Superior as measured by an FFT analyzer? If so, maybe that would be something like superior in the sense of, "looking for the car keys under the streetlight." The illumination is unrivaled, no disagreement about the great sensitivity of FFT.
Please don't exaggerate. Never said anything about magic.
If tails are masked, could be because something is involved which does not show up well on an FFT. If so, we needn't resort to magic as an explanation. Some people have complained that I have talked enough about things like noise modulation and or signal-correlated noise, so I won't go there for now. Could be some other things involved as well. Nobody said there is one and only one single cause of remaining imperfections. Most of the easy problems in dac design are well understood and already fixed. Now are left with the ones not so well 'illuminated by the steady-state FFT streetlight.'
Tell you what, rather than argue now I would like to wait until the next version AK4499 comes out. The two chip solution should help answer some questions regarding IC substrate coupled noise. Then maybe PCM will become more competitive with DSD, at least when human perception of SQ is of most interest. Don't know what to expect, myself. In any case it should be interesting to see if there anything to learn from it, or else if there isn't.
Last edited:
D
Deleted member 537459
AD1862 and TDA1541A aren’t discrete DAC. Voltage output discrete DACs don’t need I/V conversion, the -85dB THD converter-buffer was used for the current output DACs and not for the discrete DAC.
D
Deleted member 537459
You cannot get -120dB with the TDA, it’s a 16 bit DAC. But ok, -85dB is worse than the theoretical -96dB of the TDA. Simulated THD is mostly 2nd harmonic, 3rd harmonic is -92dB, no higher order harmonics. Maybe is the price to pay applying zero feedback and inductive power supply without regulation.
You may be right there Marcel in that aspect of my complaint. Conceptually I have been thinking that louder peaks are going to be more audible, but I suppose as distortion is a relative measurement (relative to signal/fundamental), rather than absolute, the distortion character and amount relative to signal are the same?
regardless, for me ending up with a good 12db more than necessary is not OK and I dont understand why you would want to do that, when it is simple to do otherwise, but hey as I said, enjoy.
regardless, for me ending up with a good 12db more than necessary is not OK and I dont understand why you would want to do that, when it is simple to do otherwise, but hey as I said, enjoy.
Last edited:
Are we discussing total harmonic distortion or SINAD? There is no theoretical limit to how low the distortion of a 16 bit DAC can go with proper dithering, although there are practical limits due to DAC imperfections. SINAD can't go above approximately 98 dB unweighted without dithering, 93 dB unweighted with triangular dithering.
Last edited:
I understand that distortion, particularly even order harmonic is one of the more benign sources of error. As mentioned for me personally the worst part if it was to be used in my system is it produces distortion above the level of the driver it would be tasked to correct. FIR would be attempting to correct error with something that is less accurate than that used to recreate the waveform. I'd be interested to see what measurements of such a system would look like and what the residual would look like.
I understand this is not the usual use case for such a dac, nor the intent.
I understand this is not the usual use case for such a dac, nor the intent.
WRT zero feedback and low distortion, it depends on the design and how well you understand the mechanisms of non linearity.
I can achieve well below 0.001% in my mic pre designs and that is with 60dB of gain. It's also virtually flat to 200kHz - no typo there.
I'm not saying using FB in a high gain mic pre is bad but having an open loop design allows other sonic options such as adding various types of distortion that are flat vs freq.
Having said the above there are certain types of distortion that open loop designs can fall prey to such as junction thermal modulation which IME are detrimental to sound.
TCD
There is a lot of anecdotal stuff about people liking short impulse responses even though they are completely in the supposedly ultrasonic range, but I don't know that many serious articles about it.
There is a Japanese article from the 1990's in which the authors claim that Japanese gamelan players listening to Balinese gamelan recordings could subliminally hear the difference between playback with or without a supertweeter reproducing sound above 26 kHz. The authors used electroencephalograms to establish that. Mind you, it isn't clear from the description in the article (not to me anyway) whether it was a double-blind test or only single-blind.
Tsutomu Oohashi, Emi Nishina, Norie Kawai, Yoshitaka Fuwamoto and Hiroshi Imai, "High-frequency sound above the audible range affects brain electric activity and sound perception", Audio Engineering Society preprint 3207, presented at the 91st Convention, October 1991
Then there is the Acoustic Renaissance for Audio DVD-Audio proposal where it is claimed that a small percentage of young people can hear well above 20 kHz. I can't find it on the internet anymore, maybe I still have a hard copy somewhere. I don't recall if they just mentioned it in passing or had a reference to a serious listening test.
Finally there is the open access paper about the MQA system, https://www.aes.org/e-lib/browse.cfm?elib=17501 It has a lot of suppositions in it, and two literature references to articles that supposedly show that the envelope of ultrasonic signals affects the perception of sound, [86] and [90]. I have no access to either of them, but from the abstract of one of them I gather it is about barn owls rather than humans, so there must be some extrapolation involved.
That DVD-Audio proposal can be found here: https://www.yumpu.com/en/document/v...gh-quality-audio-application-of-high-meridian (and in my attic). The authors state that: "The main body of psychoacoustics literature shows that the audible frequency range of steady tones for young adults extends from below 10Hz to 18kHz mean, with 1 % able to hear 25kHz tones." There is no reference to that main body of psychoacoustics literature.
Good concert piano recordings seem to be a particularity good tool for discriminating DAC timbral capability Some DACs make pianos sound like a digital piano. R2R DACs often get piano timbre right.
With a good DAC its possibly to pick the piano maker eg the classic Steinway signature vs a Bosendorfer
Please can I ask if you might consider starting a new thread to just discuss the DAC-Lite listening experience and the DAC-Lite build? It would be a relief to have a thread free of theoretical antagonism.
Seems like that's probably what the OP had in mind for this thread. Why would a new thread be different?
Besides, this thread is actually pretty polite for one that's getting as much attention as it is.
@MarcelvdG,
Oohashi et al. was the first publication trying to find corroboration for the hypothesis that ultrasonic content is perceptible by humans (and has positive effects as well). Quite a few papers were published subsequently, mostly combining sensory experiments (including controlled double-blind tests and body measurements (like EEG, PET-Scans, or even fMRI, iirc) to explore the reactions of the listeners.
Results were inconsistent, sometimes methodological flaws or weaknesses were present; Reiss calculated in his meta-analysis an overall probability for discrimination of 52,3% (CI(95) 50,6 - 54.0), but that number is questionable due to the heterogeneity of the included papers.
Oohashi et al. was the first publication trying to find corroboration for the hypothesis that ultrasonic content is perceptible by humans (and has positive effects as well). Quite a few papers were published subsequently, mostly combining sensory experiments (including controlled double-blind tests and body measurements (like EEG, PET-Scans, or even fMRI, iirc) to explore the reactions of the listeners.
Results were inconsistent, sometimes methodological flaws or weaknesses were present; Reiss calculated in his meta-analysis an overall probability for discrimination of 52,3% (CI(95) 50,6 - 54.0), but that number is questionable due to the heterogeneity of the included papers.
It was your assertion and now you should be able to bring up the evidence supporting your assertion.
At the risk of contradicting you, my view on "certain types of listening tests" is just based on science and, as you know
, I've provided the relevant literature backing my view.You should be able to do the same.
You seem to believe, that a special definition for the "audio field" must exist, but that isn't true. As usual, the real/true parameter is the one, experimenters are trying to find out by doing good quality experiments.
- Status
- Not open for further replies.