This is gonna be a long post. Have the exact same setup as you (with a tiny modification) and faced the exact same problem. Did the same solution as you and it's a lot better now. Hope this can help out some people in the future. I put a lot of detail and go into a lot of side notes, so skip to end for TLDR.
My setup is PC USB out > USB into Topping D50s > RCA out from D50s into a 2-way RCA splitter > 1 pair of RCAs into SMSL SP200 THX 888 / 1 pair of RCAs into DarkVoice 336SE.
A single grounded electrical outlet feeds a large power strip that powers my PC, 2 monitors, a router, a fan, and a daisy chained secondary power strip. That secondary power strip is for my D50s, SP200, and a DarkVoice 336SE.
I realized if I power my D50s through my PC USB, it always sounds like a poorly tuned AM radio from a 1940s VW Beetle. I bought a generic 12000 mAH 2A USB power bank and it can perfectly power the D50s for maybe 2 hours, but can't charge the battery bank at the same time it is being used, so I just plug the D50s into a generic phone USB wall adapter from (a very cheap Samsung phone) to supply 5V DC at 1A over USB. There is no discernible difference between using a battery bank or a wall adapter.
With the D50s (probably) properly powered and grounded with the SP200, half of the noise problems were solved.
Running the D50s at 0dB and SP200 alone with PC USB disconnected (and even over Bluetooth from my Galaxy Note 8!), there was absolutely no audible noise whatsoever until I cranked the SP200 beyond 70% volume on high gain. I was able to cleanly listen to songs over Bluetooth at -45dB on the D50S and 70% volume on high gain on the SP200. Even with an unrealistic listening setup, this is a very impressive showing for both devices.
Setting the SP200 at 100% volume on high gain, the crackly static of the noise floor is only just starting to transition from faintly audible to moderately audible but still impressively quiet nonetheless. I've listened to hours of worse audio quality over FM radio. Of course, 100% volume on high gain from a SP200 is unrealistic for any normal purposes. I have no headphones that ever need more than 45% volume on low gain, so that noise floor at full crank high gain was never going to be an audible issue. I'll very happily call this "dead silence" because it's the best I'm going to get.
If I turn on my DarkVoice 336SE with PC USB disconnected, there is an additional layer of noise through the SP200. However, I haven't tested it to see if it just needs to warm up or if it is persistent. Plus, the reverse is not true. At 0dB on the D50s, my DarkVoice 336SE has ZERO noise at full volume even when the SP200 is on with PC USB disconnected (maybe because of the huge input transformer it has? Just a guess but let me know if I'm wrong). It's a non-issue either way, because I never use both at the same time.
Two side points that obsessively compulsive listeners will probably dislike me for letting them know:
Firstly, if you are not frequently using D50S Bluetooth, you may want to turn it off. Even when the unit is in another mode (USB, Optical, Coax), the Bluetooth receiver is still active and having Bluetooth on will cause occasional radio noise in the output. I noticed this when I did my noise floor testing and heard very slightly audible radio interference at -45dB on the D50s and 70% volume on high gain on the SP200. Objectively, it is impossible to notice at normal listening levels but it is interference nonetheless, and it subconsciously bothered me to learn of its existence so I now disable Bluetooth when not in use.
Secondly, during noise floor testing, I was
shocked to discover that literally having the screen of the D50s active and interacting with the menus/settings on it actually create an audible tone. I always had my D50s screen in Auto mode, so I never really noticed it before because the screen is usually off, but when I was disabling Bluetooth using the knob, I heard the screen turn on, heard each time I made an input or scrolled the menus, heard very clearly when the Bluetooth chipset was disabled, and heard the screen turn off through the SP200 at sufficiently high volume. Try it for yourself, but I know for a fact that I heard it and I'm not crazy.
Back to the ground loop issue...
The instant the PC USB was connected, I got continuous ground loop noise you described when any processing at all happens. Mouse movements, file transfers, background processes, and pretty much anything that causes the CPU led on my PC's front panel to light up. In fact, just having the PC on at all causes noise. Different frequencies were being transmitted and very audible through the D50s at 40% volume on low gain on my SP200. Disconnecting the PC USB = dead silence. Strangely, booting straight into UEFI BIOS and not doing anything results in
almost dead silence as well. There will be the very occasional crackle every 30 seconds or so, when the processor does something but it's close to the unconnected PC dead silence benchmark.
I disassembled my PC, rerouted wires, disconnected the front audio inputs and USB, crossed data wires and power wires at right angles to reduce EMI, reseated the GPU, moved all possible wires as far away from it as possible, and put everything back together. Aaaaaaaand............ USB ground loop still present, but my PC looks a little nicer now so hey, I've got that going for me.
After pulling my hair out for 2 days, I had narrowed it down to six options:
1) Invest $50-$150 and get a PCI x1 sound card which can output optical to completely bypass USB ground loops. Not possible in my case, because my motherboard has 1 PCIE x16 and 2 PCIE x1 slots, and the graphics card is a 2.5 slot which completely covers the PCIE x1 ports. This may be an option for other people who have better motherboards.
I'm talking completely from speculation, but even if I could get a PCIE x1 riser cable and connect a sound card, I would be afraid that that being in such close proximity to my graphics card (less than an inch) would induce EMI and noise in the sound card itself, and the optical out would still carry a noisy signal. Plus some people say that some sound cards process the signal even if it's optical out, which will bottleneck sound quality.
Someone knowledgeable should correct me on this if I'm wrong. My PC is a handbuilt bucket of scrap from a decade ago, so I'm not planning to invest any further $$$ in it, but I really do want to know if quality sound cards are worth it for my next build.
2) Buy a new, less noisy motherboard. If I decide to replace the motherboard, I'm spending $1500+ and replacing the entire PC, which is an overkill solution for this problem. Also, there is no guarantee that the issue would be solved. Even if someone gives an anecdote that their particular motherboard works flawlessly, unless I have a duplicate of their PC down to the peripherals and monitors (literally everything connected to the PC) there is no guarantee that the parts in my specific PC will play nicely with the power supply, each other and the USB chipset in the motherboard.
3) Invest $30-$200 on a USB to Optical converter. Again, if the USB ground itself is problematic, I don't know if it would solve the issue or propagate the noise through optical. Not willing to buy it unless I can guarantee it will work.
4) Buy a cheap powered USB hub for $20-$100. This is the option most suggested as a solution. However, as an electrical engineer who has made circuits and created devices that were powered/communicated over USB, I am skeptical that this actually does what I need it to do. I could be completely wrong and if any computer engineers, audio engineers, or electronics engineers want to correct me, please do. My theory is:
At the principle level, I believe all a powered hub does is pass through the two data lines from the PC USB and shunt the PC USB power and ground lines through some sort of auxiliary circuit so the PC says "hey something's connected, let me send data on the data lines". The powered hub then rectifies +5V DC from mains power to pass through to the hub bus. Thus cleaner 5V power is supplied that is free of motherboard noise, and each device in the entire hub has simultaneous access to the full 500mA specified in the USB protocols in comparison to what a normal PC USB root could produce (500mA maximum shared between all devices on that root).
Since this 5V is grounded on the same potential as the DAC, this
would solve some ground loop issues that are caused by
powering the DAC from the PC. But I already solved that problem by simply externally powering the DAC from a mains wall USB adapter. All I need now is clean data over USB. Problem is that the USB data is always referenced between the motherboard USB 5V (which I think comes off the USB controller chip) and the ground of power supply unit (PSU), and is
NOT and can never be referenced to the 5V and ground generated by a mains adapter or a powered USB hub.
A PSU does not need to produce pure, clean and unwavering voltage for the PC as a whole to function because component tolerances are fairly forgiving. Most components/peripherals will usually work fine on up to +/- 1V because not everyone has $300-$1000 to spend on an 80+ titanium certified PSU. I don't even know if those never ever produce consistently level power. Of course, when any device is used (e.g mouse movement, keystroke, fan ramp, hard drive spin up, processor activity, etc), the total current draw on the PSU will change and so the USB voltage will also change slightly since everything is electrically connected and usually poorly isolated.
Now that I think about it, I realize that the motherboard is equally culpable because even if the PSU is supplying surgically precise 12V into the motherboard, it then has to generate the USB 5V from its main power distribution circuit which is shared with all components. In budget motherboards those parts are usually made to conform as minimally as possible to spec to save money. Ain't no one making affordable USB controller chipsets to supply 5.00V +/- 0.01% tolerance because it isn't a scientific instrument or signal generator designed by the Texas Instruments R&D department for NASA that needs to be connected to the grid of the International Space Station.
A powered USB hub will do nothing for data integrity since the 5V from the PC USB root is not the same as the 5V from mains. There will always be some slight difference in potential, maybe a few millivolts, maybe even up to 1/4 volt if your PSU is from a Chinese bargain bin or on it's last lap before it dies. There are then three options:
One, connect the PC USB ground and powered hub ground together. This is the worst option, as the ground loop is not resolved and is functionally the same as using a non-powered USB hub.
Two, connect the PC USB 5V to some "smart" circuit so it can handshake with the PC and declare itself as a device to your operating system, then close that part of the circuit to PC USB ground. I don't know why it would matter for a USB hub to do this, but I can imagine a couple manufacturers doing it regardless.
Three, leave the PC USB ground floating (unconnected) to prevent the ground loop. If the data bus is referenced to PSU ground and now you're leaving the PC USB ground floating, when the DAC references the data input to mains ground there
will be an elevation or dip in the signal, depending on which side is higher or lower potential. That's just plain electrical theory. E.g: if the data+ pin of the PC USB is 400mV referenced to PSU ground, and the ground of the PSU is 5mV higher potential than the mains ground that the DAC and powered USB hub are connected to, then isolating the PC USB ground will result in the data+ now being 405mV with reference to DAC and USB hub ground.
My knowledge on this is rusty but after googling a bit, it theoretically
shouldn't be a problem because at the PC USB side, the data pins operate at a differential of 400mV, while at the receiver side only needs to acknowledge a difference of 200mV betwen data+ and data- to register a bit. The PC USB voltage power pin is still at its 5V potential (with respect to PSU ground) fluctuating slightly with processing tasks (duh, that circuit is what caused all the noise in the first place). So you will have a signal that is unaffected by motherboard noise because the 5V has no ground to go to. You still still have to connect that 5V to something without it affecting the signal or the PC USB connection won't work. I have no clue how actual powered USB hubs handle this part (shunt USB 5V to mains ground and return hub 5V to PC USB ground? pass PC USB 5V through resistive circuit back to PC USB ground? ICs? filters? set the starting parameters of a newly universe to result the development of a better connector standard than USB and then tunnel that data through extradimensional portals directly to the output data lines?).
The only way I can see a powered hub solving this issue is if it continuously identifies the tiny potential differences in PC USB and powered USB hub ground and somehow perfectly reproduces and reclocks the PC USB data in reference to the mains ground that it itself is referenced to, which I find difficult to believe works well (or at all) in anything costing less than $100. These cheap powered hubs are made for college students to throw in their backpack so they can plug in their iPhones, battery banks and external drives into their MacBook with a single thunderbolt to USB adapter or whatever. No way they're designed to isolate, clean up and perfectly reproduce a data signal at flucuating potentials over USB. Plus that would necessarily add latency and alter the signal, which leads to the next option...
5) Gamble $100-$600 on a "black box" USB cleaner/isolator/reclocker/decrapifier/purifier/filter/panacea solution (e.g. Schiit Wyrd, iFi iPurifier / iDefender, etc). The kind of products that people swear makes the USB signal purer than a consecrated virgin drinking icewater freshly melted from a crystalline polar glacier whilst playing the harp at midnight during the celestial alignment of the 9 planets in the solar system.
I'm obviously not a seasoned audiophile and I'm low on the totem pole of equipment so I'd rather put that money towards a nicer pair of headphones in the upper mid-fi range instead of trying to "clarify" my dirty USB ports with the power of nether energies and astral physics. I've never used one before and I'm sure it would do... something... maybe.... Fix the issue? Possibly, but who knows for sure. In my opinion, this is treating the symptom rather than curing the root problem.
6) Buy a Corning USB 3.Optical cable and feed the D50s through a powered USB hub grounded to the same circuit as the D50s which will 100% solve the ground loop by bidirectionally optoisolating the PC USB from D50s USB input, transcoding the USB signal into an optical signal, and then re-transcoding it again into USB at the other end (
https://www.corning.com/microsites/coc/ocbc/Documents/CNT-075-AEN.pdf). It's basically a single port self-powered USB to Optical and back to USB solution which supposedly works with "most" USB 3.0 and 2.0 devices (I wonder why not all...).
This is an excellent article with measurements where I got the info from:
http://archimago.blogspot.com/2015/05/measurements-corning-usb-3-optical.html#more.
However this cable is
unreasonably expensive, only comes in 33 ft. and 50 ft. lengths, and is apparently prone to unexpected failure in less than a year based off the 2.7 star Amazon ratings over 49 reviews. Plus it is borderline unusable as a regular USB 3.0 cable since it basically throws the USB 3.0 standards in a fire and proceeds to do its own thing. This $154.41 cable is literally only useful as an audio interconnect into an external USB hub to filter USB ground noise. But apparently when it works and does what it supposed to do, it is the perfect solution and there is nothing better except literally sitting in the studio (or you know, using better equipment without ground loops).
Spending close to what I paid for my DAC or more than what I paid for most of my headphones (HD 6xx, K7xx, 1More Triple Over Ears, 1More Triple IEMs, SHP9500s, Tin T2 IEMs, Mdr-cd900st) on a fragile cable and a random powered USB hub is not the most appealing option to me, even if it will 100% work. These are hard choices for a new audio lover trying to maintain some money in the pocket. I'll buy it for sure it if it ever goes to $99 or they make a shorter one. If anyone knows of a similar product that is cheaper, I beg you to
please let me know. This choice may be different for others.
So what to do?
This is the straightforward (and basically free) "solution" that at least eliminated the motherboard processing noises by leaving the USB ground isolated (big thanks to Audimon). It is not a fix, but it dramatically reduces the problem to a manageable one. Make sure you plug out the cable from the DAC and PC before you do this, or you might short out something.
Use a sharp scissors to cut a very thin strip (maybe 2-3 mm) of electrical tape that is large enough to cover the ground pin (pin 4) of the male USB A cable that plugs into your PC.
With the USB symbol pointing upwards, pin 4 is on the extreme left (top left in diagram). Use a rigid and thin object to gently push the tape all the way to the back and flatten it out over the entire length of pin 4 (like a paper clip, toothpick, or one blade of the scissors). It should not cover pin 3. Shine a light into the connector and look to ensure it is fully covered. Then simply attach any excess tape to the bottom of the connector and slowly insert into the PC USB port. Nothing needs to be done at the D50s end except reconnect it.
If pin 1 is covered by mistake (or intentionally in my case), the D50s will not receive a signal because the PC simply will not acknowledge a USB device is connected if the 5V is isolated. However, there should be no damage caused if this is done.
View attachment 70596
Plugging in this ungrounded PC USB to the D50s is a night and day difference. There is no more PC processing noise at all. Gone completely.
However, there is not dead silence either.
Instead of signal static, there is now a constant tone. It is not as oppressive as the USB ground loop static, and becomes audible at 55% volume on low gain on my SP200. Even going to -99dB on the D50s does not affect the tone in any way. I can live with this. At least until I figure out a permanent solution.
The tone is present at 50% volume on low gain on the SP200, but only when nothing is playing. Above that, I would blow out my eardrums before the tone overpowers the music. So I am satisfied with the outcome not because the problem was resolved but only because it is now so hard to notice that it might as well just be added to the "dead silence" benchmark. I'll call it "mediocre silence".
After testing, I found out that shutting down/sleeping the PC or the D50s while in use, then turning it back on will remove the tone UNTIL the next sound is played. Then it will return indefinitely until the PC or D50s is shut down/sleeping.
Why has severing the grounds between PC and D50s created a new constant tone? Because unfortunately, the PC USB still has 5V going to the D50s. It
must be doing something with it because the PC USB is working. If the PC USB 5V was unconnected then it would refuse to communicate with the D50s.
Because it is a single frequency tone that isn't flucuating discernibly, I would think that this confirms that yes, the USB ground is the root cause because it is is a horrifically noisy reference and the 5V (at least from my motherboard) is actually fairly stable and decent. Kind of raises questions such as why is the ground apparently fluctuating instead of the 5V but an audio engineer, electronics engineer or computer engineer would have to hypothesize on that.
Almost positively I suspect what is happening is that the D50s ties the PC USB 5V to the external power adapter bus 5V, causing a minor voltage imbalance of a couple mV somewhere in the internal circuitry, but I don't have the motivation to use my multimeter to verify if there is voltage on the unconnected side and I don't particularly want to brick the D50s by poking around in the USB receiver and bridging it with a voltage lead.
This is a hypothetical as I have no clue about DAC design, but perhaps one of the ICs in the DAC that is responsible for a dead silent noise floor needs exactly 3.33V +/- 0.05V with respect to ground and the internal power supply is quality tested in the lab and verified to produce that voltage within a tight margin of tolerance. Throwing it off by 0.08V or whatever by having two slightly different input voltage inputs might significantly throw off the op-amp gains and cause it to heighten the noise floor by several or tens of dB, which could be the source of the tone.
Either that, or the two voltage pins are not separated with diodes and the 5V from the PC USB and the external wall adapter are causing some minor current flow from the wall adapter 5V side up through the PC USB and to motherboard ground, or from the PC USB 5V through the D50s and to mains ground. If that is the case, that current flow might be getting amplified through the op-amp gains and might be what is resulting in a tone being generated at the noise floor of the D50s.
Pure speculation on my part, but the (almost but not quite) inaudible tone at normal listening levels is here to stay for the time being.
EDIT:
It appears from persons more knowledgeable than I that the 8 kHz noise is likely data packet noise along the USB bus, and not the fault of the 5V USB input? I am inclined to say I don't think I heard that tone before I taped the ground over...
TLDR
1)
USB noise via ground loop is a problem with the D50s and certain PCs.
Any activity that the PC performs can be heard through the D50s as a raspy electronic noise. Disconnecting or shutting down the PC results in dead silence.
2)
Solving the problem requires full galvanic or optical isolation of the PC USB ground and the D50s ground.
This could be done in several ways.
-Getting a sound card with optical out might be one.
-Getting a USB to Optical converter might be another.
-Getting a better motherboard is an option, but I don't know if it's possible to know if it will 100% fix the issue before you buy it.
-A powered USB will clean up the D50s signal significantly if it was being powered from the PC. Otherwise, just connect the D50s to a phone wall USB adapter that can provide 5V DC at 1A.
-Buying a fancy USB cleaner/purifier/decrapifier/isolator might work. It might not. I don't know. Let me know if you do.
-
For certain, buying a Corning 3.Optical optoisolated USB cable and feeding it into a decent powered USB hub (not the cheapest Chinese no-name garbage you can find) will supply the D50s with clean, ground loop free, noise free input. It is a relatively expensive option.
3)
Treating the problem is a 5 minute near-zero cost process.
Simply follow the instructions outlined a little further up that explain how to isolate the ground loop by putting a small piece of electric tape over the ground pin of the USB connector on the PC side. This will isolate the D50s from the PC's noisy ground, and you will no longer hear processing noises. You will however introduce a new tone (apparently at 8300 Hz). The new tone is not offensive in normal listening conditions, but it is not dead silent either. You will be able to hear it at dangerous listening levels (please don't do that) or maybe if you have ridiculously hard to drive headphones/IEMs. I don't know if there is a solution to this or if using a powered USB hub will fix the issue.
Hope this helped.
