Pavel,
So the fact that fewer samples strongly increase the noisefloor is "nothing"?
Can you think what the noisefloor would be like with 256 Samples? What would the noisefloor be with an analogue system that has a 130dB FFT noisefloor for 256 samples in the FFT?
Did you ever try the same set of measurements on a DAC that does not work on the principle of adding "Fuzzy" Distortion?
More to the point, did you ever listen?
BTW, I am familar with that Cirrus Logic Chip and as it so happens I also know the lead designer (long story, we bumped into each other at the queue at the "foreigner registration" of a Police Station in Beijing) and we actually even discussed his DAC designs in some detail. We even discussed the design rationales, driving factors and so on all the way to the kind of HiFi System he listens to at home (all strictly Sub Rosa), over some truely delicious Shaanxi style roast duck, cat ears and very quaffable Beijing Beer at the YinYang restaurant, previously frequented by Bush senior when he was US Ambassador in Beijing, riding his Bicycle there. Six degrees of separation and all.
Anyway, all this was really off topic.
Well, if you like this DAC and the way it sounds, who am I to tell you should not?
I see little point wasting my time convincing people of things they refuse to know and understand, there is no gain in it for any side, either materially, educationally or spiritually.
Ciao T
So the fact that fewer samples strongly increase the noisefloor is "nothing"?
Can you think what the noisefloor would be like with 256 Samples? What would the noisefloor be with an analogue system that has a 130dB FFT noisefloor for 256 samples in the FFT?
Did you ever try the same set of measurements on a DAC that does not work on the principle of adding "Fuzzy" Distortion?
More to the point, did you ever listen?
BTW, I am familar with that Cirrus Logic Chip and as it so happens I also know the lead designer (long story, we bumped into each other at the queue at the "foreigner registration" of a Police Station in Beijing) and we actually even discussed his DAC designs in some detail. We even discussed the design rationales, driving factors and so on all the way to the kind of HiFi System he listens to at home (all strictly Sub Rosa), over some truely delicious Shaanxi style roast duck, cat ears and very quaffable Beijing Beer at the YinYang restaurant, previously frequented by Bush senior when he was US Ambassador in Beijing, riding his Bicycle there. Six degrees of separation and all.
Anyway, all this was really off topic.
Well, if you like this DAC and the way it sounds, who am I to tell you should not?
I see little point wasting my time convincing people of things they refuse to know and understand, there is no gain in it for any side, either materially, educationally or spiritually.
Ciao T
No I haven't seen them myself but I have no reason to believe Thorsten was lying when he mentioned the measurement he made. I'll see if I can find the link to the thread where he said what his technique was.
<edit> The discussion I was recalling is here : http://www.diyaudio.com/forums/digital-line-level/160928-open-source-dac-r-d-project-6.html#post2102675
<edit> The discussion I was recalling is here : http://www.diyaudio.com/forums/digital-line-level/160928-open-source-dac-r-d-project-6.html#post2102675
Last edited:
Pavel,
You have read the CS4398 Datasheet, right?
The modulator is 25 Level at 128 Times oversampling, so giving 3200 possible distinct levels. This is a dynamic range of 70dB.
As the Part is advertised as 120dB DNR we need to apply 50dB worth of noiseshaping to get this. Now how many samples of this 50 dB noiseshaping are needed to get the FFT Noisefloor for 120dB?
In other words, how many samples do we need to average to get 2 ^ 20 levels (which is what is basically 120dB DNR) and reverse, what is the maximal accuracy with which a single sample can be represented?
One of the methods that suits itself to revealing "hidden" behaviours of circuits is noiseloading, which is REALLY an old hat, I mean completely antique (it was antique when I started out in electronics in the 80's)and AP Equipment CAN (with substantial effort) be programmed to do something approaching this.
Anyway, all this stuff around dither and noiseshaping/modulation is really basic, the Lipshitz/Vanderkoy article on DSD is a good summary of the issues, though I find myself disagreeing with some of their conclusions (including, but not limited to the one that asserts that because they measure the same with a lot of averaging noiseshaped low bit converters perform identical to non-noiseshaped "Multibit" converters.
Ciao T
You have read the CS4398 Datasheet, right?
The modulator is 25 Level at 128 Times oversampling, so giving 3200 possible distinct levels. This is a dynamic range of 70dB.
As the Part is advertised as 120dB DNR we need to apply 50dB worth of noiseshaping to get this. Now how many samples of this 50 dB noiseshaping are needed to get the FFT Noisefloor for 120dB?
In other words, how many samples do we need to average to get 2 ^ 20 levels (which is what is basically 120dB DNR) and reverse, what is the maximal accuracy with which a single sample can be represented?
One of the methods that suits itself to revealing "hidden" behaviours of circuits is noiseloading, which is REALLY an old hat, I mean completely antique (it was antique when I started out in electronics in the 80's)and AP Equipment CAN (with substantial effort) be programmed to do something approaching this.
Anyway, all this stuff around dither and noiseshaping/modulation is really basic, the Lipshitz/Vanderkoy article on DSD is a good summary of the issues, though I find myself disagreeing with some of their conclusions (including, but not limited to the one that asserts that because they measure the same with a lot of averaging noiseshaped low bit converters perform identical to non-noiseshaped "Multibit" converters.
Ciao T
Hi,
Well, DSD was "invented" when the output of a Delta Sigma AD designated for CD release was stored directly and played back directly, without following conversion to PCM.
However, the classic DSD Modulator has less information content (information theory) than can be encoded in classic 4 * Oversampled 16 Bit/44.1KHz Audio.
It is interesting to compare prior good CD releases with the CD-Layers of many SACD's and the SACD, using high grade equipment.
Many years ago, when I modified Shanling CD-Players I used to demo them (PDM200 & PCM1704 * 2) against Sony SACD Players modified similar to Allen Wright's mods. The differences in Sonics where quite instructive.
Let me say these demo's helped tremendously in selling my modded CD-Players - they where fully level matched, but not blind, FWIW, as a rule, all people who auditioned where severely predisposed towards SACD...
Ciao T
Well, DSD was "invented" when the output of a Delta Sigma AD designated for CD release was stored directly and played back directly, without following conversion to PCM.
However, the classic DSD Modulator has less information content (information theory) than can be encoded in classic 4 * Oversampled 16 Bit/44.1KHz Audio.
It is interesting to compare prior good CD releases with the CD-Layers of many SACD's and the SACD, using high grade equipment.
Many years ago, when I modified Shanling CD-Players I used to demo them (PDM200 & PCM1704 * 2) against Sony SACD Players modified similar to Allen Wright's mods. The differences in Sonics where quite instructive.
Let me say these demo's helped tremendously in selling my modded CD-Players - they where fully level matched, but not blind, FWIW, as a rule, all people who auditioned where severely predisposed towards SACD...
Ciao T
How can oversampling put more information on a CD? It just interpolates between points doesn't it - "Bogus Bits" as some say.
I have a copy of one of those experiments and I disagree. The frequency response anomolies between the two are so gross that the simple "there is no difference but and A/D - D/A in the path", can't possibly be true. I would be glad to investigate this someday.
I have Teflon caps, in my experience "gimmic" caps don't really have low Q but that was RF. I might be able to find a 50pF vacuum cap.
Hi,
In my opinion, it cannot.
I merely pointed out that correctly implemented 16/44.1 with the correct digital filter with oversampling (in order to leave for now the Oversampling/No-Oversampling controversy firmly locked in it's little jar) can carry more information than the original "DSD 1X" modulator used for SACD can produce.
Note all this was strictly from an information theory viewpoint and contains many caveats when it comes to applicability in practice...
Ciao T
In my opinion, it cannot.
I merely pointed out that correctly implemented 16/44.1 with the correct digital filter with oversampling (in order to leave for now the Oversampling/No-Oversampling controversy firmly locked in it's little jar) can carry more information than the original "DSD 1X" modulator used for SACD can produce.
Note all this was strictly from an information theory viewpoint and contains many caveats when it comes to applicability in practice...
Ciao T
Funny, I had to google DSD. I have not searched but I find it hard to believe that the different digital systems have not been compared on first principles of information theory at least somewhere.
And as I said before it does bother me just like it bothers Thorsten that the massive averaging to make nice pictures (maybe) does not address the instantaneous performance. Luckily I can walk into Bob Adam's office and have him try to explain.
Hi,
Speak for yourself Martian, I'm from Europa!
Un SanG real? Really? THE royal blood? We all have it...
And on Europa we have long (and I mean long) known that there is no San Graal of Audio. The best there is a convincing illusion, in the same kind and manner as that practiced by those who call themselves Magicians (with a c, not a k, the difference is profound).
This is very true indeed. However, from where I stand it is not a bad thing. All my music that does not exist on those 12" Diameter plastic disks that where used in the analogue age for storage of acoustical information has eXistenZ only in cyberspace, as have most of my books and magazines, movies and TV Series,
Yet it means even my 19 Month old daughter can play it. I doubt she gets "genre" yet, but she can identify what she likes by the coverart on the screen and point at it, with the screen being "touch" it means "play"...
I could not agree more with you. I find much of these debates totally pointless, as they lead nowhere. However, as long as there are some who cannot let viewpoints and ideas they have problems of moral affordability with just stand and go elsewhere and do their own thing, but instead argue everything to death and beyond, regardless of evidence, regardless of truth this will sadly only stop if one of the sides in the argument just walks away and leaves the field to the "other side".
Well, the death of many studio's is sad indeed. However, seeing how much music has become overproduced, I think it is a good thing that Young musicians take thing into their own hands. And many of them (even some of those who run Garage Band on a Mac) do care about quality.
It has lead to a great deal of readily available and fairly inexpensive gear of high quality. It is thanks to that several commercial operations I had hands in now use prosumer sound cards with appropriate breakout boxes and software for quality testing, instead of buying several Audio Precision units...
This should not come as a surprise. Since the 80's we have seen record companies turning more and more to overproduced, valueless "music" which if were food would be so low in Quality and Nutrition that even the likes of MacDonalds and KFC have to high standards of corporate ethics to sell...
It is not as such the recording that is the problem, it is the content. If you hire the greatest french chef and only give him garbage (real garbage, not the stuff Supermarkets throw because it is past sell by but otherwise good) to cook with, the food will taste like garbage, in fact it will be garbage...
Ciao T
Speak for yourself Martian, I'm from Europa!
Un SanG real? Really? THE royal blood? We all have it...
And on Europa we have long (and I mean long) known that there is no San Graal of Audio. The best there is a convincing illusion, in the same kind and manner as that practiced by those who call themselves Magicians (with a c, not a k, the difference is profound).
This is very true indeed. However, from where I stand it is not a bad thing. All my music that does not exist on those 12" Diameter plastic disks that where used in the analogue age for storage of acoustical information has eXistenZ only in cyberspace, as have most of my books and magazines, movies and TV Series,
Yet it means even my 19 Month old daughter can play it. I doubt she gets "genre" yet, but she can identify what she likes by the coverart on the screen and point at it, with the screen being "touch" it means "play"...
I could not agree more with you. I find much of these debates totally pointless, as they lead nowhere. However, as long as there are some who cannot let viewpoints and ideas they have problems of moral affordability with just stand and go elsewhere and do their own thing, but instead argue everything to death and beyond, regardless of evidence, regardless of truth this will sadly only stop if one of the sides in the argument just walks away and leaves the field to the "other side".
Well, the death of many studio's is sad indeed. However, seeing how much music has become overproduced, I think it is a good thing that Young musicians take thing into their own hands. And many of them (even some of those who run Garage Band on a Mac) do care about quality.
It has lead to a great deal of readily available and fairly inexpensive gear of high quality. It is thanks to that several commercial operations I had hands in now use prosumer sound cards with appropriate breakout boxes and software for quality testing, instead of buying several Audio Precision units...
This should not come as a surprise. Since the 80's we have seen record companies turning more and more to overproduced, valueless "music" which if were food would be so low in Quality and Nutrition that even the likes of MacDonalds and KFC have to high standards of corporate ethics to sell...
It is not as such the recording that is the problem, it is the content. If you hire the greatest french chef and only give him garbage (real garbage, not the stuff Supermarkets throw because it is past sell by but otherwise good) to cook with, the food will taste like garbage, in fact it will be garbage...
Ciao T
Hi,
I appreciate your decision and I do in part share your reasons, but I would ask you to reconsider.
A little later, when I have more time, maybe tomorrow, I would like to share a simple short story that will explain WHY I am still here as part of this forum, though I have sure been given enough reasons to throw my toys out of the pram and stamp my little feet and cry like a baby and then make like a tree and leaf...
However, every now and then something happens, something small, nothing earthshaking, but it makes a difference.
Whichever way you go, may Fortuna smile on you, do leave us a link, if you go.
Ciao T
I appreciate your decision and I do in part share your reasons, but I would ask you to reconsider.
A little later, when I have more time, maybe tomorrow, I would like to share a simple short story that will explain WHY I am still here as part of this forum, though I have sure been given enough reasons to throw my toys out of the pram and stamp my little feet and cry like a baby and then make like a tree and leaf...
However, every now and then something happens, something small, nothing earthshaking, but it makes a difference.
Whichever way you go, may Fortuna smile on you, do leave us a link, if you go.
Ciao T
- Status
- Not open for further replies.