Its the original music signal in the time domain, A, plus an error term arising from the imperfection the dac. The dac imperfection has slightly distorted A, partially in phase, and also added noise to it.
Could be. Jitter, Vref noise, variations in shift register output levels and output impedances, resistor tolerances (not to mention tolerances of components in the differential summing circuit which may also include audio band filtering).
There is no reason to assume the system is perfectly LTI either. It isn't.
There is no reason to assume the system is perfectly LTI either. It isn't.
Has this been confirmed by some other party? If not, why are we discussing this? It is about as beneficial as discussing Bybees or Myrtle cable lifters.
I think you need more than some second-order effects to get the musical details to be in phase and the rest of the music in antiphase. Maybe some sort of coarse quantizer with a couple of adders and subtractors would do the trick.
Marcel,
Its like this: If you listen you listen on your simple headphone system, and if you try balanced vs SE, and if you find there is more low level musical details in SE than in balanced, then the logical thing to do is deductively work backwards to find out the explanation. Some low level musical details have gotten lost. There has to be a reason or a set of reasons that are causal.
Again, the brain deconstructs a single time domain waveform into a 100-piece symphony orchestra. We don't know how that works, we only know that it does.
After differential summing we find something has happened to the original signal, A, that has caused it to sound lossy, something like an mp3.
We have to ask ourselves what physical process must be behind the loss. If the differential summing is ideal, then the conditions necessary for loss during differential summing must have occurred before the differential summing takes place.
That is the problem we are trying to understand. No matter how odd or mysterious it may seem, there has to be a physical explanation. If you have a better one than mine, then please let's here it.
Its like this: If you listen you listen on your simple headphone system, and if you try balanced vs SE, and if you find there is more low level musical details in SE than in balanced, then the logical thing to do is deductively work backwards to find out the explanation. Some low level musical details have gotten lost. There has to be a reason or a set of reasons that are causal.
Again, the brain deconstructs a single time domain waveform into a 100-piece symphony orchestra. We don't know how that works, we only know that it does.
After differential summing we find something has happened to the original signal, A, that has caused it to sound lossy, something like an mp3.
We have to ask ourselves what physical process must be behind the loss. If the differential summing is ideal, then the conditions necessary for loss during differential summing must have occurred before the differential summing takes place.
That is the problem we are trying to understand. No matter how odd or mysterious it may seem, there has to be a physical explanation. If you have a better one than mine, then please let's here it.
There are two obvious ones: your listening tests are unreliable or you like even-order distortion.
One of the things I used to do is investigate dangerous medical device problems. Especially when there was in intermittent problem reported by medical professionals, it was not uncommon for the design engineers to respond by sayings, "That's impossible. The system can't do that. The users are doing something wrong or imagining something that isn't real."
Over the years, in all the times something like that happened, in every single case it was the engineers who were suffering from human bias and denial. They didn't offhand understand how the system could do what was reported, so they blamed it on the users. Once it was proven to the engineers that the problems was real, then they always shifted their thinking from denial to troubleshooting. In the end they always found the problem.
Don't know why its that way with engineers. Because the are expert with transistors and what not, they think its just common sense to be a cognitive psychologist and diagnose the users as hallucinating. That is a diagnosis by people that have zero training and or expertise in the field in which they are claiming expertise, cognitive psychology.
Its a human bias to think its always the other guy that's wrong. Engineers are just as guilty of making that type of error as anyone else.
Research to date shows that there is no such thing a generalized critical thinking. People can't be taught general critical thinking. All efforts to teach that have failed. People can only think critically in areas where they have expertise. Engineers should be smart enough to understand what I just said and stop jumping to conclusions about human error in other people who report problems with the things engineers design.
EDIT: For the record, I hate distortion. I told you before that I sorted recording of non-inverting unity-gain audio opamp buffers in order of distortion by ear double-blind. The recordings were made by PMA, who also measured the distortion of each buffer. I sent him a list of the sort order before he publicly revealed which opamp was in which file. You may not remember that, but at the time I told you, you accepted that it happened as I described.
Moreover, I asked you to do your own listening test earlier in this thread with your headphone amp in a bucket thing and a set of cans. IIUC you found just as I stated, and later remarked that what I wrote regarding a flaw was true. Or, did you in fact listen? If not, why did you say there was a flaw I wrote about?
Over the years, in all the times something like that happened, in every single case it was the engineers who were suffering from human bias and denial. They didn't offhand understand how the system could do what was reported, so they blamed it on the users. Once it was proven to the engineers that the problems was real, then they always shifted their thinking from denial to troubleshooting. In the end they always found the problem.
Don't know why its that way with engineers. Because the are expert with transistors and what not, they think its just common sense to be a cognitive psychologist and diagnose the users as hallucinating. That is a diagnosis by people that have zero training and or expertise in the field in which they are claiming expertise, cognitive psychology.
Its a human bias to think its always the other guy that's wrong. Engineers are just as guilty of making that type of error as anyone else.
Research to date shows that there is no such thing a generalized critical thinking. People can't be taught general critical thinking. All efforts to teach that have failed. People can only think critically in areas where they have expertise. Engineers should be smart enough to understand what I just said and stop jumping to conclusions about human error in other people who report problems with the things engineers design.
EDIT: For the record, I hate distortion. I told you before that I sorted recording of non-inverting unity-gain audio opamp buffers in order of distortion by ear double-blind. The recordings were made by PMA, who also measured the distortion of each buffer. I sent him a list of the sort order before he publicly revealed which opamp was in which file. You may not remember that, but at the time I told you, you accepted that it happened as I described.
Moreover, I asked you to do your own listening test earlier in this thread with your headphone amp in a bucket thing and a set of cans. IIUC you found just as I stated, and later remarked that what I wrote regarding a flaw was true. Or, did you in fact listen? If not, why did you say there was a flaw I wrote about?
Last edited:
Mark,
Let’s go back on track.
I suppose you’re waiting for new parts.
In the meantime you could do a few interesting simple tests like listening to the other diff channel with interchanged wires, 2x8 instead of 4x4, interlink termination or whatever else you can think of.
Hans
Let’s go back on track.
I suppose you’re waiting for new parts.
In the meantime you could do a few interesting simple tests like listening to the other diff channel with interchanged wires, 2x8 instead of 4x4, interlink termination or whatever else you can think of.
Hans
It's philosophy of science rather than psychology, Ockham's razor. Simple hypotheses are to be preferred over far-fetched ones, until there is a good reason to refute the simple ones.
I don't know what flaw you refer to, I still have to do the experiment. Actually there are three problems with it:
1. I can always get a null result by not listening well enough. It would be better to have someone who is not sceptical about the effect as a test person.
2. I don't know what kind of music is preferred for it, but that can be solved by asking you. Mark, what kind of music is preferred?
3. The relays in my ABX box make an audible clicking sound. To make identifying which relays click as difficult as possible, I put them as close together as possible, in pairs in a common centroid layout. I may also have put some damping material in the ABX box, I don't remember that.
Hans,
I have a problem. Its true that X5R caps in the dac make the stereo spatial effects 'better.' Its also true they make male voices sound more correct in the midrange. However, I can't stand to listen to the dac for very long because of the distortion they produce. Its ugly. And I'm not the only one who hears it either.
The only suggestion I have from Marcel is to load the presently unused inverting dac outputs to match the output in use. At audio frequencies the reflected load from the transformer is easy enough to approximate with passive components. To match load impedances at RF might be more complicated and time consuming. Also, if it doesn't sufficiently improve the cap distortion then I still have the same problem.
Of course, I knew the film caps were not ideal sound-wise. I was just hoping not too much of their sound would come through to the dac output if they were only used for bypass.
Probably some other solution, such as electrolytics with sufficient damping in parallel with a small film or C0G cap, could turn out to sound closer to correct. However, I don't know if there is physical room for that kind of stuff. Also it could take a lot of trial and error to find a good combination.
Maybe the most expedient thing to do at the moment would be to put the film caps back and live with reduced spatial effects and openness, and with thinner male vocal midrange sound.
Or I could just leave it for awhile. I have some work to do on the other dac that might take few days.
"As far as I'm concerned, the only thing that is flawed about it is the stuff Mark wrote."
https://www.diyaudio.com/community/threads/return-to-zero-shift-register-firdac.379406/post-7427735
What did you mean by "it" if not the dac. Did you mean my report of the sound of the filter board was the "it" that was flawed?
It would have to be someone who has a dac board. Besides, you said you did go to the Purifi website and listen to the two files that had the FFT spectrum yet sounded different. If you can hear that, you should be able to hear what I'm talking about in this case.
It should be music you know intimately, that you always use for test. Its not just the music either. Ideally you need to prepare yourself by clearing you mind. Stop thinking while you're listening. Just listen to everything at once in as much detail as your brain can take in.
Please don't use an ABX box, at least not to start with. I asked you to keep it simple as possible. Everything you add to the signal chain risks adding distortion, coloration, and or masking. Pots distort, relays can distort if the gold plating wears off the contacts, etc. The less of opamps, class-d amps, etc., and the more of simple discrete amplification and linear power supplies, probably the better.
Once you have noticed what if anything is different between A and B with a simple setup, then add an ABX box if you want. But make sure it doesn't color the sound. As a test of the ABX box, can you still hear the exact same difference as you could without the ABX box circuit? (this is before you start the ABX test, its just to make sure the ABX setup isn't coloring or distorting the sound in some way.) If the difference disappears merely by hooking up the ABX box then you have a problem to troubleshoot. Go back and try again without the ABX box. When you do, just listen. Don't try to think or analyze, as it will bias your perception (more on that below).
Besides checking out and preparing the playback system, you also should prepare yourself. The goal is to empty your mind, relax, maybe take a few slow deep breaths, then just listen deeply and fully, experience the full depth of the sound of the music. That may be the hardest part. In particular, please don't listen for distortion, noise, FR, things you know about. If you do then already you are programming part of your brain to filter out other differences, such as small details of how the musical instruments themselves and the space they were recorded in sound together. If you close your eyes and try to imagine you are at an unamplified acoustic performance in a good room, is the illusion of being there credible? Are the instruments in different locations on the stage? Or does it sound different from real in some way, or ways? If not real sounding, which is less real A or B? What is difference in lay terms? Is A better in some ways, and B better in other ways? Does it sound crunchy, grainy, open, etc. IOW use words that describe perception of sound.
Then come back the next morning when you are fresh. Empty your mind again and repeat. Just experience the music, don't try to confirm what you heard yesterday. Just listen fresh and notice if there is any difference between A and B as though it was the first time you listened. Only after you are done listening, then compare your results to see if your perception is consistent. It probably won't always be consistent, especially if you ever listen when you are tired, or distracted, or thinking about something while listening. But on average you may notice a trend that there is a real difference. If you find there is, then try to find simple lay words to describe the sound, try not to use the technical engineering terms and factors you think may be responsible for any differences in sound. IOW, when you are in listening mode, let go of thinking like an engineer. The analysis of why A and B sound the way they do is something to do later. It may require some thinking. Listening requires not thinking, just listening.
Aside: if you want to know what specific music I usually use and why, please PM. NDA.
Last edited:
Ran out of editing time. Something I forgot to say is about the DUT. Before starting to listen, its important to let the DUT up warm up and stabilize thoroughly. Overnight is usually good, if you can do it. Clocks can take longer. Some things may need to burn in for days or longer before fully stable (e.g. electrolytics in the sound path). When I am doing these kinds of tests, I try to leave the DUT running 24 hours a day, just turn down the volume for silence. If I have to power a DUT down to make a change, the clocks typically remain powered up.
Once again this silly stuff. As I have explained several times already the reason for those 2 files sounding different is simply that the amplitudes are not the same. Human hearing is very sensitive to amplitude changes. In this case the amplitude variation is about +-3dB. If you have trouble hearing such variation then I guess you should not really be doing any listening tests.
Attachments
Last edited:
Some people have reported the frequency modulation to be equally audible. Depends on the person. At least we aren't getting arguments that the files have to be listened to double blind, nor arguments that people claiming to hear a difference must be doubted. Why not such arguments? Most likely its because the skeptics can hear the difference themselves. If they couldn't, then they would demanding double blind from everyone else. It suggests the skeptics believe only they perceive reality as it truly is, its other people who aren't objective. That type of belief has a name, its called "Naive Realism."
The research shows that humans aren't as objective as they tend to believe, and they have a strong tendency to be overconfident. This is for all humans, of course. I know I'm human too and all that it implies. Its the condition we find ourselves in, is all.
https://rb.gy/1d7yy
The research shows that humans aren't as objective as they tend to believe, and they have a strong tendency to be overconfident. This is for all humans, of course. I know I'm human too and all that it implies. Its the condition we find ourselves in, is all.
https://rb.gy/1d7yy
Last edited:
Quotes?
As I said the amplitude difference is +-3dB. That is well above hearing threshold for normal people if not for you. So nobody but you would question this.