Filtering PC "processor" noise through speakers

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Looking for tips in tracing down and eliminating "Processor" noise coming through the speakers. A bit hard to explain, but it's the high pitched "modem-like" noises when GFX are in motion on a screen or the processor is busy.

I've heard old remedies like turning off inputs and BOIS tweaks, but I'm not hearing much on what actually causes it on any given motherboard. Is it often the audio chip? Bleed on the 12V from the video line?

Putting a scope on it this afternoon, but wanted to gather advice.


Thanks.
 
No, USB isn't isolated by nature. If you're not going beyond 96kHz sample rate, the isolation can be incorporated fairly cheaply as there are single chip solutions (e.g. Analog Devices). For higher rates there is one single chip solution available (Silanna) but its fairly pricey.
 
I agree...ground noise is the usual culprit. As an experiment, simplify the system as much as possible...just the PC and the amp...then, as an experiment, use a ground lifter to disconnect GWG (green wire ground) from the amp. Doing so will usually quiet things down a lot. Important: As to lifted GWG/safety, some folks would complain.

However, if it fixes your problem, you could build a ground lifter that maintains the safety connection while improving the audio noise situation.
 
About the best you can do is turn the volume up on the computer and reduce it as necessary on the powered speakers or headphones. That maximizes the signal to noise ratio. I assume it's caused by a failure to provide the DAC chip with clean power.

The best is to use an internal or external sound card or device. Decent PCI cards are dirt cheap at computer recyclers. At least, if your computer still has a PCI slot... Or if it has s/pdif output, use an outboard DAC or AV receiver.
 
Last edited:
Member
Joined 2017
Paid Member
There are many posts about "modem-like" noise " on youtube. It usually comes from ground loop and USB. An isolating transformer for ground loop and USB isolator for USB are usually effective. Many countries use three terminals power; two is for power, one is for earth ground. Earth ground is for safety. That's why a ground lifter is not recommended. But some countries use two terminals power e.g.japan. I don't know many people get an electric shock in Japan. I have doubts about that, IMHO.:)
 
I think the problem is digital engineers who don't know that codecs require clean regulated power to achieve the performance numbers on the data sheet. If you're lucky, they'll copy the reference design, and then the numbers they've copied off the data sheet will actually be real.
 
Integrated soundcards can provide surprisingly decent analog output, provided the ground loop is avoided.

Unlike simple audio components (CD player, tuner, etc) x86 PC components run at very low voltages which generates extremely large currents. A 20W CPU powered by 1.5V requires > 10 amps. The same for even low-powered GPU. North bridge, south bridge - all powered by small voltage -> high currents. These huge currents flow only between the components and the on-board step-down regulators but if the 0V ground trace for the soundcard codec is fetched on the board from a point close to these currents, the voltage difference between 0V on the soundcard output and 0V in the PSU (directly connected to the PC earth wire) is nowhere close to zero. It reflects all the currents related to CPU load, GPU load, USB activity, hard drives activity etc. - the typical modem-like noise.

In my small all-in-one project Home * pavhofman/plabs-player Wiki * GitHub I hooked a small tripath amp (TK2050) to soundcard output, powered it from the PC input jack for the 20V ungrounded power adapter.... and could clearly hear every PC activity (reading from the CF card, letters appearing in linux console, etc.). Totally inacceptable. I had to deploy another ungrounded adapter just for the amplifier... and the noise was completely gone. Measurements (ESI Juli@ + jaaa at another thin client) at speaker terminals when paused are enclosed, they are under -110dB for integrated soundcard -> cheap digital amp powered by a cheap adapter.

But this requires an additional adapter and relay to power-down the adapter when the PC is sleeping/off. I do not like it.

For second version of the device (the dirt-cheap all-in-ones are this year's project of a few school kids who like to play with computers and soldering iron) I was looking for an amp with balanced inputs. Tripaths do not offer the negative input pin, but the newer TP3116 chips do. The $7 amp DC 12V 24V TPA3116 Dual Channel Stereo 50W*2 BTL Mono 100W Audio Amplifier Board | eBay looked promising - all four inputs coupled with 1uF capacitors - one to the input terminal, another to ground.

First we hooked the amps to the soundcard jacks unbalanced and powered all by one PC adapter - the expected modem noise, no way.

Then the grounded side of input coupling capacitors on the amp was unsoldered and output jacks were connected in sort of balanced way - live soundcard jack terminal -> amp IN+, 0V at the soundcard jack -> amp IN-, shield of the balanced cable connected only on the soundcard side. And the noise is gone, nice sounding $7 balanced input amp, both the amp and the thin client PC powered by a single adapter. I do not have the measurements yet as guys are still working on their devices.

Of course this works only for analog outputs of internal soundcards, it would not help with a USB DAC. But perhaps it will solve the ground-loop problem for someone as it did for us.
 

Attachments

  • paused.png
    paused.png
    59.2 KB · Views: 245
  • TP3116.png
    TP3116.png
    71.5 KB · Views: 254
  • TP3116-img.jpg
    TP3116-img.jpg
    72.6 KB · Views: 265
Another answer that may work is to eliminate those high current found on PCs but simply using a Raspberry Pi as the server or slave.

The ARM chip in my 3B barely gets warm and it's powered off a small USB charger so I know there's not much current there. Also it's fanless and therefore totally silent (and cost less than my quiet PC's PSU).
 
I second the recommendation of a Toslink (optical) connection to an outboard DAC. I have never had a processor noise issue then. Sometimes noise from a dubious USB power source, though. Currently I am using TOSLINK to an eight channel balanced USB DAC but that may be overkill for your purposes :D
 
Could this be worth trying - John Swenson's theory about leakage currents & how they cause audio issues? One easy solution to try is - to ground the negative side of any SMPS powering your PC speakers - remember there's also an SMPS in the PC too & the negative output of this may also have to be grounded?

Here:
SMPS and grounding - Page 9 - UpTone Audio (Sponsored) - Computer Audiophile
"In the audio relm the effects of leakage that were important concerned generating voltages across loads and sources, even with tube circuits these are usually significantly less than 1 Mega Ohm, thus in what I am calling the "low impedance" range."

"
SMPS have been extensively tested for leakage, but it has been done with all the existing test equipment and methodologies, thus focusing on low impedance leakage.

Unfortunately it turns out that SMPS also include a high impedance component to their leakage, this is way above 1 Mega Ohms. The problem is that the existing test equipment and methodologies shunt this high impedance leakage to ground so they never see it. So nobody knew it was there. This high impedance leakage is significantly higher in intensity than the traditional low impedance leakage, so it can actually have a significantly larger affect on audio systems than traditional leakage, but nobody knew it was there.

Do not confuse the high impedance with high frequency. The SMPS contains high and low impedance components at all frequencies. Thus even at 60 Hz, there are both high and low components. This MUST mean that there are at least two different mechanisms contributing to the leakage simultaneously. One with a high impedance and one with a low impedance. The same thing happens at the higher frequencies. That amplitude ratio between high and low impedance varies with frequency (this is varies radically from one model to another), but both components seem to exist across the frequency range"
And here:
A novel way to massively improve the SQ of the microRendu / ultraRendu - Page 11 - Sonore (Sponsored) - Computer Audiophile
"SMPS leakage current contains two parts: a very high impedance part that goes through just about anything, including Ethernet transformers and a low impedance part that is easier to block.

The high impedance part is over 300 Mega Ohm which takes something in the multi Giga Ohm region to block. BUT it is very easy to shunt, this is what the grounding of the SMPS negative DC output does, it shunts the high impedance part, leaving the low impedance part. It seems that the low impedance part can easily be blocked by the Ethernet transformer. This happened in every switch I tested, the leakage coming into the switch from the power supply that powered the switch was fully blocked by grounding."​
 
I intially tried 2 different USB isolators for the same problem. While they worked for the ground loop and got rid of the noise, my dac would have periodic clicks and drops (that my older dacs didn't).

The ultimate solution was changing my laptop's 3 pin ac-dc brick power supply to a doubly insulated 2 pin ac to dc supply (aftermarket, reputable brand). I was able to get rid of the USB isolator, zero noise.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.