Troubleshooting PC audio noise

[mihaelbele]

Hi,
stereo amp I'm using with PC is missbehaving, and instead of fixing it, I decided I'll replace it with small D class amp (10w, one pot, powered from 12V PC PSU rail). Instead of building one, I found "perfect" ~$2 thing on aliexpres (PAM8610 board). Amp works OK, but I'm having a problem with induced noise (static like crackling influenced by CPU load and mouse movement, pulsing in ~4 pulses per second). This noise is present with the original amp also, maybe little less loud, and was one of the reasons why I wanted to power the new amp from PC PSU to avoid (longer) ground loop. PC is brand new and I think it has high quality components.
Here's what I've tried:
1. stereo cable with ferite beads on both ends - no audible difference compared to plain cable
2. ferite bead on 12V amp power line near the plug - no audible difference
3. ground lift on stereo cable (amp shorts DC 0V with audio GND, and it's shorted on motherboard side too) - didn't fix it, maybe the noise is a bit louder
4. analog audio output from monitor DAC (digital via DP) - no audible difference compared to PC analog output (stereo cable is far away from any power cables, but near DP and USB)
5. front vs rear analog output from PC - no audible difference
6. different analog source (smartphone) - acceptable noise level (almost none)
7. headphones instead of amp on PC front panel connector and monitor analog audio output - there's no noise! (or I can't hear it)
8. turning of both monitors (and unplugging HDMI/DP/USB cables), so no USB hub/wireless mouse/HDMI/DP - no audible difference
9. unplugging audio cable from PC - no noise, hums when I touch connector's hot terminals
Using optical output from PC and feeding the amp from external DAC solves the problem when using separate isolated 5V (phone charger) power supply for DAC (noise levels very low), while switching to power supply via PC USB noise level raises (but it's acceptable).
I don't have 12V/3A isolated (DC 0V not connected to protecive earth) PSU for the amp, so I used 12V/1A just to test the noise with PC analog output, and it's acceptable as long as I don't touch the metal pot (whole amp should be isolated and pot should have plastic knob) - just some low level buzz and hum. With non-isolated power supply, the noise is the same as when using PC PSU.
Can someone explain what is happening?

 

[Newmarket]

For clarity - are you amplifying analogue audio from the PC or feeding a digital signal to a DAC/Amp combo ?

 

[mihaelbele]

I don't have an amp with built-in DAC, these are two separate components, but I tried both digital and analog PC outputs (amp is powered from PC PSU 12V power rail):
- analogue out from PC is noisy (3.5mm front and rear jacks)
- digital out via displayport to monitor -> monitor's built in DAC -> 3.5mm headphone out - noisy
- digital out from PC (optical/toslink S/PDIF) -> standalone DAC (powered by floating 5V power supply) - OK
- digital out from PC (optical/toslink S/PDIF) -> standalone DAC (powered by PC USB) - acceptable

 

[Khron]

Ground loop. Especially since single-ended (ie. not balanced) audio connections are employed.

- digital out from PC (optical/toslink S/PDIF) -> standalone DAC (powered by floating 5V power supply) - OK

All other variants have a common ground/earth connection between the DAC and the rest of the components (computer, display. Although powering the amplifier from the computer's own 12v rail is..... not the best solution, to begin with.

 

[mihaelbele]

Thanks, so, "ground loop problem" is not happening only via protective earh (both PC DAC and amp have the same reference to PE), but also via DC 0V lines? If there are multiple DC0V lines, do they act as antena (eg. if their resistance is different, or, how/where is the noise induced into audio signal)? Also, ground loop symptoms are not limited to AC hum and buzz only, but can also cause system to be susceptible to EMI?

 

[Khron]

Computers themselves are an utter EMI hellscape, and since there's barely any analog circuitry in there, noiseless-ness is pretty low on the list of priorities.

https://en.wikipedia.org/wiki/Ground_bounce

This describes a similar phenomenon. Yes, there are multiple and/or shared ground return paths, and a higher-current return WILL modulate the "ground" / reference voltage of less power-intensive circuits, leading to noise signals added onto audio.

At least that's my rationalization / understanding, but I'll be the first to say i could very well be talking out of my rear end on this...

 

[Mixermend]

Transformer isolation will probably cure this problem - it certainly works when interfacing lap top computers often employed at conference venues, to audio systems.

 

[mihaelbele]

Thanks, I'll probably use external DAC with floating PSU, but I'm really interested in what and why is happening...

 

[richiyobs]

In my old i7 desktop PC i had noise problems, a sort of coil wines.... Glitches and High freq noise moving mouse...
Last week the screen turned off when working... Replaced GPU with an old One from another dismissed PC and now noise has gone away.... Before this replaced Psu too but nothing changed... I was thinking the old mobo was causing the problem instead It was related to a faulty GPU

 

[Newmarket]

Thanks, I'll probably use external DAC with floating PSU, but I'm really interested in what and why is happening...

Much can, and has been, written on this stuff and I daresay plenty of YouTube content of varying depth and quality.
But in summary - you have an unbalanced output with a noisy reference (the 'ground' from the computer). This ground is one of the signal conductors and causes the noise problem.
For analogue audio from the computer itself you can do various things:
Best for noise - transformer isolation. Use one of the commercially available "Hum Busters" from eg Behringer or ART. For high performance in terms of distortion / Freq response etc you'd need to go expensive but depending how fussy you are the cheaper ones should give a major improvement wrt noises.
Alternatively you could try in setting an impedance in the ground signal. Say start with 10R resistor and hear what happens.

 

[mihaelbele]

Thank you, I understand it on that superficial level, but I'm banging my head arround following:
- if PC has "dirty/noisy" ground, how come this dirty ground does not generate artefacts in audio signal in the amplifier if decoupled analog source is used (eg. phone/case #6, or DAC with floating power supply) since amp is powered by PC PSU?
- if PC analog audio is allreday "dirty", why I can't hear it with headphones?
First case can be explained with "there's no DC ground loop between the amp and a source" (is there such thing as "DC ground loop"?), and second case that PC DAC audio output is not noisy to begin with (otherwise it would be a bad motherboard).
On the "wire level", when PC powered amp is used with PC analog source, amp has DC GND connected through power cable and through audio cable (both are almost the same length), and PC DAC has it's ground connected once through PC motherboard to PSU (short) and once again through the audio cable via the amp's power cable back to PC PSU (long). Is this "long" wire the culprit? eg. if I break the loop by disconnecting amp's power DC0V wire and let the amp "reference itself" through audio ground, and that fixes the problem, that would tell me that this long wire somehow introduces noise in the system (if so, how? I think this is crucial for me to understand). I will not try that because I think PC DAC GND trace on the motherboard cannot handle amp's 3A max current draw, but then again, if for some reason audio cable provided less resistive path to close the circuit, would that mean I would burn the motherboard?
On the other hand, if I lift the audio ground on amp input (case #3, audio cable is shielded on DAC end), then audio circuit is closed via amp's "dirty" power supply ground and that makes noise even worse (eg. the way it's happening in the guitar tube amp bad grounding scheme)?

Preemptive answer: "You know nothin' Jon Snow!"

 

[Newmarket]

Too many words there for me tbh. Suggest diagrams.
But - wrt headphones - the headphones are passive. The headphone transducer sees the signal differentially. Unlike feeding an unbalanced amplifier. A balanced input amplifier would give you a better result. This is basically what you are doing with a "hum buster" transformer device (1:1 transformer). The headphone transducer is basically an inductive coil. Like a transformer winding.

 

[rh001]

Hi!

Best chance of a near 100% solution when using ADC and DAC that's inside the pc or notebook chassis is to use transformer isolation, on input and output, like "Hum-Busters"...I use "sum-Eliminator". Often these things are 2 channel so you'll need 2. You could buy one and give it a try....I suppose there is some small chance it alone will do the job and it deosn't hurt to try. But you may find no benefit with just one...two (thus 4 channels in total) is most likely needed. No amount of toying around is likely to fix the problem.

I use a Focusrite Scarlett 2i2 and use a USB isolator on the USB cable and Hum-Eliminator on the input to the Scarlett. With this setup the output is hum/buzz free.
If I eliminate either of the remedies mentioned there will be hum when monitoring the Scarlett output. The character of the hum/buzz is a bit different depending on which remedy I've eliminated. "Hum-Eliminator" is far from perfect. It's seems fine and one may not notice unless you look at it with audio diagnostic software: transformers begin saturate at around 12db and more so at low frequencies than high. One day I'll get around to replacing the Hum-Eliminator with large transformers that don' saturate until 20 or 24 db but too busy these days to start a new project.

USB isolator is from HiFi Me.

 

[mihaelbele]

Thanks guys, but I've allready sovled the problem with floating DAC ($6). I just wanted to understand what's happening behind the scenes, but I guess it's complicated and that I'll have to read on it in tech books (most info I have, deals with noise in industrial automation systems and guitar tube amps).

I use a Focusrite Scarlett 2i2 and use a USB isolator on the USB cable and Hum-Eliminator on the input to the Scarlett

That's whole another story, let me share... I recently bought (and returned) Tascam 16 channel USB audio interface with mic preamps. It was rich in features and price was great. Software was bad - but I could've live with that. What I could not live with is the noise, it's like there was DC present on almost all channels and there were resonances at multiples of 4kHz, screenshots:
noise profile:
normalized inputs recording silence:
Tech support said that that's normal?!? so I went back to my old PC with 2 internal PCI Sek'd Sienas (unbalanced line inputs). Now, that system has multiple PE ground loops (PC, mixing console, powered monitors/speakers), and it's quiet. It wasn't quiet in the beginning (same issues I'm having now with the new PC) when I tried to avoid ground loops between the console and the PC, but as soon as I connected the multicore shields on both ends it went quiet. No "hum eliminators".