Investigation Into Effects Of PC load On DAC Analogue Output

[March Audio]

3 minutes ago, jabbr said:

 

 

If you think that common mode noise is just nebulous hand waving you need to go back to school -- of course this is an example, a plausible example nonetheless. But man its just freshman electronics...

Eh? I never said that.

[jabbr]

1 minute ago, March Audio said:

Eh? I never said that.

Sorry "again its just nebulous hand waving" was why I concluded that

[March Audio]

5 minutes ago, jabbr said:

idk balanced does differential mode noise rejection.

Seriously, you think that balance connection dont perform common mode rejection and perform differential mode rejection?

 

Sorry but you haver just given the game away.  You might want to re-visit your "freshman" lectures.

[jabbr]

7 minutes ago, March Audio said:

Seriously, you think that balance connection dont perform common mode rejection?

 

For the purposes of our discussion here USB is balanced. Copper ethernet is balanced. Neither entirely block common mode noise. The topic is too complicated to debate this way. Ground loops are one type of common mode noise that balanced connections block. I'm referring to the broader wide band common mode noise that isn't blocked simply by balanced connections i.e. the kind that would be relevent to this discussion.

 

Peace.

[March Audio]

3 minutes ago, jabbr said:

 

For the purposes of our discussion here USB is balanced. Copper ethernet is balanced. Neither entirely block common mode noise. The topic is too complicated to debate this way. Ground loops are one type of common mode noise that balanced connections block. I'm referring to the broader wide band common mode noise that isn't blocked simply by balanced connections i.e. the kind that would be relevent to this discussion.

 

Peace.

That is irrelevant.  We are only interested in what comes out of the DAC.

 

Please go back and read the thread.  I have mentioned the effects ground loops, and noise currents multiple times, even in the OP.

[March Audio]

In post 4? I even showed the effects that can happen.

[jabbr]

17 hours ago, March Audio said:

That is irrelevant.  We are only interested in what comes out of the DAC.

 

Please go back and read the thread.  I have mentioned the effects ground loops, and noise currents multiple times, even in the OP.

I really don't see your point. Sure what matters is what comes out of the DAC, agreed. At this point if what comes out of the DAC is common mode noise, then we can look at the output of the amplifier to see if its an issue. Yet that entirely misses the point I made. My point was about common mode noise on the DAC input i.e. carried between the PC and DAC across the USB cable.

 

Also the DAC you've used is SOA so perhaps they have a great USB input that isn't sensitive to common  noise, or ringing or jitter on the USB input ... terrific. 

 

**That said** a conceivable physical mechanism whereby the USB output of a server **might** affect the output of a DAC includes variations in the USB signal that fall within the USB spec, those include rise time, allowable jitter and common  mode noise. Those are possibilities. Jitter has been debated ad nauseaum and I'm a bit skeptical to what extent it matters because great clocks are cheap these days. I'd be happy to see a measurement using two different USB clocks that show a difference at the output of the DAC. No one has shown this. I can say that the Ethernet standards have measurably low jitter for 10Gbe and up so with 10Gbe and up Ethernet its essentially not an issue. The subjective debate remains. Rise time variations: I have seen these cause significant changes in ringing at the USB receiver ... so lets see if cables for example either affect rise time or the output of the DAC. Lastly is common mode noise (not merely ground loops): this can sail down a USB cable. It can even sail across the tiny little isolation transformers on a copper Ethernet PHY.

 

These are all physically possible mechanisms that might affect the DAC output and remain untested. When people claim  that there are no plausible mechanisms which fall within the law of Physics they are wrong. When people claim that because USB can reliably transfer data then there cannot be a SQ difference regardless of the exact USB connection including server cable and receiver, they are wrong. Physics doesn't say that and never has.

[PeterSt]

@jabbr, as long as it is allowed ... I would grant you so much more show time ... well, you know ...

Regardless the camp you're in, I'd really wish all posts go as balanced as yours. 

[idiot_savant]

Erm,

 

sorry to be pedantic, but this thread is about PC load? Can that change common mode noise? If so, surely we can measure it? As for ringing, it’s not showing up on the DAC output? The thread is about common conceptions  - if the DAC used in the test is “too good”, can someone suggest one more likely to show differences?

 

 

 

 

 

your friendly neighbourhood idiot 

 

[March Audio]

1 hour ago, idiot_savant said:

Erm,

 

 As for ringing, it’s not showing up on the DAC output? The thread is about common conceptions  - if the DAC used in the test is “too good”, can someone suggest one more likely to show differences?

 

 

 

 

 

your friendly neighbourhood idiot 

 

And doesn't this just reinforce the comments earlier in the thread, ie that it's obviously quite possible to engineer a dac that isn't sensitive to these problems.  It isn't a virtue for a dac to be sensitive to these problems.

 

It appears to me that people look at the problem backwards.  Pcs are noisy. Period.  They are also very variable in this respect. If you really are hearing problems then dump the offending dac. 😉

 

Anyway, to be clear I haven't finished looking at dacs, I can test as many as I can get my hands on.  This is an investigation, not a statement of "I have proved X".

[One and a half]

@March Audio, In revisiting this thread and the graphs on page 1, the PC is shown at idle. 

Most DACs I know of remain muted and if there's no incoming signal to process and display the last current state and don't perform any work.

 

The graphs would be more meaningful if the PC was actually playing a file (doesn't really matter what type) and the DAC is processing and performing what it does. The idle PC shows, predictably no difference in the DAC output cause there's no work to do.

 

Or am I missing something?

 

[March Audio]

1 minute ago, One and a half said:

@March Audio, In revisiting this thread and the graphs on page 1, the PC is shown at idle. 

Most DACs I know of remain muted and if there's no incoming signal to process and display the last current state and don't perform any work.

 

The graphs would be more meaningful if the PC was actually playing a file (doesn't really matter what type) and the DAC is processing and performing what it does. The idle PC shows, predictably no difference in the DAC output cause there's no work to do.

 

Or am I missing something?

 

Sorry but you misinterpreted the plots.  The dac is playing a -60dB 1kHz tone, so it's working away 😀

[March Audio]

The tone was being played by Roon software.

[opus101]

3 hours ago, idiot_savant said:

Erm,

 

sorry to be pedantic, but this thread is about PC load? Can that change common mode noise?

 

That is an interesting question and worthy of a thread by itself I'd hazard.

[March Audio]

7 hours ago, jabbr said:

I really don't see your point. Sure what matters is what comes out of the DAC, agreed. At this point if what comes out of the DAC is common mode noise, then we can look at the output of the amplifier to see if its an issue. Yet that entirely misses the point I made. My point was about common mode noise on the DAC input i.e. carried between the PC and DAC across the USB cable.

 

Also the DAC you've used is SOA so perhaps they have a great USB input that isn't sensitive to common  noise, or ringing or jitter on the USB input ... terrific. 

 

**That said** a conceivable physical mechanism whereby the USB output of a server **might** affect the output of a DAC includes variations in the USB signal that fall within the USB spec, those include rise time, allowable jitter and common  mode noise. Those are possibilities. Jitter has been debated ad nauseaum and I'm a bit skeptical to what extent it matters because great clocks are cheap these days. I'd be happy to see a measurement using two different USB clocks that show a difference at the output of the DAC. No one has shown this. I can say that the Ethernet standards have measurably low jitter for 10Gbe and up so with 10Gbe and up Ethernet its essentially not an issue. The subjective debate remains. Rise time variations: I have seen these cause significant changes in ringing at the USB receiver ... so lets see if cables for example either affect rise time or the output of the DAC. Lastly is common mode noise (not merely ground loops): this can sail down a USB cable. It can even sail across the tiny little isolation transformers on a copper Ethernet PHY.

 

These are all physically possible mechanisms that might affect the DAC output and remain untested. When people claim  that there are no plausible mechanisms which fall within the law of Physics they are wrong. When people claim that because USB can reliably transfer data then there cannot be a SQ difference regardless of the exact USB connection including server cable and receiver, they are wrong. Physics doesn't say that and never has.

Then you haven't understood why I talked about measuring single ended.  Can I politely suggest you review your knowledge of common mode noise, how balanced will have some  rejection of this and SE won't.  

 

You are also not listening or reading what's already in the thread . I have already pointed out to you that I talked about noise currents travelling through usb connection and even shown examples.

 

I have also said I will be looking at several different dacs and pcs  not just one SOA unit.

 

You don't need to look at the clock to see jitter.  If there is jitter is visible in the dac output.  Please refer to the previous posts on this.

 

Most dacs use a very stable internal clock to drive the data timing into the dac chip so that data connection is irrelevant to this, unless it can somehow corrupt the behaviour of the dac word clock.  If it dies it *will* be visible in the dac output.  It's not invisible magic.

 

Again this is all hand waving of might this, might that.  We are looking for evidence.

[One and a half]

Is there a way to measure common mode noise only on the input to the DAC and the output of the DAC?

[One and a half]

4 hours ago, March Audio said:

Sorry but you misinterpreted the plots.  The dac is playing a -60dB 1kHz tone, so it's working away 😀

Tones. No music?

[March Audio]

10 minutes ago, One and a half said:

Tones. No music?

We are looking for any effects of PC load.  Music would make it more difficult to see.  Music  does not bring any extra challenges over tones, thats a myth.  However a multi tone test was also performed and no different results were seen

[March Audio]

22 minutes ago, One and a half said:

Is there a way to measure common mode noise only on the input to the DAC and the output of the DAC?

Why?  We are only interested in the output of the dac.

[March Audio]

16 minutes ago, One and a half said:

Tones. No music?

Just to add to this, I intend to organise some controlled listening tests at some point

[One and a half]

16 hours ago, March Audio said:

Why?  We are only interested in the output of the dac.

The images of the waveforms don't tell the whole story. A PC is a common mode generator par excellence. There are so many smps and pwm controllers on the motherboard, that anything connected will receive a fair dose of noise, this includes USB, AES3 cables.

The images don't show how the soundstage is set, how wide or narrow the instruments are placed in music. From my humble experiments with the DX ISO filters on ethernet transmissions, there's more up the upper end of the conducted EMC spectrum that causes soundstage errors in DACs. Yes these are heard and are subjective. By removal of upper audio band noise does the audio soundstage improve.

 

Here's a paper on common mode noise on AES3 transmissions for interest.

 

 

AES2003.pdf

[March Audio]

3rd time lucky, please see my comments on single ended measurement.

 

Soundstage is an artificial construct. Firstly a function on the recording engineers choices.  A multi miked performance is panned and level mixed by the engineer/artist/producer.  

 

Secondly, how this is replayed in your home is very dependant upon the speaker dispersion and room acoustics.

 

Thirdly, with respect, your anecdotal subjective experience isn't adequate in an objective thread.

 

Can you explain how you quantified the upper emc spectrum?  What RF spectrum analyser you used?

 

How did you quantify the upper audio band noise?

 

How did you quantify the soundstage change in a controlled way?

[March Audio]

Must to add, for soundstage to change the frequency response, phase (timing), or amplitude would need to change.  The tests look at all of those things.

[manueljenkin]

2 hours ago, March Audio said:

Must to add, for soundstage to change the frequency response, phase (timing), or amplitude would need to change.  The tests look at all of those things.

If you think only these three factors measured through a sine sweep analysis contribute to spatial perception and accuracy of a real world sound signal, you clearly haven't understood anything about "the last 100 years of psychoacoustics research" you so often quote! I repeat for the n'th time here, human hearing works on a very complicated adaptive time-frequency manner, and this has been established fairly conclusive (and hence making fft analysis inconclusive) for more than 10 years now.

[March Audio]

2 hours ago, manueljenkin said:

If you think only these three factors measured through a sine sweep analysis contribute to spatial perception and accuracy of a real world sound signal, you clearly haven't understood anything about "the last 100 years of psychoacoustics research" you so often quote! I repeat for the n'th time here, human hearing works on a very complicated adaptive time-frequency manner, and this has been established fairly conclusive (and hence making fft analysis inconclusive) for more than 10 years now.

Of course they are measured with this.

 

Your hearing works spatially by assessing timing, amplitude and frequency response.  Responses that are modified by your head, torso and ear shape.

 

How do you think that things like Q sound Work?  They manipulate these factors.

 

OK Manuel, would you like to tell us all about your knowledge of the subject?  Tell us how it works?