Oh FFS it's a GROUND LOOP!

[just brew it!]

For the past couple of weeks I've been intermittently exploring a rabbit hole that I started going down due to an objectionable amount of static I was getting from my (work-issued) MacBook Pro's analog audio output when using external speakers. Since I did not seem to have any static issues with my main desktop when playing through the same speakers, I assumed the issue was the MBP.

Using an external USB audio adapter from my junk box helped, and using the external USB audio adapter through a USB power isolator completely got rid of the static. End of story? No, far from it...

I then noticed that my new 3700X desktop build (which I had not yet switched over to as my daily driver) was also exhibiting minor issues with audio static - not nearly as bad as the MBP had before my external adapter "fix", but it was definitely there. I tried an external USB adapter there as well, and like the MBP it helped some... but didn't completely fix the issue, and I didn't want to hack around with another USB power isolator. I then experimented with using the optical S/PDIF out to an inexpensive DAC, with good results. End of story NOW? No, still not yet...

It was bugging me that the noise issue seemed to be so prevalent; it just didn't seem right. So yesterday after work I decided to do a bit more experimentation using my "smoke test" box. I tried a discrete soundcard; I tried a hack-ish experiment hanging an audio DAC off of a spare HDMI port (using one of these); I tried a couple more external USB DACs (a cheap-ass one and a slightly better but still not great one). Similar results across the board. None of them sucked, but none were great either - the noise floor seemed higher than it should be.

Then FINALLY... I tried one of these. Holy crap, the static was gone! I tried the ground loop isolator on the analog outputs of each system in turn (including the MBP), and in every case it reduced the noise level to pretty darned close to zero.

It was a f*cking ground loop all along!

Other random thoughts and discoveries that resulted from all of this...

You may be wondering how I ended up with a ground loop. Well, I have multiple systems all running through a KVM. So everything is grounded through that, but also grounded through the ground connections of the PSUs. The desktop 2.1 speakers I use also have two inputs, which are normally connected to different systems. So there's likely another path to ground through the speakers. So ground loops galore (one of which runs directly back to the audio interfaces of the PCs). Whee!

Ground loop issues aside, I still intend to add a better audio interface for my 3700X system (either moving over the Xonar DSX from my old daily driver or an external solution), since the on-board ALC1220 on the 3700X system doesn't play nice with the Linux JACK audio stack (random audio stutters and hangs).

Still puzzled why my current daily driver doesn't seem to have static issues at all. Maybe the Xonar DSX card has integrated ground loop isolation on its analog outputs? I kind of doubt it. More experimentation seems to be in order...

As part of my experimentation, I also did some informal tests of mic input noise. Turns out that mic inputs on on-board audio solutions and even some discrete sound cards can be stupidly noisy, with tons of hiss. But the mic input on this cheap-ass USB audio dongle is surprisingly quiet (go figure).

Guess I'll be ordering a pile of extra ground loop isolators...

[SecretSquirrel]

Still puzzled why my current daily driver doesn't seem to have static issues at all. Maybe the Xonar DSX card has integrated ground loop isolation on its analog outputs? I kind of doubt it. More experimentation seems to be in order...

Points towards the loop path being through your daily driver.

--SS

[just brew it!]

Points towards the loop path being through your daily driver.

Wouldn't it tend to be affected also? All of the KVM and audio cables have ground paths in them. I would think the KVM's and/or speakers' ground potential would be bouncing all over the place with respect to EVERYTHING. (And FWIW using the KVM's audio path definitely makes the issue worse, but running the cable direct only eliminates about half of the noise; definitely need to cut the direct ground path in the audio cable with one of those isolator doohickeys or a fiber optic S/PDIF connection to truly make it go away.)

[SecretSquirrel]

Hard to say, without knowing exactly how everything is connected and which systems have actual AC ground connections. The MacBook, for example, will be completely isolated from AC ground. If running it on battery, with USB/Video disconnected, and only the audio running through the KVM, still results in noise, then the loop problem is in the KVM and speakers. I don't remember if the ProMedias have a fully isolated supply or not -- I'm assuming the 2.1s use the same general design as the 4.1s I have. If the DC "ground" for the power supplies for the KVM, and speakers and such, isn't isolated from AC neutral, then you can get a pretty big ground differential between them and something that has a true AC ground, like a computer power supply. The loop goes all the way back to the breaker box. If the second computer you tested wasn't plugged into the same power strip/outlet, then it's ground path to your daily driver would also potentially go all the way back to the breaker box.

Figuring out the exact path/cause is crazy hard when you have so many different devices involved.

--SS

[Oops, hit edit rather than quote by mistake and mangled your post. I think I got the part you posted back the way it was. --JBI ]

[just brew it!]

Speakers, KVM, and MBP all have isolated supplies, so the loop(s) must run through some combination of the desktop systems and monitors. Untangling the rats nest of cables and power cords to figure out where everything is plugged in is probably part of the next phase of investigation.

For now I'll probably just order a couple more of the audio ground loop isolators. They're cheap, and (based on a few quick listening tests) don't seem to be degrading the frequency response noticeably.

[SecretSquirrel]

[Oops, hit edit rather than quote by mistake and mangled your post. I think I got the part you posted back the way it was. --JBI ]

With great power....