If I understand the schematic correctly, your dither has a probability density function consisting of four Dirac impulses rather than a triangle (only four possible values instead of a continuous distribution). That probably explains the distortion in your later simulation.
Marcel,
What I did was generating two uncorrelated white noise sources, each quantized to +/- 0.5 LSB, Vrms=0.5 LSB
The sum of both noise sources that were used, was regarded to be Tri dither, having a rms value of 0.71 LSB.
If this is not good enough, please tell me what to change.
Hans
P.S. Should I use the un-quantized dither instead ?
What I did was generating two uncorrelated white noise sources, each quantized to +/- 0.5 LSB, Vrms=0.5 LSB
The sum of both noise sources that were used, was regarded to be Tri dither, having a rms value of 0.71 LSB.
If this is not good enough, please tell me what to change.
Hans
P.S. Should I use the un-quantized dither instead ?
Last edited:
You haven't answered my question of whether you selected all those components without even the briefest of audition, or other subjective evaluation.
Wrong with? Personally, I don't think anything is 'wrong' with them. My home theater system is based around a Denon AV receiver. However, for a pure objectivist, the answer must that everything is so right with them that you've necessarily based your music listening system around one. Since mass market AV receivers are, by far, the best objective value in audio amplification. Which was my point. So, again, which AV receiver is your music system based around?
Time, yes. Which is why I'm engaged in DIY.
Money, no. Which is why I'm engaged in DIY.
My listening room, among other factors, ensures that it is not flat, of course. Why?
Hans, our thread is fortunate to have you and Marcel contributing to it.
So, the simulation predicts the Anagram technique reduces DAC harmonics to the same degree as does non-subtractive dither, but additionally restores about 10dB of SNR which is otherwise scarified via non-subtractive dither?
Okay, I think that I see.
However, let me ask my question a bit differently. Does the 10dB difference essentially represent the full contribution of the dither noise? In other words, is essentially all of the added dither noise being subtracted by the process?
Why 0.5 LSB RMS? Normally it should be the sum of two uniformly distributed random signals of 1 LSB peak-peak, 1/2 LSB peak, 1/sqrt(12) LSB RMS each, so the sum is 2 LSB peak-peak, 1 LSB peak, 1/sqrt(6) LSB RMS.
In this specific case you can use half those values if the rounding of the sum of the signal and dither is done as in the appendix of my document. You can also use some integer multiple of the normal dither levels, if you want to randomize the errors due to DAC dynamic non-linearity.
I did answer. Other than speakers and room acoustics (products), I don't use audition or other subjective evaluation for buying those audio electronics.
I use one of those mass market preamp / DAC.
That's odd.
More time will be spent in diy-ing than buying a finished product and given how cheap DAC section is these days (most preamps and digital players have one built-in), a dollar spent on diy DAC will be an extra dollar spent.With which method?
Alright well, then. I’ll accept your word about that.
That's odd.
I own those as well. In fact, both commercial OS, and commercial NOS DACs are easy to locate for purchase. However, as should be obvious by the very existence of this website, there are any number of reasons for why people engage in DIY instead. Some do it out of curiosity of the technology, some for the handcraft of building things, some do it in order simply to obtain exactly the set of product features they want, and some do it because they have learned that the sound they desire is too costly to obtain commercially. All such reasons sacrifice a time resource. So, which is your reason?
Again, why?
Hi Marcel,
The noise that I used was only slightly higher with 0.5 LSB rms instead of the 0.4 LSB rms that you mention [= sqrt(2)*sqrt(1/12) LSB rms].
So I reduced the noise to this level, but was unable to see a relative difference between the two noise spectra, but the absolute level went down by 1.9 dB for both.
Hans
Something I only realized a few minutes ago, after thinking about Hans's -100 dB:
Noise modulation is inevitable with this scheme. When the input signal is constant at 0 or any other integer number, the signal+dither will always be rounded in the same direction for the positive and negative DAC, so the noise will disappear with integer input. That also means there is no point in using triangular dither, you might as well use rectangular. (Mind you, the dither level is critical for rectangular dither.)
Noise modulation is inevitable with this scheme. When the input signal is constant at 0 or any other integer number, the signal+dither will always be rounded in the same direction for the positive and negative DAC, so the noise will disappear with integer input. That also means there is no point in using triangular dither, you might as well use rectangular. (Mind you, the dither level is critical for rectangular dither.)
Last edited:
Sound they desire? You mean self-fi? Well, there is hi-fi which has to do with the likeness to the reference material and then there is self-fi which has to do with one's own desire. Is that what DAC should do? The answer is resounding "no". If sound manipulation to suit one's own taste is the goal, then speakers, room acoustics and equalizer are the tools to do it.
I thought your answer (time & money) was to my question, "what do you think is the best way to achieve hi-fi, spending more time and money on already matured technology like DAC?", which turns out to be a waste of both when it's spent on DAC.
I'm just as curious as you are. Which method do you use?
While there is much that could be challenged about such notion, this isn't the place for a, pure objectivity vs. objectivity guided by some amount of subjectivity, debate.
The real point of your question was, of course, that there is nothing more which can be obtained from DACs, because you judge them to be perfect, mature technology. I ignored that false assertion, and simply answered why I choose to spend mostly time (and a little money) on DAC DIY.
This will be the final time I will ask you to declare exactly why you wish to know. It's only too obvious that you desire to maneuver me in to some silly corner regarding it. So, you either fully come clean, or drop it.
Last edited:
You can make DAC to "color" the sound if that's what one desires but it has so much more limitations in that aspect than equalizer, speakers and room acoustics. However, I'm not trying to stop you from doing what you enjoy doing.
False assertion? There are level matched double blind listening tests reporting audibly indistinguishable between high-end DAC and cheap mass produced receiver DAC. I see that you left out the word "audibly" which I used. It is the primary aspect when dealing with audio gear, no?
You asked me something out of curiosity and I answered. I asked you something out of curiosity as well.
Which method do you use for ensuring that, computer, your own ears or something else?