Actually, adding twice as many DACs improves the signal to noise ratio by 3 dB, not 6. The signal is exactly additive so its power increases by 6 dB, whereas noise is uncorrelated so its power increases by 3 dB for two sources, hence the difference is 6 dB - 3 dB = 3 dB SNR improvement. That applies only if noise is uncorrelated from one DAC to another, as for Johnson noise: power supply induced noise woud be correlated for similar DACs so that assumption goes out the window.
Do please explain how and why op-amps are so bad, by the way. What are their sonic flaws, precisely? Enquiring minds want to know.
Not prejudice, arrogant stupidity! Why are you still here Kurt? GO AWAY
I repeat!
Op-amps sucks.
[/QUOTE]
Kurt everyone knows where you stand on OpAmps, DAC's, dairy cows, transportation, etc. Please go away! You add nothing to this thread other than your echoing voice with the same crap over and over again. We are not deaf and we are tired of your drivel.
Duncan Idaho , with an IP adress originating from planet Arrakis !
Almost right!
Try again, it´s no shame to miss
Kurt everyone knows where you stand on OpAmps, DAC's, dairy cows, transportation, etc. Please go away! You add nothing to this thread other than your echoing voice with the same crap over and over again. We are not deaf and we are tired of your drivel.[/QUOTE]
I´m glad you got the point about the op-amps, but I didn´t know I´ve said anything about dairy cows.
Indeed, but you are evading my point, which is that it has some unique technology NOT available elsewhere.
Well tough. You've set yourself up for this very obvious criticism.
I wonder, do you understand why you are generating so much hostility?
All DAC chips nowadays are differential, hence the 6dB DNR gain.
About my full hearted operational amplifier hate, I only have small hints of scientific reasons for that. Since they do measure much better than any discrete design anyone can come up with anyway.
No discrete design can be expected to cope with 136 dB of THD+N, as the BB OPA 211 does.
And if you are trying out with non feed back design, the whole thing turns completely upside down.
So any good book of electronics would tell you to use op-amps, as you can achieve THD+N more than 1.000 times lower than any non feed back discrete design could ever achieve, and that is the truth.
But! That goes only for THD+N.
The whole idea of op-amps is that they have almost infinite gain, which can be cancelled out by negative feed back. Imagine what happens when you throw away 100-120 dB of signal level in every op-amp stage, because that is what you actually do. Engineers told me, that this is not a problem, but I easily can hear that this is not true. Op-amps has no low level resolution compared to strong non feed back discrete design at all. I think this stems from the linearising feedback.
Op-amps are also designed to cope with lousy supplies, so that bad supplies will not interfere with the performance of the op-amp. To me this is not a good idea, the supplies should be strong and noiseless from the very beginning, so that no correction is needed.
As a result of the above, op-amps are not really strong voltage sources. They are simply not capable of driving cables and inputs of i.e. pre - or power amps. Discrete non feed back designs can be designed to do so, and that is eventually the most important difference.
The sonic difference between the best op-amps I´ve encountered, and superb discrete design is, that the discrete non feed back design, simply is unrestricted and amplifies the signal in the same way from one end to another. Op-amps are almost always depending on the sound of their feedback capacitor, which is an innevitable source of colouration.
But also non feed back op-amps sound mostly like normal op-amps, although NFB opamps can be found with obscene data i.e. AD844.
We tried the AD 844 out, and it may be the best op-amp available at all, but the sound of it is still small, squeezed, thin and uninvolving.
Maybe it has some quality thrughout the frequency range, but unfortunately music does not reward that kind of stuff.
Kurt,
I dont need it kept a secret I asked to see the schematic. I have a high interest in seeing what you have come up with just like everyone else. BTW, Im still wating to see it
You can ad 6 dB of DNR each time you put in another chip.
/Its really only 3dB
Last edited:
Kurt,
I dont need it kept a secret I asked to see the schematic. I have a high interest in seeing what you have come up with just like everyone else. BTW, Im still wating to see it
Is this it on page 192? http://www.diyaudio.com/forums/showthread.php?t=138230&page=20
And the thread carries on , the pages increase yet still no closer to an actual sound comparison
I know someone who upgraded from a CS4396 based DAC to a Buffalo (24 bit) and noticed a substantial improvement. The old Crystal DAC was of his own design and used synchronous reclocking, low noise supplies etc. I mention this as he was able to use the exact same analog stage for both ....
I know someone who upgraded from a CS4396 based DAC to a Buffalo (24 bit) and noticed a substantial improvement. The old Crystal DAC was of his own design and used synchronous reclocking, low noise supplies etc. I mention this as he was able to use the exact same analog stage for both ....
I've messed about with CS4397/8 too, diy and and non diy, for me both Buffalo's was a large enough step up to convince me which is the better .
I think this hyped up one on here is supposed to add some magical circuit changing the average out of the CS4398 to something special
Its really only 3dB
Sorry!
What I thought about was giving it 8 DACs as the Sabre also has.
Then you can ad 6 dB extra DNR to the CS4398.
That is not possible, as the Buffalo is an I out DAC and the CS4396 is an U out type. There has to be some kind of I/V conversion for the Sabre.I know someone who upgraded from a CS4396 based DAC to a Buffalo (24 bit) and noticed a substantial improvement. The old Crystal DAC was of his own design and used synchronous reclocking, low noise supplies etc. I mention this as he was able to use the exact same analog stage for both ....
I've messed about with CS4397/8 too, diy and and non diy, for me both Buffalo's was a large enough step up to convince me which is the better .
I think this hyped up one on here is supposed to add some magical circuit changing the average out of the CS4398 to something special
There has never been arguments about the qualities of the Sabre DAC, since We´ve never tried it out, and do not know it. But what we did was trying out almost endles numbers of different op-amp configurations and types. They do have some advantages, as it always works at first shot, they are cheap, easy to get and there is a lot to chose from. Just audio types are pretty rare.
The disadvantages is deffinately sound quality, at a point it was almost like if somebody describes the sound of an amplifier, I could guess which op-amps was used. So it will be with the Buffalo, the sound you like is by far mostly the sound the opamps. In fact the op-amps most often used in CD players and DACs, are the components at which you can addres the whole idea about digital sound.
Why is it that leo has offered to build your DAC so many times and continuously asks for the price of the PCB and what not and you just skip over it as if he didn't post at all? If you really want to prove your DAC is better send him the PCB let him build it and make a comparison. Heck maybe he'd let some other trusted members of the forum try it out as well and we can get a consensus. It seems you are all talk with absolutely no action. You said you posted schematics but you really didn't you posted partial schematics and then they weren't even correct according to your thread.
The Sabre DAC output looks like a voltage source with 780 Ohm output impedance. It may be used in both "voltage mode" where you driving it into a high impedance, or in "current mode" (like the Buffalo implementation) by driving it into a low impedance. A large difference is in that it doesn't have a switched cap filter in the DAC section like some others do. Not that there is anything wrong with those, I just did it differently. As a by product of the way its done, as many many people have pointed out, rightfully, the DAC is a bit more sensitive to power supplies. At first I thought of this as a weakness, but after a while I have come to realize its actually the right thing to do. The tradeoff would have been to give up some performance for PSRR, or have the ability to get good performance, but that will require a good power supply technique. This is where manufacturer's can some in and add their own expertise for product differentiation. If every implementation sounded the same independent of the power supply, what fun would that be for all the tweakers and modders? I know I have fun with it myself.
KvK, Hurtig, you really owe it to yourself to atleast just listen to Russ and Brian's Buffalo32. If you dont like it after, rip on it then. At some point I would like to listen to your "no global feedback" design. I am very curious. Having a design that is linearized along the way does have the advantage that the ringing will be minimal. I think your design deserves a listen, but afterall, so does the Buffalo. Since I have never heard a CS4396/CS4398 based design, maybe yours will be the first for me. How much can I buy one for, and when will I be able to get one?
Dustin
KvK, Hurtig, you really owe it to yourself to atleast just listen to Russ and Brian's Buffalo32. If you dont like it after, rip on it then. At some point I would like to listen to your "no global feedback" design. I am very curious. Having a design that is linearized along the way does have the advantage that the ringing will be minimal. I think your design deserves a listen, but afterall, so does the Buffalo. Since I have never heard a CS4396/CS4398 based design, maybe yours will be the first for me. How much can I buy one for, and when will I be able to get one?
Dustin
- Status
- Not open for further replies.