Mains conditioner for SMPSs?

[mansr]

Unplug the SMPS strip and see if you can hear any difference in noise from the speakers with the volume turned up to 11. If not, a power conditioner will be of no benefit.

[mansr]

If one fears that a device might be harming the sound quality, is it not a valid test to unplug said device and check if the sound improves?

[mansr]

3 hours ago, Blackmorec said:

Again, this my layman’s understanding. For a more thorough explanation you’d need to talk to a specialist who understands both digital and analog domains.

Yes, someone like marce.

[mansr]

1 hour ago, Blackmorec said:

OK. Which pieces of top class audio gear has marce developed?

I have no idea, but he seems to have a decent understanding of EMC issues.

[mansr]

49 minutes ago, Blackmorec said:

Sorry to be so blunt, but that’s just rubbish.  Did you know that there are 4 levels of knowledge....

1. Unconscious ignorance...where you don’t know what you don’t know
2. Conscious ignorance....where you know what you don’t know
3. Conscious knowledge....where you know but need conscious effort to recall
4. Unconscious knowledge....where you know it so well it is automatic and requires no effort

With a great deal of certainty I can say that the above statement falls into category 1. 

You forgot:

 

5. Wilful ignorance

 

You can lead a fool to knowledge, but you can't make him think.

[mansr]

48 minutes ago, semente said:

Does everyone agree that every commercial equipment is built to a price point,

Everything is built with some budget in mind, be it that of a plumber, a lawyer, or the DoD.

 

48 minutes ago, semente said:

its performance in practice is limited by the knowledge/ experience/expertise of the designers,

Also physical reality.

 

48 minutes ago, semente said:

and can be improved?

Depends on the nature of the device, which aspect you'd like to improve, and how much you're willing to spend. For some items, even a modest investment gets you close to what is physically possible (cables come to mind).

[mansr]

27 minutes ago, semente said:

So you agree.

Not in the way you think.

 

27 minutes ago, semente said:

So you agree that commercial hi-fi electronic equipment and speakers can be improved.

I said it depends on what you want to improve. Take the Voyager probes, for example. They are still working after spending 40 years in space with no maintenance. Would any hi-fi equipment (commercially available or not) fare as well? Unlikely. A few hundred million dollars (the cost of the Voyager programme) might buy you that improvement. On the flip side, if reducing cost by, say, 90% only requires giving up a marginal amount of performance, that too could be considered an improvement. Saying that "everything can be improved" is meaningless without some context.

[mansr]

2 hours ago, Blackmorec said:

But hi-fi ...used in a domestic environment to connect all kinds of solid state and tube equipment; some hot, some cold; some balanced, some single ended; some close together, some far apart, some with low output impedances, some high; some requiring cable with low inductance, some perform better with high; some extremely expensive, high resolution components some cheaper less revealing units.

And still it all just works. What does that say about hi-fi?

[mansr]

8 minutes ago, Blackmorec said:

I don’t want to study electronics. Not remotely interested.

Then perhaps you should stop making such grand pronouncements about it.

[mansr]

42 minutes ago, semente said:

Is there a difference in the way a computer sends data to another computer by say Ethernet (could it be data verification?) and the way a computer sends a data stream to a DAC?

Is the DAC also using Ethernet? Then the lower layers (Ethernet, IP, TCP/UDP) are exactly the same. Above this, protocols may vary. UPnP streaming usually uses HTTP requests to fetch the audio files just like a regular web browser would. There are also audio specific protocols such as RTP and Dante.

 

42 minutes ago, semente said:

Is it correct to differentiate between the two data transmission and data streaming?

No. The network doesn't know or care about the purpose of data packets.

 

42 minutes ago, semente said:

Is there a way to determine/measure if the data stream reaching the D/A chip is exactly the same which was transmited by the processing software?

Yes, monitor the I2S pins going into the DAC chip and compare the data with what was sent.

 

42 minutes ago, semente said:

Are there way to test the possible impact of noise/interference being sent along with the digital signal (voltages?) in D/A performace?

Sure, inject known noise into the interface and measure the output of the DAC.

[mansr]

5 minutes ago, semente said:

I was thinking about USB.

 

Again, I was thinking about a computer feeding a DAC using USB.

You mentioned Ethernet, and I assume you know that USB is a rather different connection.

 

5 minutes ago, semente said:

Can you somehow determine how clean and crisp the analogue DC voltage levels are and if they are affected by noise?

Which DC voltages?

 

5 minutes ago, semente said:

What would you measure and what would you expect to find if noise were to affect the D/A conversion?

I don't think that it would be possible to compare the input with the output, would it? Could you meaningfully compare conversion (analogue output) of a clean vs a dirty signal, not static test tones but live streaming music or at least something more complex and varied?

If there is a problem, it will be revealed by a standard suite of tests.

[mansr]

5 minutes ago, semente said:

I though that the digital signal consisted of changes in voltages when transmited through wire.

Changing voltages are not DC.

[mansr]

4 minutes ago, semente said:

My bad.

Can those voltage changes be affected by noise in a way that affects D/A conversion, that was my question?

If we're talking about the data inputs, too much noise will cause errors (bit flips). This would be immediately audible, so assume it's not happening. Noise on the clock input will increase jitter somewhat. In addition to these primary effects, it is possible that high-frequency noise can couple through to the D/A stage and somehow affect the output. The exact mechanism as well as the nature of the effect will depend on the architecture of the DAC. Most likely, we'd be looking at a slightly increased noise level.

 

What matters is that any audible change in the output is readily measurable, as is whatever caused it.

[mansr]

8 hours ago, Blackmorec said:

So let me understand this. You can look at the analog output of a DAC...a music signal and measure any additions caused by noise or jitter and use that measurement to diagnose what caused that change to the analog signal? Is that correct?  I’m wondering; within that analog signal, how do you differentiate whats music and supposed to be there from whats been caused by the jitter and Hf contamination and not supposed to be there? Can you for example take an analog output signal and directly measure the contribution of phase noise or would you need to set up a test jig to inject jitter and see how the analog output signal was modified? And once your know the answer, how would you know how much jitter was actually present in a given system i.e how would a user know if they had a problem with Hf noise or jitter?

i guess what I’m trying to understand is, how would a user know whether they had a problem or not with their system in their environment. What would the measurements look like, how would they be made? 

Measuring such things using purposely designed test signals is much easier than using music which is already a very imperfect signal. To check for jitter, we can simply play a pure tone and measure the spectrum of the output. Ideally, we'd then get a single narrow peak. If there is jitter, the spectrum will show some combination of additional side tones and widening of the peak. Here's a horrible example:

 

If we want to know what is causing this, we have to measure the inputs to the device. For instance, we can check for jitter on the clock input. If the device doesn't have any jitter attenuation circuitry (a PLL), clock jitter will pass through to the output since there is no better reference. If there is a PLL, we can test its effectiveness by varying the amount and type of jitter at the input while checking the effect at the output.

[mansr]

11 minutes ago, Blackmorec said:

That was a very reasonable and helpful answer, which I understand perfectly. So engineers can measure these types of noise and resulting distortions, but it is, I’m sure you’ll agree, beyond the scope of your average audiophile, given that the measurements take some fairly specialised instruments, test tones with added jitter and specialised knowledge on how to conduct the test and evaluate the results.

Of course the average person won't be able to perform these tests, nor should they have to. However, someone producing and selling devices intended to address supposed problems really ought to be able to show that the desired effect is actually delivered. When we say show us the measurements, it is not directed at you, the end user, but at the likes of Synergistic Research and Uptone.

[mansr]

38 minutes ago, fas42 said:

What isn't measured is how components interact with each other - one obvious example is when all the boxes are plugged into the same effective socket, and the varying current draws by the individual power supplies add constantly changing noise to the mains voltage - power amplifiers are a major contributor here ... no-one is telling you anything about the robustness of the designs here.

Sure they are. It's called power supply rejection ratio.

[mansr]

1 minute ago, semente said:

Is there a chance that ultrasonic rubbish could be affecting the amplifier's performance?

It could. One possibility is intermodulation creating audible artefacts. Even if not, driving high-level ultrasonics into the tweeters burdens the amp for no benefit.

 

1 minute ago, semente said:

How often is the range from 20 to 1000 kHz measured at the output of the DAC?

Is 3 GHz enough for you?

 

1 minute ago, semente said:

And at the output of the amplifier?

Few amps have a frequency range extending much beyond 100 kHz, so there's not going to be anything of interest to measure.

[mansr]

15 minutes ago, semente said:

Would this cause a "bit flip"? (red channel at 1 msec, both channels at 4.65 msec)

 

Prism Sound Callia, waveform of undithered 1kHz sinewave at –90.31dBFS, 16-bit TosLink data (left channel blue, right red).
Read more at https://www.stereophile.com/content/prism-sound-callia-da-headphone-amplifier-measurements

That's a noisy rendering of a low-level 1 kHz tone. It has nothing to do with digital data transmission.

[mansr]

3 minutes ago, Summit said:

Jriver and many other music players are transferring music data from SSD to RAM and have done it long before Innuos, so what’s you point?

The way computer systems are designed, it is impossible to access a single bit of data from a storage device without placing it in RAM first. When an application requests data from a file, the OS works out where on the storage device this file resides. Then it sends a request to the storage controller with an instruction to read the relevant block and place it at a specified location in RAM. When the transfer is complete, the controller raises an interrupt, and the OS lets the application continue.

[mansr]

4 hours ago, semente said:

With a bit rate of 5,644,800 bits per second per channel for DSD128 (I hope I didn't get this wrong) what's the chance of noise interference flipping a few bits over USB (I'm assuming perhaps wrongly that USB doesn't check for the accuracy of the transmission)?

The USB 2.0 bit rate is always 480 Mbps. Since the data rate of audio is lower, the interface is idle most of the time. As for chance of bit flips, the USB 2.0 spec recommends that implementations achieve an error rate of 1 in 1e12 bits or better.

 

4 hours ago, semente said:

What would a flipped bit sound like?

A single bit flip in DSD isn't very noticeable. In PCM, it depends on which bit. With 24-bit samples, any of the lower 8 or so bits can be flipped without anyone noticing. Flipping the higher bits will produce a tick, louder the higher in the value of the bit.

 

5 hours ago, semente said:

Would the extra effort required to process a noisy signal affect D/A conversion?

What type of interface are we talking about here? The idea that a noisy signal takes more "effort" to decode is very often simply wrong.