I am not responding to the power supply/noise thing. I was responding to someone saying you can't build a music server with state-of-the-art performance. I showed a quick example I had online that said otherwise.
Baffled about computer power
- Thread starter mauidan
- Start date
- Status
But that's part of the argument the transport proponents are making here - that PCs are too noisy. And a single sine wave stimulus won't show up the difference between a dedicated transport and a PC. A multitone just might though.
So no, your measurement doesn't show that your PC isn't reducing dynamics coz its the wrong measurement.
So no, your measurement doesn't show that your PC isn't reducing dynamics coz its the wrong measurement.
Because they already have a good DAC in their current system that only takes S/PDIF or AES??? Regardless, I was commenting on whether you can have a high-end transport using a PC. That by definition calls for an external DAC.
There you go
. I don't think async USB is remotely a standard feature on DACs even though I believe it should be. Some still think that any "computer interface" has the opportunity to degrade DAC performance.The only comparison we can make is for same measurements. Do you have published reports for the high-end transport using the stimulus you suggest?
No, but why would I need them? I'm just falsifying your claim here that displaying a single sine result somehow 'proves' that your PC is delivering SoTA performance. It does nothing of the kind.
Because they already have a good DAC in their current system that only takes S/PDIF or AES??? Regardless, I was commenting on whether you can have a high-end transport using a PC. That by definition calls for an external DAC.
There you go
To clarify..
-you would only a recommend a converter if someone had a perfectly good DAC they were very happy with without a USB input.
As far as new DACs are concerned.. honestly, I can't name a manufacturer who has put out a DAC in the last 2 years without an Async USB input. Can you?
When (or if) they do power a speaker they'll quickly degrade that level of performance because speakers are non-linear loads. Meaning to get THD at that level the output impedance presented to the speaker terminals would need to be below a micro-ohm. Since that's totally impractical to achieve (would require feedback, 4-terminal sensing) chasing such low THDs is pointless in practice.
A few years ago I pointed this out on a forum where the engineers present passed their time talking about the design of poweramps with sub 0.001% THD. I received a lifetime ban for my contribution
You proposed a test. Yet you have no results to share using it to show its efficacy and the bar set by any product. You need that if want me to care about what you proposed.
Same back to you
. You throw out some vague metric for which you have no data to share. So even if I had my data for a PC, you wouldn't have anything to compare it to. What I shared is industry standard data and is the most common measurement for transports. I am open minded to anything new you propose but please put aside the personal tone and be constructive.
I prefer to call it an 'experiment' rather than a test (I have an aversion to tests). And yes its true that I haven't done the experiment myself using a multitone stimulus to discern the differences caused by CM noise pollution.
I have no such want. If you'd like to make progress in understanding SQ differences, you might wish to perform the experiment. If not, then don't.
Same back to you
But our two situations are not symmetrical. You're putting forward measurements which you claim say something relevant about your kit. I explained why they do not. I put forward an experiment you could perform should you so desire. A proposal for an experiment is not a claim.
This is nonsense. I agree I was vague about the number of tones so let me add detail. Try with a stimulus of 100 or more tones. Do you need any more detail to perform the experiment? If you still consider the experiment a 'vague metric' then explain what's missing.
I don't follow what you are saying.
Firstly why do you need to do a comparison? You could begin by comparing the noise floor with a digital zero stimulus with that under a multitone stimulus. The digital zero would give you a baseline for determining the degree of noise modulation.
Secondly, just because something is 'industry standard' doesn't mean its any practical use. In this case I've explained why not, if you disagree then what have I said so far which isn't correct? Show me my faux pas please.
Thirdly you're claiming something about 'personal tone' - chapter and verse please as this looks to be your own personal perceptual distortion to me. If you have evidence for the 'personal tone' then I'd like to see what it is.
Finally, I'm being constructive here so don't understand why you request me to be constructive. What in what I have so far written isn't constructive - chapter and verse again please.
Only if you accept that this jitter measurement graph says it all about the sound quality of a music server and decide to never look (or hear) back ...
I am not an expert and do not pretend to be able to interpret jitter measurements. But please look at the jitter measurements of the Vivaldi DCS stack taken from the exhaustive suite of measurements (I think more than an hundred ...) by Paul Miller for HifiNews and available on line in his site. Everyone I know about who listened to both USB and CD transport playing the same recording (ripped and verified CD) preferred the transport sound quality. In my ignorance I would say a different think if just looking at the graphs. Can I have your comment on them?
Attachments
Indeed, & I wonder has Amir tried listening to any other asynch USB input DAC Vs Audiophileo feeding the DAC? I would be interested in his impressions!
Objectively, the measurements show superlative results when the right gear is used. Here is my standard work-laptop, with a ton of apps and stuff running on it, driving a Berkeley Alpha USB to my more than 10 year old Mark Levinson 360S Dac using async USB:
Only two correlated spikes are are visible at <-120 dB. That is distortion product below the noise level of a 20 bit signal. Again, this is driving an old but well designed DAC.
There is no cheaper and more convenient way to get into "high-end" audio than getting a good async USB and pairing it with the DAC of your choice. These devices get the best out of any DAC.
Is this with digital silence? If so, would one also be able to measure noise with a complex music signal? And can we establish that USB is a superior interface over AES/ABU with respect to noise? Also, this graph is really half the picture - just noise, no jitter info. A state of the art transport would excel at both, and so would a PC-based server...
The problem is separating out the noise from the complex music signal. With silence, or with a (few) sine tones you can tell which is which, but how do you know what is music and what is noise?
Why would it be? From a noise point of view, optical TOSLINK ought to be the best one, as there is no galvanic contact at all between the source and receiver.
No, that is the J-test signal. J-Test is a (near or full scale) square wave that has its low-order bit toggled. The square wave once through the DAC gets filtered to a single fundamental that shows up as that tall spike. An ideal system would only put out that one spike (i.e. the signal itself). A non-ideal system adds more spikes for various reasons (jitter, hardmonic distortion, correlated noise, random noise, etc.). In order to pull out more of the jitter, J-test uses a rather high frequency of Sampling Rate/4 so in the case of 48 Khz, it is 12 KHz.
You would have to factor out the music to have the noise fall out. You further need to dial out the noise of the instrument itself.
The graph showed two distortion products. One of which I have the cursor on at 15.618 Khz. The other is at 8.382 KHz. Both differ from the principal tone (12 KHz) by a frequency of 3618 Hz so we know that is jitter since we have symmetrical distortion tones. There may also be jitter that is random and hence lost in the noise floor. Or there is more deterministic jitter that also gets lost in the noise floor of the system.
I had one such device in my testing from which I am posting these results (my WSR magazine article on HDMI vs S/PDIF performance). It was the Peachtree Decco65 which is an integrated AMP/DAC (I only tested the DAC portion). I tested it three ways, with its own async USB input, with the Audiophilleo and with Berkeley:
The tallest spikes in gray are that of Decco's own USB. The next tallest spikes are Audiophilleo in red and the lowest in green, are the Berkeley. All of these are of course at exceptionally low levels once more so the distinction there is matter of engineering pride, than meaningful differences. The regular set of spikes in Decco's own USB input are at 60 Hz so appear to be power supply related.
As to listening, no I did not have enough time. I only had the unit long enough to do the tests and had to return it.
No it is not "all" that one can say about audio quality. I am not writing a thesis on the topic
. Simply adding some data points, taking us from mere words to something concrete. With respect to transports, where digital samples cannot change, measurements of jitter is a super strong component and that is what I shared. To prove the sensitivity of the test in showing such differences, this is my same laptop, driving an AVR using its HDMI input:
We see a substantially degraded (measured) performance. BTW, the two traces are the same measurement just changed in time! Once the AVR gave me the top results, next the lower. Such variations (triggered by activities internal to the machine) did not exist with S/PDIF.
From an objective point of view, those are superb results. They are somewhat incomplete however in that the stop short at maximum jitter of 3.5 Khz. I rather see the full spectrum as effect of masking diminishes with increased jitter frequency.
If you are asking me to make sense of why people have the preferences they do, we get into the same classic argument of not knowing if those observations, even if expressed by a million people, are accurate. We know the measurements however, are accurate. So in order to get some place rather than turn this into the classic back and forth, we should focus on the data. Once we establish that the measurements show state-of-the-art performance, then the area of disagreement becomes smaller.
I had one such device in my testing from which I am posting these results (my WSR magazine article on HDMI vs S/PDIF performance). It was the Peachtree Decco65 which is an integrated AMP/DAC (I only tested the DAC portion). I tested it three ways, with its own async USB input, with the Audiophilleo and with Berkeley:
The tallest spikes in gray are that of Decco's own USB. The next tallest spikes are Audiophilleo in red and the lowest in green, are the Berkeley. All of these are of course at exceptionally low levels once more so the distinction there is matter of engineering pride, than meaningful differences. The regular set of spikes in Decco's own USB input are at 60 Hz so appear to be power supply related.
As to listening, no I did not have enough time. I only had the unit long enough to do the tests and had to return it.
Thanks Amir, it's a pity you didn't listen. I predict that the Audiophileo would have sounded noticeably better than the Decco's Asynch USB but not because the plots reflect this. If we were to predict from the plots, we would probably come to the same conclusion as you "exceptionally low levels once more so the distinction there is matter of engineering pride, than meaningful differences." But if you listened (yes blind or whatever failsafe way you wanted) you may have come to another conclusion.
I tell you what, I'm sending one of my asynch USB DAcs to Andre Marc for review - if you want to listen to it & test it, I can have it passed onto to you?
Thanks for the offer John. I will take you up on it later. Right now, I am up to my eyeballs trying to get my boat electronics to work. And the boat itself! . Until I get through that, I won't have much time to do anything else. But I am curious about your product.
. Until I get through that, I won't have much time to do anything else. But I am curious about your product.It won't be available to pass onto you until about a month away anyway! Look forward to your impressions!Thanks for the offer John. I will take you up on it later. Right now, I am up to my eyeballs trying to get my boat electronics to work. And the boat itself!
As Opus111 has studied noise for a while now, I'm surprised nobody has asked him if he has any such graphs/measurements to backup his theories?
I will - any meat to throw on the bones of this idea that changing the PS reduces (or affects) the CM noise? Indeed any measure of CM noise on USB lines from a PC?
I will - any meat to throw on the bones of this idea that changing the PS reduces (or affects) the CM noise? Indeed any measure of CM noise on USB lines from a PC?
- Status
Similar threads
