PC noise through internal sound card

Kev06

Member
2006-10-05 11:42 am
Midlands
Hi,
I've got an Asus Xonar Essence STX sound card which I've been very impressed with, and I like it even more after installing the third party UNi Xonar drivers mentioned elsewhere on this site. But I find that theres still some noise making its way into the RCA audio path, which is noticeable during quiet tracks or silences in the music.

It could be something to do with the grounding, but its certainly not just that as I can hear sounds that directly correspond to actions on the PC. The card already has a metal case around it, and swapping around the PCIE slots hasn't helped. I'm thinking of upgrading my hi-fi amps and speakers, and this is something that would take the shine off it.

Is there anything else I should try to solve it, or would I be right in thinking that using the card's S/PDIF output to an external DAC would be the best bet?

Thanks
Kev
 

Kev06

Member
2006-10-05 11:42 am
Midlands
Unfortunately, you won't solve the problem. I have the same noise using an USB DAC...
Searching for a solution :confused:
Gianni
Oh dear, I'd been hoping it would work - maybe the interference isn't necessarily though the analogue output stage as I'd imagined, then. A switchable S/pdif dac would be a neat solution for integrating some of my other sources so I'll probably try it anyway, but it appears I'll probably still have this problem to wrestle with.

Guess I could try to build a dedicated, electrically quiet media PC; aside from this small issue I'm so pleased with the move to the computer that I can't see myself wanting to go back to disk spinners.

Cheers
Kev
 

Kev06

Member
2006-10-05 11:42 am
Midlands
What frequency is the noise?
There are two types of noise really:

When the PC is at idle, there's a sort of general background 'data' type noise - I remember years ago some early computers loaded games through a cassette player and it made a noise like a very fast fax machine, this is similar only considerably faster - its much higher frequency than a 50hz buzz (although theres a faint one of those as well if I really crank up the amp). If I run processes on the PC (like play a video) this noise changes - it can pause for moments, get louder, and usually gets higher in frequency; almost becoming a whooshing hissing noise.

The other noises are varied, I get crackling/hissing noises and pops and clicks but only when the PC is working at something; opening/closing windows for example. When its idle these never occur.

Apologies for my rather rubbish descriptive abilities!

Cheers
Kev
 
Cheers, trying to decipher all the noise sources from a PC is interesting, all audible noise will be 20-20KHz, which means the digital noise is getting rectified and adding to the noise floor. The other problem is economics, one of the main culprits being the main SMPS (SMPS PSUs are not noisy if done right) but a PC one will be often built to a cost, single or double sided PCB with PTH components, compared to a top rate industrial one (where power failure is more costly than a PC going down), multi layer in built planar transformer, capacitive screening between layers, etc. The motherboards don't help, they get away with as few layers as possible so a lot of high speed signals run on the outer layers, the pretty squiggly patterns!
So a combination of noise sources, exasperated usually by a power supply well below the standard required, a comparison, again some more critical stuff, the power supply was 40% of the cost.
Oh often over specifying a SMPS can cause problems, depending on the topology, during low current requirements they can enter burst mode which lowers the frequency of the noise, even into the audio band in some cases.
 

Kev06

Member
2006-10-05 11:42 am
Midlands
Interesting stuff; thanks! Not sure how much control I have over those things, though when I built the PC it wasn't intend for music so i didn't try very hard I guess. I did get a Corsair SMPS rather than some unbranded effort, but I've no idea what kind of noise performance it has.

Cheers
kev
 

wwenze

Member
2008-03-07 12:46 pm
It is definitely something to do with the grounding.

I have an STX and a DX. Both cards measure nicely as they are known for in loopback mode. But if I let them measure each other, I get an SNR of only ~100dB and affected by computer activity. Same pattern (looking at the spectrum) can be seen in USB DACs and SPDIF DACs without an isolated SPDIF connection. If I use a sound card with transformer for SPDIF output to connect to an SPDIF DAC, then I get the ~120dB SNR (or somewhere there) that the DAC is capable of.

(The onboard sound also measures around that value with similar noise floor spectrum. Makes me wonder, if this noise is suppressed, onboard sound might actually be at the same level of sound cards.)

Optical out into a DAC would be your best bet at minimizing noise. Otherwise depending on the grounding configuration of the DAC/amp (or speaker) that noise may or may not show up, and you got unlucky this time. I once recommended somebody using studio monitors and USB DAC to "play cheat" and lift the ground on the speaker's power socket (along with the necessary warning that this is entirely not safe and should not be used for normal operation) and that did the trick.

P.S. I'm using Antec Truepower New
 
Last edited:
It is definitely something to do with the grounding.

I have an STX and a DX. Both cards measure nicely as they are known for in loopback mode. But if I let them measure each other, I get an SNR of only ~100dB and affected by computer activity. Same pattern (looking at the spectrum) can be seen in USB DACs and SPDIF DACs without an isolated SPDIF connection. If I use a sound card with transformer for SPDIF output to connect to an SPDIF DAC, then I get the ~120dB SNR (or somewhere there) that the DAC is capable of.

(The onboard sound also measures around that value with similar noise floor spectrum. Makes me wonder, if this noise is suppressed, onboard sound might actually be at the same level of sound cards.)

Optical out into a DAC would be your best bet at minimizing noise. Otherwise depending on the grounding configuration of the DAC/amp (or speaker) that noise may or may not show up, and you got unlucky this time. I once recommended somebody using studio monitors and USB DAC to "play cheat" and lift the ground on the speaker's power socket (along with the necessary warning that this is entirely not safe and should not be used for normal operation) and that did the trick.

P.S. I'm using Antec Truepower New
Very interesting - many thanks indeed!!!

I'd assumed that (as I was clearly hearing computer activity) it wouldn't be simply a grounding issue. But from your measurements it looks like I'm probably very wrong in that assumption! It also confirms that just using the S/PDIF (or USB) may not necessarily solve the problem as I'd imagined.

Presumably the optical out would completely overcome ground loop problems? I haven't chosen or built a DAC yet so, with a bit of planning that would be a very easy solution. I seem to recall that the STX has a similar spec for both optical and coax so it should be comparable quality hopefully.

I also understand the method of isolating/floating the signal using a transformer, which would get over the need for a DAC. But I've not actually tried this in Hi-fi audio and never on a digital output. Can anyone suggest what sort of transformers should I be looking at?

Many thanks
Kev
 
impedance matching is nothing without the cable and connectors all matching the same impedance, as well as termination at the PCB on both ends. thats the last thing to worry about.

what you describe is called blitter-noise and is caused by the graphics processors (GPU) generally, when writing 2D graphics to the screen, which is at lower frequency and this is rectified by chips internal diode junctions, as well as ripple imposed on ground by same.

its conducted primarily through the air, inductively and capacitively, so that it can also be induced on cables. its primarily common mode, but not all. use optical connection, transformer coupled spdif,or isolated USB to avoid it.
 
Last edited:
I have a couple of PCs going atm. Main PC uses an Asus rampage gene 4, micro ATX with shielded and separate ground plane'd audio card. Really notice the difference between that and a cheap Asrock MB I had before. Both of these were on an OCZ 1KW PSU. The PSU makes a difference too, as I believe what you're hearing is also loading / ripple.

I also have a Biostar T-power X58 coupled with a Corsair 850W PSU on the other PC, and that's as ripple / noise / hiss free as I can hear; I'm listening to that rig with a homebuilt LM3886 "gainclone" driving JBL infinity 6x9s. (yeah I know. I'm upgrading bits as I go. Built it out of what I had laying around)

In short, you could try another PSU before you consign the whole thing to the bin :)
 
Excellent - thanks very much for helping me understand this! I've learnt a new term too - 'blitter-noise' isn't one I've run across before.

There does seem to be a definate correlation between whats happening on-screen and the pitch of the noise so it matches what you say about graphics card activity really well. Its encouraging to hear that its likely to be an ambient/environmental type effect, it sounds less fundamental and potentially more solvable than direct electrical problems.

Okay then I'll either try the transformers (observing also the various line impedances) and/or the optical connection and see what happens. The optical one is sounding simpler at the moment, and I quite like it for other reasons too - without ground loop and electromagnetic issues it may be worth looking at repositioning my pre-amp (and/or new DAC) to make the digital optical path longer and so keep any other electrical/analogue cabling shorter. My TV also has an optical output, so I could make/but a DAC that can act as a sort of optical hub or pre-amp.

Thanks again,
Kev
 
I have a couple of PCs going atm. Main PC uses an Asus rampage gene 4, micro ATX with shielded and separate ground plane'd audio card. Really notice the difference between that and a cheap Asrock MB I had before. Both of these were on an OCZ 1KW PSU. The PSU makes a difference too, as I believe what you're hearing is also loading / ripple.

I also have a Biostar T-power X58 coupled with a Corsair 850W PSU on the other PC, and that's as ripple / noise / hiss free as I can hear; I'm listening to that rig with a homebuilt LM3886 "gainclone" driving JBL infinity 6x9s. (yeah I know. I'm upgrading bits as I go. Built it out of what I had laying around)

In short, you could try another PSU before you consign the whole thing to the bin :)
Thanks - I'll keep that in mind in case the optical/isolated approach doesn't work. IIRC my corsair SMPS is the 850w version so probably from what you say about your other PC perhaps I have something not too bad to begin with, though. More by luck than judgement in my case!

Cheers
Kev