Computer Performance Can Make A Difference in DAC Performance

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
My son bought a $400 DAC that claimed to have async USB support but immediately found out that he got PC noise through it just like his other solutions. So I instrumented it and this is what I found.

Feeding it a single tone over S/PDIF:



I wouldn't expect this kind of jitter in a $25 DAC let alone one for $400.

So we decided to test it with the PC driving it over USB. We played the tone using Windows Media Player:



Didn't think there would ever be a DAC that had more problem with S/PDIF than USB but here we are and the huge jitter spikes are gone. But they are replaced with low frequency ones around our main tone.

To test my son's experience of him hearing system activity, we played the identical track using another player, the Media Player Classic:


All the jitter is gone!!!

We went back to WMP player and let it read the file for a bit and it too settled down to the same performance as Media Player Classic. Clearly system activity disturbed this DAC.

Its brochure brags about "isolation" but the word "galvanic" is not there. This means that it is not isolated from the PC ground and that is feeding into the oscillator causing those correlated jitter spikes.

He sent the DAC back to get his money. When the check arrives, I will mention which DAC it is :).
 

jkeny

Industry Expert, Member Sponsor
Feb 9, 2012
3,374
42
383
Ireland
Interesting post, Amir
Can I ask you about this?
In previous discussions about software playback (Jplay), the contention was made that a software player's influence on the audio output would be swamped by the OS activity. In other words no bit-perfect software player could really have any influence on the output sound.

Is this post showing a measurable difference between playback software (irrespective of the DAC being used)?

"We went back to WMP player and let it read the file for a bit and it too settled down to the same performance as Media Player Classic. Clearly system activity disturbed this DAC."
Have you done these same measurements with other DACs (ones that don't resample) using WMP & MP classic & any idea why there would be a difference in the output between these two players & what is the cause of that difference?
 
Last edited:

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
Sure you can ask :). If you recall our conversations, you always see me preface it with having a high quality interface between the PC and your DAC. Once there, then I question the value of different player and indeed I ran the same test before and could not make the variations happen on other hardware that had proper isolation. Indeed, I suggested that an easy win for Jplay would be to use one of these suspect DACs to make their point.

Shame I did not think of running their player before my son decided to return his DAC.

If I get time, I will repeat the same test with Jplay with my current hardware and see if there is a difference.
 

jkeny

Industry Expert, Member Sponsor
Feb 9, 2012
3,374
42
383
Ireland
Sure you can ask :). If you recall our conversations, you always see me preface it with having a high quality interface between the PC and your DAC.
MY takeaway from our Jplay conversation was that the inherent noise of a computer's OS/hardware would swamp any possible improvement that software could make. I remember the logic being along the lines that the OS activity itself was far more critical than & swamped any small contribution that application software had. So I was surprised to see your measured differences between WMP & MP classic which seems to point to a very significant difference between the USB output when different playback software is used. I'm also not sure how "let WMP read the file for a bit & settle down" resulted in a better performance? Any ideas?

A "high quality interface" can mean all sorts of different things to different people. For instance, as most asynch USB DACs now operate at USB2 high-speed, I think you will find it difficult to find one which galvanically isolates at the 480Mbps USB input speed (although it's possible to isolate the signal & grounds further downstream but how many DACs do this?). I'm not sure what criteria you would use to qualify a USB DAC as "high quality"?.
Once there, then I question the value of different player and indeed I ran the same test before and could not make the variations happen on other hardware that had proper isolation. Indeed, I suggested that an easy win for Jplay would be to use one of these suspect DACs to make their point.
I'm not sure what you mean by "proper isolation" but in my experience, software & hardware improvements on the PC front end usually translates to better sound through any DAC I've heard & not just "suspect DACs". One common characteristic of all these DACs was that none used ASRCs - in my experience ASRCs mask the audible benefits that can be realised from improvements to the front end system.

Shame I did not think of running their player before my son decided to return his DAC.

If I get time, I will repeat the same test with Jplay with my current hardware and see if there is a difference.
Yes, pity you didn't get to run JPlay on that DAC but maybe differences will show up in measurements using another DAC?
 

Elberoth

Member Sponsor
Dec 15, 2012
2,007
253
1,170
Poland
The problem is, there is no such a thing as 100% isolation - unless you go optical interface route (but then, the jitter problems multiply). So there is always room for improvement on computer side.

I have recently switched from Win 8 to Windows Serwer 2012, running the Audiophile Optimizer software, and found the improvement to be noticable. Both on the AQ DragonFly (no isolation) and my Trinity DAC (galvanic isolation on I2S lines inside the DAC). The BADA ALlpha USB I used to use also wasn't 100% immune - even though, its designers have made some heroic attempts to eliminate not only primary couplings, but also secondary and tertiary couplings.
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
The problem is, there is no such a thing as 100% isolation - unless you go optical interface route (but then, the jitter problems multiply). So there is always room for improvement on computer side.

Although I also often say that there is no such thing as 100% isolation - you can not have a system that transmits the logical information and timing without transmitting some noise - I can not see why a a properly designed optical interface system, having the master clock at the DAC side, multiplies the jitter problem.

When using USB DACs we can not exclude that the PC noise is affecting the performance of the whole system - after all people use fine tuned power cables, ICs, power conditioners and even grounding devices in their systems, and then introduce a lot of wide bandwidth noise through the signal and ground wires.
 

jkeny

Industry Expert, Member Sponsor
Feb 9, 2012
3,374
42
383
Ireland
Agreed - there is no such thing as 100% isolation, although it's not clear why.
The isolation of I2S + ground lines inside the DAC followed by a buffer & reclocking with local clock should result in complete immunity to any upstream conditions but still appears not to do so.
It may well be that the signal waveform itself is not as immune to noise as is usually attributed to a digital signal. The conversion from digital to analogue seems to be more sensitive to influences than first suspected & the conduits for all these influences hasn't been fully identified yet.

The electrical to optical conversions at transmitter & receiver seems to introduce it's own jitter in consumer implementations, anyway.
 
Last edited:

Phelonious Ponk

New Member
Jun 30, 2010
8,677
23
0
Agreed - there is no such thing as 100% isolation, although it's not clear why.

What makes you think the isolation is incomplete, then? Something you hear? Could it be something else you're hearing other than upstream noise?

The isolation of I2S + ground lines inside the DAC followed by a buffer & reclocking with local clock should result in complete immunity to any upstream conditions but still appears not to do so.

Again. "Appears." How so?

The electrical to optical conversions at transmitter & receiver seems to introduce it's own jitter in consumer implementations, anyway.

Maybe. Maybe it's even audible. How is this related to isolation and electrical noise?

Now we're back where we left off, John. :)

Tim
 

Elberoth

Member Sponsor
Dec 15, 2012
2,007
253
1,170
Poland
Although I also often say that there is no such thing as 100% isolation - you can not have a system that transmits the logical information and timing without transmitting some noise - I can not see why a a properly designed optical interface system, having the master clock at the DAC side, multiplies the jitter problem.

We are probably discussing two different things. I was talking about sending a data via TosLink interface.
 

jkeny

Industry Expert, Member Sponsor
Feb 9, 2012
3,374
42
383
Ireland
What makes you think the isolation is incomplete, then? Something you hear? Could it be something else you're hearing other than upstream noise?
Of course it could but noise is the best current guess (ignoring the usual - "it's all in your imagination")

Again. "Appears." How so?
I don't really need to explain, do I?

Maybe. Maybe it's even audible. How is this related to isolation and electrical noise?
Attempts at optical isolation using TOSLINK can cause it's own audible issues so why would we want to substitute one set of problems for another? Maybe a better optical transmission technology exists (some people say that ST OPtical sounds much better but I haven't heard this yet)

Now we're back where we left off, John. :)

Tim
Cool, Tim, thanks.

Edit: There are 2 optical solutions that I know of which will work with USB2 hi-speed - Adnaco & a new one just released from Corning which handles USB3 over optical
 
Last edited:

Elberoth

Member Sponsor
Dec 15, 2012
2,007
253
1,170
Poland
Talked to dCS people at lenght regarding the ST interface - according to Martin Reynolds 'the problem with optical interfaces is that you do not know what the light intensity in the fibre will be, so the receiver needs an AGC system (Automatic Gain Control). Such systems tend to be quite jittery due to the AGC circuit continually hunting for the best threshold setting.

In listening tests, the wired AES and SPDIF interfaces sounded better than either Toslink or ST for this reason. Both of these interfaces are transformer coupled (another way to galvanically isolate), although the SPDIF standard spoils this by requiring a chassis connection at both ends
'.

I think a dual ST interface (with a separate ST line feeding clock signal back to the transport) should be the one least compromissed.
 

Phelonious Ponk

New Member
Jun 30, 2010
8,677
23
0
Attempts at optical isolation using TOSLINK can cause it's own audible issues so why would we want to substitute one set of problems for another?

Well, because one set of problems may be significantly less audible than the other. But you know how that is determined - measurements - and some folks around here break out in hives at the thought.

Tim
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
Talked to dCS people at lenght regarding the ST interface - according to Martin Reynolds 'the problem with optical interfaces is that you do not know what the light intensity in the fibre will be, so the receiver needs an AGC system (Automatic Gain Control). Such systems tend to be quite jittery due to the AGC circuit continually hunting for the best threshold setting.

In listening tests, the wired AES and SPDIF interfaces sounded better than either Toslink or ST for this reason. Both of these interfaces are transformer coupled (another way to galvanically isolate), although the SPDIF standard spoils this by requiring a chassis connection at both ends
'.

I think a dual ST interface (with a separate ST line feeding clock signal back to the transport) should be the one least compromissed.

The problem of the variable threshold using ST is well known - but if designers wanted they could solve it easily defining a fixed threshold at switch on of the equipment or when switching the inputs. VTL uses a similar procedure to set the bias for their tube amplifiers at turn on. The listening tests you refer were probably carried using the standard ATT interface modules used for telecommunications and computers, and by digital audio designers.

It is only faith - I do not have any real support for it - but currently my "preferred" digital connection is also the AES/EBU.
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
Well, because one set of problems may be significantly less audible than the other. But you know how that is determined - measurements - and some folks around here break out in hives at the thought.

Tim

Tim,

Nice to know that you agree with us that the levels of jitter we are debating are audible - most people who find the solution for everything in audio using simple measurements will say it is not.

BTW, any one wanting to research why there is no 100% isolation can find the answer in "Grounding and Shielding: Circuits and Interference" by Ralph Morrison. http://www.amazon.com/Grounding-Shielding-Interference-Ralph-Morrison/dp/0470097728#reader_0470097728
 

Phelonious Ponk

New Member
Jun 30, 2010
8,677
23
0
Tim,

Nice to know that you agree with us that the levels of jitter we are debating are audible - most people who find the solution for everything in audio using simple measurements will say it is not.

BTW, any one wanting to research why there is no 100% isolation can find the answer in "Grounding and Shielding: Circuits and Interference" by Ralph Morrison. http://www.amazon.com/Grounding-Shielding-Interference-Ralph-Morrison/dp/0470097728#reader_0470097728

Not sure that I do, micro. What levels are you debating?
 

amirm

Banned
Apr 2, 2010
15,813
37
0
Seattle, WA
MY takeaway from our Jplay conversation was that the inherent noise of a computer's OS/hardware would swamp any possible improvement that software could make.
Only in the context of high performance systems. Here is an example post from me:

Here is the other problem: the audience for this player is not the guy who is running the motherboard S/PDIF or a DAC on the computer. Instead, it is someone who is likely using a high-quality interface that is not dependent on the quality of the clock out of the PC. Assuming so, then the situation gets hard, really hard. That is because a proper async interface does not use any clock out of the PC. It has its own system similar to what I described above where the DAC calls the shots as far as when it needs the next chunk of data. In that sense, it has isolated itself from the PC clock altogether.

And I actually provided a similar scenario to what I tested here to Jplay author:

As I noted to John, you are making specific technical claims, connecting certain dots. That theory can be easily tested as the bloggers did. And their results, says that your theory is not valid in their tests. Of course they did not test all scenarios. So maybe it matters elsewhere. Problem we have is that you, as the proponent of these changes, have no data either. I even gave you the low hanging fruit: take the PC motherboard sound and see if your changes make a difference there. I think there is some chance that it might there. But you are not going there. And certainly have not in the past.

I remember the logic being along the lines that the OS activity itself was far more critical than & swamped any small contribution that application software had. So I was surprised to see your measured differences between WMP & MP classic which seems to point to a very significant difference between the USB output when different playback software is used. I'm also not sure how "let WMP read the file for a bit & settle down" resulted in a better performance? Any ideas?
Answering the last part, WMP's system load is a lot higher at the start than the rest of the stream.

A "high quality interface" can mean all sorts of different things to different people. For instance, as most asynch USB DACs now operate at USB2 high-speed, I think you will find it difficult to find one which galvanically isolates at the 480Mbps USB input speed (although it's possible to isolate the signal & grounds further downstream but how many DACs do this?).
Using a USB isolator is the cheap way out. As you say, there are existing single-chip solutions for that. There is however a discrete implementation where you decode USB and then put isolators in your data path to the DAC side.

BTW I just wrote an article on this for WSR. Should be in print in a few weeks and then I will post it.

Yes, pity you didn't get to run JPlay on that DAC but maybe differences will show up in measurements using another DAC?
The real test is with high quality interfaces anyway which I still have.
 

jkeny

Industry Expert, Member Sponsor
Feb 9, 2012
3,374
42
383
Ireland
Only in the context of high performance systems. Here is an example post from me:

And I actually provided a similar scenario to what I tested here to Jplay author:
Yea, I looked back over that Jplay thread & saw those posts.

Answering the last part, WMP's system load is a lot higher at the start than the rest of the stream.
OK, I see


Using a USB isolator is the cheap way out. As you say, there are existing single-chip solutions for that. There is however a discrete implementation where you decode USB and then put isolators in your data path to the DAC side.
Yea, but not a lot of DACs do this & it's debatable that they offer 100% isolation, anyway.

BTW I just wrote an article on this for WSR. Should be in print in a few weeks and then I will post it.
Great, I look forward to your articles as they are always very interesting


The real test is with high quality interfaces anyway which I still have.
Interested in your results if you get the time to do some measurements.
 

Elberoth

Member Sponsor
Dec 15, 2012
2,007
253
1,170
Poland
Only in the context of high performance systems. Here is an example post from me:
Using a USB isolator is the cheap way out. As you say, there are existing single-chip solutions for that. There is however a discrete implementation where you decode USB and then put isolators in your data path to the DAC side.

This is what Berkeley is doing in the Alpha USB interface. It is interesting to note though, that even the Berkeley designers thought that isolation the data lines* is not a silver bullet and many extra steps have to be taken to properly isolate the compouter from the DAC.

* Berkeley is using one of the ADuM chips between the USB receiver chip and the SPDIF output. The ADuM chips are digital isolators based on the Analog Devices iCoupler® technology, that combine high speed CMOS and monolithic transformer technologies.
 

jkeny

Industry Expert, Member Sponsor
Feb 9, 2012
3,374
42
383
Ireland
This is what Berkeley is doing in the Alpha USB interface. It is interesting to note though, that even the Berkeley designers thought that isolation the data lines* is not a silver bullet and many extra steps have to be taken to properly isolate the compouter from the DAC.

* Berkeley is using one of the ADuM chips between the USB receiver chip and the SPDIF output. The ADuM chips are digital isolators based on the Analog Devices iCoupler® technology, that combine high speed CMOS and monolithic transformer technologies.
Yes but looking at that Berkely implementation, it looks like the I2S lines are asynchronously reclocked after the ADUM isolator - is it using an ASRC chip there?. As you know, I'm not talking about asynch USB - that is a totally different thing.

To me, this suggests that it will tend to hide/mask the differences between software players (& other upstream improvements). I have heard this on a number of devices that have asynch reclocking - if it can be turned off, the sound is improved if you are feeding it a high quality signal. When ASRC is turned on, the sound becomes very similar for all sources i.e the better sources don't reveal their qualities. ASRCs improve the sound (up to a point) for lower quality (jittery?) source signals but limit the sound (to this same point) for better quality source signals. Even the ESS DAC chips, which have one of the most sophisticated ASRCs around, sound better when operated synchronously (once fed a quality signal).

It's one of the great gotchas in digital audio - if all sources sound the same, is it because they are all being brought to this SQ point determined by an ASRC or some other limiting factor in the build or is it because the system is truly optimal for all sources?

I have experimented with a number of isolators on the I2S lines (NVE, Silicon Labs, ADUM) to completely isolate the USB receiver from the downstream DAC or SPDIF transmitter chip. In order to avoid asynchronous reclocking (the XMOS definitely needs reclocking on it's jittery I2S output signals) you need to have a number of isolated signals besides the I2S signals - an isolated clock enable signal (from the USB side) to select the correct clock speed family (two audio clocks on the "clean" side); the isolated clock signal needs to be fed back to the USB receiver/I2S generator in order to use the same clock as used in the reclocker.

In my opinion, this is the optimal way of doing it.

In my experiments, I found no difference with this isolation (but my system is pretty immune to CM noise & ground loops) - the upstream software still had an influence on the sound. I double checked my isolation & ensured that it was correct. I tried with battery power Vs USB power on the "dirty" USB side & always used battery power on the "clean" side of the isolator, so no possibility of any "sneaky" ground connections via mains connections

Edit: Looking closer at the internal pics of the Alpha, I don't see any ASRC chip & possibly a clock enable line but I can't see how the clock signal is fed back to the XMOS chip - there are only 4 lines into/out of the ADUM isolator
 
Last edited:

jchong

New Member
Aug 29, 2013
2
0
0
My son bought a $400 DAC that claimed to have async USB support but immediately found out that he got PC noise through it just like his other solutions.

To test my son's experience of him hearing system activity

Good post and topic Amir. Being new to computer based audio via USB, this is of interest to me.

What kind of PC noise or system activity was he hearing? I'm getting some noise too particularly when my HDDs are active (e.g. playing a movie), it's like a low volume high pitched noise. When the HDD is not active the noise goes away. Is this what your son experienced?

Secondly, what is your take on what is causing the noise? From your article it seems you ascribe it to the software player.

It's often said that the USB bus is noisy (from all the devices that might be connected to it) more particularly due to contamination of the USB power line in the cable. Is this the root of the problem? Thinking about my case, I play a movie which loads from an external USB HDD, this introduces noise into the USB bus which then gets into my USB DAC and hence causing the noise. Valid thinking?

If the noise is power related, would even an async USB DAC be able to eliminate it?
 

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing