120 Db SNR
At last most dac have a exaggerate SNR if you have a normal output 1V max on
output your dac
You have down to 1 uV of noise at 120 dB of SNR where output stages is hard have
best of 90db of SNR .
I think is too hard to hear 1uV of noise at normal reproduction level ....
And an exaggerate dynamic range put the listen at very low levels .....🙄
At last most dac have a exaggerate SNR if you have a normal output 1V max on
output your dac
You have down to 1 uV of noise at 120 dB of SNR where output stages is hard have
best of 90db of SNR .
I think is too hard to hear 1uV of noise at normal reproduction level ....
And an exaggerate dynamic range put the listen at very low levels .....🙄
Last edited:
So they can sell more stuff and make more money? 16/44 appears to be adequate as a music distribution medium.nironiro said:
No filter is perfect. I have no idea what the sine wave looked like as I don't read Stereophile. However, a reconstruction filter would have to be quite poor to deliver a 'bad' 1kHz sine wave.
Because they can, because they want new problems, because they can make more money with it, because ...
Besides, when recording or in production (with lots of processing) you can get a better final result if working at higher bit depth and even sampling rate.
For example if you record a concert the extra headroom 24 bit gives you means you don't have to worry about clipping. You can be sloppy in setting the levels without penalty.
Because you look at an (I) undithered, (II) 16-bit sine at over (III) -90 dB down.
Do you know what the original digital signal for that looks like?
0, 0, 0, 0, 1, 1, 1, 1, 1, 1, ...., 0, 0, 0, 0, 0, 0, 0, -1, -1, -1, -1, -1, -1, ...
=> I guess it is hard to tell what the output should look like if you don't even know what the input looks like.
I don't know if Stereophile ever showed the input signal and explained this.
Regarding (III), that means that for a DAC that outputs 1V full-scale this sine is at 22 uV.
If you had an amp with 30 dB gain and turned it up fully you'd get 0.7 mV vs. close to 32 V full-scale into your speakers. Into 8 ohms that's about 60 nW vs. 125 W.
(All numbers RMS)
Now plug this number into your speaker's sensitivity and compare the resulting number with your room's noise floor. 😉
And also compare this 0.7 mV (RMS) with the DC offset and noise floor of your amplifier.
Last edited:
With 16 bits there is more than enough dynamic range for headroom.
Can you tell me one song wich use 90 dB dynamic range?
Dynamic is offtopic in this thread anyway.
Besides, HiRes and DSD is long time ago out from studios, in our homes, so for consumer use.
If input signal is 1kHz sine wave, than all of us should know how that signal look.
Sorry if someone don't know how it look.
Author of this topic obviously know how sine signal look, that's the reason why he thinking about new design of DAC.
So, Stereophile guys measure signal at some pin of DAC chip? Not at the output of CD player or DAC?
Why measuring signal before the reconstruction/anti-aliasing filter? Why not at the output of the player?
Can you tell me one song wich use 90 dB dynamic range?
Dynamic is offtopic in this thread anyway.
Besides, HiRes and DSD is long time ago out from studios, in our homes, so for consumer use.
If input signal is 1kHz sine wave, than all of us should know how that signal look.
Sorry if someone don't know how it look.
Author of this topic obviously know how sine signal look, that's the reason why he thinking about new design of DAC.
So, Stereophile guys measure signal at some pin of DAC chip? Not at the output of CD player or DAC?
Why measuring signal before the reconstruction/anti-aliasing filter? Why not at the output of the player?
Since I can't edit my previous post anymore.
Here's an undithered 16-bit 1 kHz -90.31 dB sine wave:
Blue dots are the samples at 44.1 kHz.
Green is the reconstruction.
That's what happens when you don't dither near the noise floor and that's why Stereopile also labels similar looking results as "essentially perfect" reconstruction. They're right in this case.
Here's an undithered 16-bit 1 kHz -90.31 dB sine wave:
Blue dots are the samples at 44.1 kHz.
Green is the reconstruction.
That's what happens when you don't dither near the noise floor and that's why Stereopile also labels similar looking results as "essentially perfect" reconstruction. They're right in this case.
But that obviously is not the case here due to the circumstances (see I, II and III above).
He maybe knows how a sine looks like, but that's still a loooong way to understanding digital audio let alone D/A conversion.
No, they measure the "normal" output.
Last edited:
Thanks for the answers, but then PCM DAC can't make decent looking sine wave, at least at 16/44.
You either didn't read the answer or don't understand it.
Either way, your conclusion is absurd.
Btw, you really shouldn't judge by "looks" if you don't understand what you're looking at.
Last edited:
No there is something wrong here. ( as indeed i may be )
As engineers we should ALWAYS be prepared to think "out of the box" , there is a saying "from the mouths of babes" it means that the best clues to better design can come from the strangest of sources.
Although a lot of the comments have been informative i have found most to be simply re-stating whats in a tonne of text books, if had the time and the patience i could refresh on digital theory and also think like everyone else, and if everyone did that what sort of progress we going to get ?? You end up with Japan in the 1980's , lots of very clever people rolling out the same crap year after year after year, stuck in their own form of dream.
So guys , open up, relax 🙂
Back to the subject, OK may not work, ( the idea ) , i think i will have to spend some ( even though i have ZERO time ) coming up with an implementation, and just see what happens on a spice sim with just a few bits. Maybe everyone is right and its the worst idea ever !! ( oh my god, we can never have a dac without a digital filter, BULL**** , i can show you 4 in a matter of minuets ) but maybe not, or maybe it will lead to some interesting results, maybe will trigger some other line of thought . Or maybe a total waste of time, and in 6 months i will be thinking "what the f*** did i do that for" and cringe when i revisit this post 🙂 , i don't know whats going to happen, THAT'S THE POINT , that's progress, that's how electronics as whole moves forward.
PS... This is all one mans opinion obviously
As engineers we should ALWAYS be prepared to think "out of the box" , there is a saying "from the mouths of babes" it means that the best clues to better design can come from the strangest of sources.
Although a lot of the comments have been informative i have found most to be simply re-stating whats in a tonne of text books, if had the time and the patience i could refresh on digital theory and also think like everyone else, and if everyone did that what sort of progress we going to get ?? You end up with Japan in the 1980's , lots of very clever people rolling out the same crap year after year after year, stuck in their own form of dream.
So guys , open up, relax 🙂
Back to the subject, OK may not work, ( the idea ) , i think i will have to spend some ( even though i have ZERO time ) coming up with an implementation, and just see what happens on a spice sim with just a few bits. Maybe everyone is right and its the worst idea ever !! ( oh my god, we can never have a dac without a digital filter, BULL**** , i can show you 4 in a matter of minuets ) but maybe not, or maybe it will lead to some interesting results, maybe will trigger some other line of thought . Or maybe a total waste of time, and in 6 months i will be thinking "what the f*** did i do that for" and cringe when i revisit this post 🙂 , i don't know whats going to happen, THAT'S THE POINT , that's progress, that's how electronics as whole moves forward.
PS... This is all one mans opinion obviously
A few other observations.
1) Are you guys really trying to tell me there is no difference between studio quality digital reproduction ( typically at 32bit and some sample frequency considerably above REDBOOK ) and home redbook players ???
If so why do studios use the "bigger" data ? And dont for one second tell me its due to mastering concerns , cos its not !!! ( that was only an issue in the analog days ) Its so the artists and producers can head MORE of the sound in better quality and hence make better music. Same reason they use superior monitor speakers and good quality headphones.
2) There are countless threads in this forum that show that normal humans can easily distinguish between 1 bit ( actually DS ) dacs and multi-bit types, and they are capable of distinguishing a hell of a lot more ( in blind tests i myself can tell the difference, and i am no musician ), so where does that fit into the digital world ? according to the posts i have read so far , that should be impossible yes ?
1) Are you guys really trying to tell me there is no difference between studio quality digital reproduction ( typically at 32bit and some sample frequency considerably above REDBOOK ) and home redbook players ???
If so why do studios use the "bigger" data ? And dont for one second tell me its due to mastering concerns , cos its not !!! ( that was only an issue in the analog days ) Its so the artists and producers can head MORE of the sound in better quality and hence make better music. Same reason they use superior monitor speakers and good quality headphones.
2) There are countless threads in this forum that show that normal humans can easily distinguish between 1 bit ( actually DS ) dacs and multi-bit types, and they are capable of distinguishing a hell of a lot more ( in blind tests i myself can tell the difference, and i am no musician ), so where does that fit into the digital world ? according to the posts i have read so far , that should be impossible yes ?
1bit DACs are 'DSD' and its been shown mathematically that 1bit conversion can't be correctly dithered. The result of incorrect dithering is noise modulation. No surprise there, noise modulation is fairly easy to discern when you know what you're listening for. The vast majority of S-D DACs nowadays aren't 1bit, quite possibly for this very reason.
In comparing DACs you'll generally be comparing more than just the DAC chip itself, quite likely downstream opamps which may well impart their own sound signature (loss of dynamics being a common one for opamps).
Given that S-D DACs are designed by the numbers its no major surprise that they do impart their own characteristics. As Goodhart's law states 'A measure which becomes a target ceases to be a good measure'.
People often say something like that when their daft idea is rejected by engineers.vecna said:
You would get fewer silly ideas from people who can't be bothered to understand what is already known. You would get more progress, because progress usually comes from "standing on the shoulders of giants" not kicking the ankles of giants.
No, that is not how electronics as a whole moves forward. It may how one particular newbie moves forward, and eventually approaches understanding what others already understand.
More headroom for edits. Musicians mostly don't understand technology? If posts on here are to be believed, neither do some studio engineers.
It is very hard to achieve true 24 bit resolution, noise becomes a big factor, realistically more like 20-22 bits or there about, again this is discussed in greater detail on another thread. I believe 32 bit is for processing as floating point, another thread discussed this in great detail also ...
Here's what you'll find: adding a sine wave does not remove any images nor does it fix a NOS DAC. Doh. 😉
Just look at the spectrum of zero-order hold I pointed you to. What do you gain by adding a sine wave to that spectrum?
The way to achieve high SNR with inadequate/cheap interpolation (like zero- or first-order hold) is to oversample by an insanely high factor first or filtering afterwards.
20 khz
🙂 the 4 khz above 16khz is not free for dac electronics ...
Ringing and others problems are inevitable...
I think the equilibrium between solutions is the best way ....
Since I can't edit my previous post anymore.
Here's an undithered 16-bit 1 kHz -90.31 dB sine wave:
Blue dots are the samples at 44.1 kHz.
Green is the reconstruction.
That's what happens when you don't dither near the noise floor and that's why Stereopile also labels similar looking results as "essentially perfect" reconstruction. They're right in this case.
🙂 the 4 khz above 16khz is not free for dac electronics ...
Ringing and others problems are inevitable...
I think the equilibrium between solutions is the best way ....
Last edited:
Sorry, I don't understand what you're trying to say. 4 kHz above 16 kHz? What solutions?
A big problem with digital audio is that people don't understand it. For example you can't just sample a square wave that theoretically has infinite bandwidth.
Now if you look at the incorrectly sampled waveform it looks great, but what actually happened is that everything above half the sampling rate folded back to below that -- aliasing. Without filtering the result is actually really bad.
Another misunderstanding is that a really steep (and therefore "long") reconstruction filter always significantly changes the input and adds ringing.
Sadly FUD seems to be more interesting than information in audiophile circles.
wow some aggressive people in this forum, and extremely opinionated ( thats a fact ) and i would suggest rather narrow minded, like i said im a lot better off with a book rather than listening to people spouting a book. And i can get insulted in a host of places so really dont need to become the victim here, thank you and goodbye.
If you think this thread's got aggression then you really could get out a little bit more. May I suggest Hydrogen Audio for a spell, then you may return with a modified perspective? 😀
Most of us haven't been offering opinions; we have merely been presenting unwelcome facts. You will often get different opinions on here, but in this thread there have been few opinions - mostly just facts and non-facts.vecna said:
I have not noticed any aggression in this thread. I have noticed a reluctance to learn from others. Have there been any insults?
- Status
- Not open for further replies.