Cleaning USB for Bus Powered Audio Devices: Discuss

[solderdude]

Chances are it is the Mobo.
What MAY be causing this is 2 parts of the ground plane on the board (audio out or ground from the TV or monitor) may be connected to a ground on the Mobo groundplane that is physically too far away from the ground of the USB.

What happens in such case is that there is a voltage of a few mV difference between those 2 grounds and these may be connected via each other using ground connections (the common) via audio and digital (USB) signals. This can cause a current to flow through these devices that should not be there. Some manufacturers use 1 ground for audio and another for digital, others tie it together or do so using an inductor or caps.
In such case there is a voltage difference where there should not be any which is what you hear. This can also run via safety earth for instance.
The groundloop can well flow via the active monitors and safety ground for instance.
Below the PCB of the D10. One can clearly see separate ground planes (as is usually the case for DACs) and at some point they are coupled.
That coupling point may well cause this issue WHEN a groundloop exists.

Try to listen to the D10 when connected only to the PC and connect the RCA out to a portable headphone amp or something.
That SHOULD sound buzz free.

Can be quite a puzzle to find the culprit(s).

Maybe galvanic separation for USB can help here.
Use Toslink or the RCA SPDIF, that one seems separated by a pulse transformer.

 

[L5730]

I was coming to the conclusion that maybe I got two faulty units, one has missing decimal points on the display, but behaves exactly the same...that was, until now.

I jumped through the hoops with the RPi and Volumio and got it working on a static wireless IP address - all of the previous issues are gone. The static sounding noise isn't much more than the ADC self noise. I still hear movement of Windows around on the Desktop PC - so that is internal noise, nothing to do with the DAC!
When sample rate is changed on the D10, there is no buzz and zero audible change in the noise profile, no clicks nothing.

So I am declaring that the D10 units I have are not faulty, but these strange noise profiles that change with the behaviour of the D10 DAC are somehow caused by the Desktop PC or somewhere in that direction.

Same UK 240v 50Hz power > (RPi logo) 2.5A 5v PSU > RPi with Volumio 2 > USB > D10 > RCA to 3.5mm TRS > ADC
A USB keyboard is connected to the RPi and a 120mm fan on the GPIO and nothing else.

Now to start plugging things in and testing from here...
Thanks for following this, folks.

Incidentally, the other D10 connected to a cheap HP Notebook in the other room doesn't seem to make a peep either. I turned the integrated amp volume up to what would be pain inducing levels, stuck my ear right next to the drivers and heard nothing more than soft hiss from the tweeter getting louder with the raising of the volume knob - which was exactly the same noise I heard when the amp input was switched to CD (which wasn't turned on) or any other unconnected inputs. So, that setup is clean too.

 

[solderdude]

A happy audio 2019 it seems.

 

[L5730]

Haha, well my investigation hasn't finished yet.

1. Connect HDMI from TV to RPi (TV is also connected to PC graphics card) - buzz!
2. Connect Ethernet to RPi, which is connected to Router, which is connected to PC motherboard. - no change, clean.
3. Use DIY USB cable, get Power and GND from RPi whilst Data and GND from PC - buzz!

So, that USB split power cable doesn't work in my case, which may mean I might have further issues with external USB audio devices in the future.
I'm gonna need to work on getting this PC cleaned up somehow, because it is not really acceptable to have the noise that simply isn't present in the hardware audio devices.

 

[solderdude]

A PC may be hard to 'fix'.
Looks like the TV may be inducing quite a bit of leakage currents from its power supply.
I assume the TV has a 2 prong mains connector. When it is 3 prong that 3rd one may be plastic only.
What you could try is to run a wire from a ground connection on the TV (antenna input or other metal exposed connection) to safety ground.
That could be the 3rd prong or more safely the central heating (when having metal piping) or water piping (copper).
You need to remove paint and connect the wire.
This way leakage currents from the TV may find the way to that ground easier than via the PC.

It probably is not the 5V USB power. It usually isn't a problem. Most DAC's (unless really cheap) condition that 5V or lower it to say 3.3V to get rid of nasties anyway.
Only really cheap USB dongle DACs won't have any decent filtering. There it may be beneficial. To audiophiles with golden ears it always seems to bring great improvements though.

 

[L5730]

I appreciate your thoughts.
I've used a bare copper heating pipe connected to central heating before as a grounding point for connecting an anti-static wrist band to. I realised later that the static is likely a non-issue here in dull and damp England, but I thought I'd be extra cautious when handling computer components.

I just disconnected the TV, but it made no impact on the PC, whereas it did fix the RPi.

So, can I try and solve the noise from the PC by DIYing a cable? Simply providing power from elsewhere did nothing with the ground still connected. Maybe if I put in a 47 ohm 1/2w resistor on the ground between the PC and where the external PSU ground T-'s in, maybe that will help?

I've tried the ports on the back of the PC, same story unfortunately. One would think they would clean these things up more, especially as USB is accepted as a connection for so many devices. It wasn't a cheap budget motherboard (ASRock Taichi X370), and certainly one of the more expensive PSUs (Seasonic Prime Titanium 750).

 

[L5730]

Ah, in my DIY USB split power cable I am still keeping part of the shield connected from the Data providing side to the DAC.
This post indicates the shield is severed too.
https://www.audiosciencereview.com/...asurements-of-topping-d10-dac.2470/post-90089
Is that a safe thing to do?

 

[solderdude]

It looks like a groundloop issue.
It is a matter of finding it and creating a solution.

Have you tried the optical input of the D10 ? or the SPDIF ?
This might be the easiest way out as the loop to the PC is removed.

The shield is supposed to be low Ohmic and carry the majority of groundloop currents so the least amount of loop currents will flow through the common/return wire of the USB bus. It is supposed to connect the enclosures of both connected devices.
Depending on how it is actually used/built in the mobo and D10 may determine the usefulness of this intended feature.
I don't think its the 5V line.

 

[L5730]

I'm afraid I don't have another external DAC to run the optical or coaxial digital into.
To be honest, I think the D10 analogue sounds perfectly adequate, I am just trying to solve an issue that didn't and doesn't really cause me any concern unless I measure it and boost in software. I can, honestly just accept that there is some noise and move on, I don't hear it in normal use.

Had a quite lovely evening listening to the D10 through the HiFi in the other room. Some software issues to sort out with Windows dropping out the audio occasionally for no obvious reason - short tiny silences. I'll find a solution to that eventually, move priorities around, some reg tweaks and bunch of background services to ditch.

Well, I added some sticky tape around the outside of the USB plug that goes into the PC, insulating the plug itself and thus the shield, made no difference which makes sense.

The main issue for me is that discovering this noise, if I wanted to get a cheap Picoscope for fun, or an audio interface like RME Babyface Pro, which is bus powered, am I going to have issues? I'd imagine that with a Picoscope, due to it being sensitive by nature, this might cause problems with measuring anything.

 

[solderdude]

picoscope may work best on battery fed laptops or tablets.
Although I have no issues running a Handyscope but it's the only device along with an UMC204HD on the USB bus (+ USB ext 2.5" HD)

When doing measurements you could always remove the TV and other non essential devices.

Perhaps look for a 2nd hand card with optical or coax out ?

 

[L5730]

Something interesting, although not curing the problem and probably not going in a direction of doing such either:
I ran Prime 95, maxed out all cores (8c/16t AMD 1700x) and this alone removed quite a lot of the higher pitched random sounding noise, all of the scratchy crackly stuff. That leads me to believe that noise came from the CPU frequency bouncing around - a reason why it's been said that for an audio workstation to disable the lower clock states?
The main buzz sounds are still present as always though, no change there.

I'll unplug everything except the PC itself tomorrow (today) and speakers/monitor to hear/see whats's going on.

Sorry, I'm not following your thoughts on optical out. I've nowhere for any optical or coaxial s/pdif signal to go, no other DAC or ADC in my presence as yet. If I did, I reckon the D10 is probably fine on it's currently noise producing connection, optical would isolate that noise.
Odd that it changes when the screen brightness changes. It doesn't really give big enough clues.

I might grab some lengths of copper cable and try hooking up grounds. I mean the outer of plugs like HDMI or audio. Bad idea?
I mean, it'd make all of the grounds common, and if I hooked up to a central heating pipe, all the merrier?

 

[solderdude]

I would just start with the TV. The PC (when plugged into a 3 prong socket) should be grounded already.

There are PC cards that will have optical or SPDIF RCA out. Use that instead of USB. This way you are galvanically separated from the PC.
It is probably a loop somewhere that is causing this.
The D10 has optical and coax SPDIF input and the latter seems galvanically separated (don't know for certain).
These 'loops' are hard to see as sometimes there does not appear to be a loop but it is there via mains connected devices.

 

[L5730]

Sorry, they are all outputs on the D10, the only input is USB.
The TV is disconnected from HDMI and unplugged from the mains - makes no difference.

I pulled everything else from the PC, one by one and also all at once, nope, still buzzing (quietly) though D10.
It's the motherboard, PSU or the case, and I have no others to switch out, grr.

I snipped the shield on the DIY USB cable. No difference. The only other thing to do is to snip the GND, and only have Data + and Data - come from the PC, and the power come from the RPi. I don't think it;'s a good idea to do this, so will wait until I grab a resistor to put inline GNR, PC side, and maybe a 10uF cap to put between the +5v and GND.

I listened again, and at normal listening levels with my ear right next to the driver I can hear the tiniest faint buzz. At proper distance it's quieter than the background noise in the room, so I stress again it's not stopping me from using the D10 with my PC.

 

[L5730]

Interesting reading and explanation of ADuM4160 chip and why it does low and full (12Mbps) speed, not high (480mbps) speed required for Hi-er Res. audio.
https://www.analog.com/en/analog-di...isolators-in-medical-and-industrial-apps.html

These seem cheap enough (on ebay etc.) to see if it makes a difference.
Other than this, there is the ifi Defender USB 3.0 coupled with external PSU, but it's almost as much cost as the DAC! Still, could be useful in the future.

 

[solderdude]

Yes, my bad.. and too bad ...thought they were inputs but they are outputs. Forgot it could also be used as an SPDIF converter.

Looks like there is an unwanted current flowing from the mains through the active speakers through the D10 and then through the PC to the mains again.
That (very small) current is causing a very small voltage somewhere between the D10 DAC chip and inputs of the speakers which is what is heard.
Chances are when the loop is broken the problem is gone.
Have you tried hooking up the speakers directly to the mobo analog outs ? Is it quiet then ?

edit: maybe interesting in this situation ?

 

[L5730]

This is basically the setup, apologies for the scruffiness.
The desktop speakers are powered with a 3-pin transformer, but I've disconnected them and it made no difference. These are connected to 3.5mm on motherboard.
The monitor is connected via DisplayPort, but I unhooked that for the sake of testing, and also unplugged from the multi-strip.

The only conclusion that I can draw from all of this is that it's either the PSU (not likely) or the motherboard that has some kind of dodgy design which is causing some kind of USB power or GND or Data noise. It seems disappointing.

I am not going to break the USB GND connection PC>DAC because it's used along with the Data lines if I am correct, and I do not want to upset things. I plan to find some cheap parts (low cap. and pot./resistor selection) and see if that helps.
I don't think ferrite chokes will do anything here.

 

[solderdude]

What's the purpose of the D10 ?
Ferrites don't do anything in the audible band, only far above it, in the MHz range.

I assume your normal setup would be USB to D10 > analog out to speakers.
Now it is analog out PC to speakers and D10 is out of the loop but shorting the analog out to the USB ground line and maybe induce currents in the common line.

When you have connected everything the way as described above and you remove the D10 (either the analog in plug or the USB or both) are the speakers quiet this way ?

 

[L5730]

Normally, I'd connect D10 to active monitors - another set of speakers, not the desktop ones in the image. I hear the buzz faintly in those, when connected to D10 directly. The active monitors each have their own 3-pin plug which connected to the same power strip. D10 connects to RCA and then Neutrik RCA>XLR adaptor, but also tried RCA>TS cable, same thing. A buzz which changes with D10 display brightness.

I am looping the D10 back to the Line in ADC on motherboard so that I can boost the signal and hear the noise easier. I had it like this with the RPi and there was no noise coming into the ADC from the D10.

If I just unplug the D10 USB connection from the PC, the buzz goes away. This is still the same case where I have 5v + GND running from RPi but no Data lines of GND connected to the PC.

If the Desktop speakers are connected directly to the D10 and they are turned right the way up I cannot hear any buzz. But I don't think they go loud enough for me to detect it, in all honesty.

 

[L5730]

I don't know if this is more confusing or not. But this is the more typical setup.
Active Speakers/Monitors only have XLR/TRS combi jacks that do accept unbalanced sources. Each speaker has it's own amp, so needs it's own connection to AC mains.
Neutrik RCA adaptor puts Pin 2 and 3 to ground, which is the shield in the case of the RCA, right?
RCA>TS cable , same story.

Purpose of D10:
Provide a dedicated DAC for active monitors.
Motherboard onboard audio goes to desktop (active) speakers which are on all the time and are not a worry for the sake of fidelity.

 

[solderdude]

Most likely an USB isolator, like the hifime can be the solution.
Can't guarantee it though.

You can see the loop in the drawing. USB to audio in of the speakers to mains and from mains to the PC power supply and from that to the PC case and the mobo which is connected to the USB.
Breaking this loop can be done with audiotransformers in behind the audio out of the D10 or the USB. In either case it isn't done cheaply and may only lower the noise or bring it below audible limits.