@Markw4
First of all, let me say that this is just my opinion.
ex, virtual sound technology requires research on human listening method,
but stereo DAC has a clear goal (converting binary code to analog based on VREF).
In this case, I think measurement is the only tool.
Just as PCM (and PDM) audio data cannot be infinite in length,
in order to open up infinite possibilities for our hearing ability,
the physical size of the body system for better hearing must also be infinite.
I cannot see the contents of a leaflet 10 km away.
On the other hand, telescopes were born for that.
In short, "rather than doubting the capabilities of the measuring equipment,
it is a priority to think about how to use this tool to obtain useful data."
First of all, let me say that this is just my opinion.
ex, virtual sound technology requires research on human listening method,
but stereo DAC has a clear goal (converting binary code to analog based on VREF).
In this case, I think measurement is the only tool.
Just as PCM (and PDM) audio data cannot be infinite in length,
in order to open up infinite possibilities for our hearing ability,
the physical size of the body system for better hearing must also be infinite.
I cannot see the contents of a leaflet 10 km away.
On the other hand, telescopes were born for that.
In short, "rather than doubting the capabilities of the measuring equipment,
it is a priority to think about how to use this tool to obtain useful data."
It kind of looks like you are confusing what a spectrum analyzer measures with what humans hear. FFT spectrum analyzers discard phase information and only show frequency magnitude of individual bins. Also, one way to perform a DFT is by measuring average signal correlation with bin frequencies. That can be important to recall later when we get around to talking about weird things Sigma Delta modulators can do.
Moreover, its probably worth thinking about noise. Taking white noise as one example, it can have a gaussian distribution, or it can be a dirac impulse function, or pretty much anything in between. A dirac pulse is obviously very loud for a very short period of time. It all looks the same as viewed in spectral analysis. For example, it can all look like a low level flat noise floor.
Again considering noise, in a dac, clock period noise and or Vref noise are multiplied by the dac output array to produce the analog output of the dac. In that case the audio signal has been intermodulated with noise and noise sidebands have been produced around the signal frequency. If you look at that in a typical audio spectrum analyzer view, you will only see the signal and never know what's happened to the sound.
Also, it is quite noncontroversial that humans are sensitive to phase at lower frequencies, but that often isn't measured. Spectrum analyzers can't show it.
In addition, humans can localize lateral sound sources, real or virtual in a stereo sound field, to within about 3-degrees of accuracy using something called Interaural Time Delay (ITD). It has to do with phase coherence between stereo channels when the channels are playing two different audio streams. The time difference can be as low as a few microseconds. It is typically not measured.
So, all the talk you hear over at places like ASR is by people who are ignorant and or pretend the above types of things are inaudible to humans. IOW, there are people who believe the only thing wrong with a dac that can audible to humans is nonlinear distortion like HD/IMD, and or a fixed noise floor. To believe that those people must be in some way deaf and or maybe their systems reproduce audio poorly because they relied too much on the wrong measurements.
So, what do you think about that?
Moreover, its probably worth thinking about noise. Taking white noise as one example, it can have a gaussian distribution, or it can be a dirac impulse function, or pretty much anything in between. A dirac pulse is obviously very loud for a very short period of time. It all looks the same as viewed in spectral analysis. For example, it can all look like a low level flat noise floor.
Again considering noise, in a dac, clock period noise and or Vref noise are multiplied by the dac output array to produce the analog output of the dac. In that case the audio signal has been intermodulated with noise and noise sidebands have been produced around the signal frequency. If you look at that in a typical audio spectrum analyzer view, you will only see the signal and never know what's happened to the sound.
Also, it is quite noncontroversial that humans are sensitive to phase at lower frequencies, but that often isn't measured. Spectrum analyzers can't show it.
In addition, humans can localize lateral sound sources, real or virtual in a stereo sound field, to within about 3-degrees of accuracy using something called Interaural Time Delay (ITD). It has to do with phase coherence between stereo channels when the channels are playing two different audio streams. The time difference can be as low as a few microseconds. It is typically not measured.
So, all the talk you hear over at places like ASR is by people who are ignorant and or pretend the above types of things are inaudible to humans. IOW, there are people who believe the only thing wrong with a dac that can audible to humans is nonlinear distortion like HD/IMD, and or a fixed noise floor. To believe that those people must be in some way deaf and or maybe their systems reproduce audio poorly because they relied too much on the wrong measurements.
So, what do you think about that?
Last edited:
Depends on what the goal is. If the goal is to "faithfully" reproduce what is on the recording then measurements are the best tool. OTOH if the goal is to have a subjectively pleasing sound then listening has greater value. However this subjectively pleasing sound may not please everybody. Needless to say both measurements and listening evaluation are needed for best results.
@Markw4
I agree, not misusing.
I think what you said is a good quality for a scholar to have.
But in reality, there are too many obstacles.
Please also mention the music files that are being distributed.
What do people in the music distribution industry know?
What is the level of the device used to create the music file?
What is the TF of the living room with the DAC and speakers?
Do you think it is more dominant than these variables?
Also, the listening evaluation that changed before and after the
double-blind test for the same product clearly showed the credibility of the ear.
If it is difficult to distinguish it from the market products in the double-blind test, It is not bad as a DIY result.
and to do that, understanding the IC well enough, make the layout beautiful enough, put it in a metal case, and measure it.
planning to purchase through numbers gather. don't have to be too tired because are consumer, not an engineer.
Before that, people who review products on YouTube should stop.
Thank you for your kind words.
I agree, not misusing.
I think what you said is a good quality for a scholar to have.
But in reality, there are too many obstacles.
Please also mention the music files that are being distributed.
What do people in the music distribution industry know?
What is the level of the device used to create the music file?
What is the TF of the living room with the DAC and speakers?
Do you think it is more dominant than these variables?
Also, the listening evaluation that changed before and after the
double-blind test for the same product clearly showed the credibility of the ear.
If it is difficult to distinguish it from the market products in the double-blind test, It is not bad as a DIY result.
and to do that, understanding the IC well enough, make the layout beautiful enough, put it in a metal case, and measure it.
How did get that fact without the machine?
The ASR community is just a place where people who want to be sure about the products they have purchased or are
planning to purchase through numbers gather. don't have to be too tired because are consumer, not an engineer.
Before that, people who review products on YouTube should stop.
Thank you for your kind words.
Last edited:
@Mins
For now I will agree with bohrok2610 conclusion in #103.
You have asked a lot of questions in #104 for which there are answers, but for the most part I would like to defer getting too deep into those things until later.
What I would like briefly mention though is that when you talk about double blind tests, you need say what kind of double blind tests. There are many ways to perform scientific perceptual tests using a double blind protocol.
Where there is a problem is not with double blinding, rather its with ABX protocol. Among professional perceptual scientists it is well known that ABX tends to be biased toward false negatives (although it can be overcome with sufficient training of test subjects and by limiting how many trials are allowed per day). What tends to happen in reality though is ABX is used by amateurs, often trained engineers, but not trained in perceptual science, who do not adequately train test subjects and or limit trials per day. Also, often the test system is not properly vetted to make sure it is capable of reproducing small differences being tested for. So the result of such "tests" tends towards an erroneous conclusion that people can't hear small differences (which test subjects actually could hear if tested properly).
To be very clear, I am not against DBT itself. I am against the misuse of badly conducted DBT to try to convince people they can't trust anything they hear.
For now I will agree with bohrok2610 conclusion in #103.
You have asked a lot of questions in #104 for which there are answers, but for the most part I would like to defer getting too deep into those things until later.
What I would like briefly mention though is that when you talk about double blind tests, you need say what kind of double blind tests. There are many ways to perform scientific perceptual tests using a double blind protocol.
Where there is a problem is not with double blinding, rather its with ABX protocol. Among professional perceptual scientists it is well known that ABX tends to be biased toward false negatives (although it can be overcome with sufficient training of test subjects and by limiting how many trials are allowed per day). What tends to happen in reality though is ABX is used by amateurs, often trained engineers, but not trained in perceptual science, who do not adequately train test subjects and or limit trials per day. Also, often the test system is not properly vetted to make sure it is capable of reproducing small differences being tested for. So the result of such "tests" tends towards an erroneous conclusion that people can't hear small differences (which test subjects actually could hear if tested properly).
To be very clear, I am not against DBT itself. I am against the misuse of badly conducted DBT to try to convince people they can't trust anything they hear.
Last edited:
Currently I am simulating a Ferrite+C filter to prevent noise from being amplified.
There is a additional 10nF C to avoid losing high frequency bypassing performance due to damping resistance.
Of course fc should be lower
There is a additional 10nF C to avoid losing high frequency bypassing performance due to damping resistance.
Of course fc should be lower
An ideal filter dint have one. But a real would implementation containing some iron do. A bead has a core or it sent one - right?
//
//
Magnetic hysteresis in ferrites is a subject of increasing research:
One overview: https://purifi-audio.com/blog/tech-notes-1/this-thing-we-have-about-hysteresis-distortion-3
Other info: https://www.google.com/search?sca_e...HXXXNPUQBSgAegQIDRAB&biw=1626&bih=898&dpr=1.5
Also, I would just mention that some people including me have found that jumpering out ferrite beads in dac clock power supplies may result in change in dac sound. Some people feel it is less distorted sounding with the ferrites bypassed.
EDIT: An article discussing the history of ferrites in PCB design and why not to use them is attached to a post at: https://www.diyaudio.com/community/threads/general-purpose-dac-clock-board.413001/post-7765137
One overview: https://purifi-audio.com/blog/tech-notes-1/this-thing-we-have-about-hysteresis-distortion-3
Other info: https://www.google.com/search?sca_e...HXXXNPUQBSgAegQIDRAB&biw=1626&bih=898&dpr=1.5
Also, I would just mention that some people including me have found that jumpering out ferrite beads in dac clock power supplies may result in change in dac sound. Some people feel it is less distorted sounding with the ferrites bypassed.
EDIT: An article discussing the history of ferrites in PCB design and why not to use them is attached to a post at: https://www.diyaudio.com/community/threads/general-purpose-dac-clock-board.413001/post-7765137
Last edited:
Ferrite bead Spice models are a combination of RLC networks. Hysteresis is not modeled.
E.g.
(from https://electronics.stackexchange.c...imulations-for-a-2-4-ghz-esp32-3v3-power-line)
E.g.
(from https://electronics.stackexchange.c...imulations-for-a-2-4-ghz-esp32-3v3-power-line)