply without the ground connected. It stops all audible static/beeping noise. I’m thinking about a new DAC, but I’d better test it with my setup.

I suspect you have a ground loop.

Try bypassing the conditioner (the use of this is most suspicious), unplugging all power cords to non-essential equipment, and connect the power of the minimal equipment for your test setup to the same power bar (calculate power needs and beware of electric overload). Unplug all HDMI cables, and coaxial antenna cable for TV (if any). Unplug all RCA cables as well except for the ones you use to test your setup. If you have a network switch, use it instead of the router, and unplug the broadband for this test.

If you get no noise from this minimal setup, then you can gradually restore things to see which gives you the ground loop.

2 Likes