Schiit Fire and Save Matches! Bifrost Multibit is Here.
- Thread starter Jason Stoddard
- Start date
Where an explanation starts, and how long it is depends on your starting point knowledge of electricity (as relates to electronics), binary representation of numbers and of building blocks such as comparators and summers.
Not meaning to dodge the question but to give an explanation in a language you understand one first needs to know what that language is.
I was involved in digital electronics in an earlier life and to me simplistically multibit is logical, DS is smoke and mirrors
Beside the multibit DS divide I think a very important consideration is the MM mega-whatever filter, but only MM knows about that.
EDIT: MM (Baldyr) just posted
http://www.head-fi.org/t/701900/schiit-happened-the-story-of-the-worlds-most-improbable-start-up/9630#post_12232121
disastermouse
Headphoneus Supremus
Where an explanation starts, and how long it is depends on your starting point knowledge of electricity (as relates to electronics), binary representation of numbers and of building blocks such as comparators and summers.
Not meaning to dodge the question but to give in explanation in a language you understand one first needs to know what that language is.
I was involved in digital electronics in an earlier life and to me simplistically multibit is logical, DS is smoke and mirrors.
Beside the multibit DS divide I think a very important consideration is the MM mega-whatever filter, but only MM knows about that.
And that's why I don't get into arguments about the technical details. It's way out of my league!
bgentry
100+ Head-Fier
- Joined
- Nov 12, 2014
- Posts
- 331
- Likes
- 127
I'm going to take a stab at this. This will be simplified and probably get some details wrong. But the basic ideas should be correct.
In a multi-bit D/A converter, each bit, in each input word is directly converted to an output voltage. So, with a 16 bit word input (like from CDs or similar files) each of the 16 bits is used directly. 16 bits gives us 65536 values and these are converted directly to output Voltage. So the middle value of a 16 bit word would be approximately 32,000 and the top value approximately 65,000.
Inside the D/A converter there is normally a resistor the corresponds to the value of each bit in the word. Which means that the biggest resistor is approximately 32,000 times the value of the smallest resistor. Each resistor represents one of the bits in the word, so each resistor is half the value of it's bigger neighbor like 1, 2, 4, 8, 16, 32, 64 .... 32678 .
As I understand it, getting these resistor values to be precise is very difficult. This (and probably many other reasons) led to the development of "1 bit" D/A converters. A 1 bit D/A only has one resistor. The smallest one with the value of 1. This is what Delta Sigma is: Using a single element, or a small number of elements, to add or subtract from the signal as it goes along. Think of Delta Sigma as a piece of chalk moving across a chalk board that's going to sketch a waveform. At each increment across the board, the chalk can move across a bit and UP one value, or across a bit and DOWN one value. To go all the way to the top, it gets to UP, UP, UP, UP until it reaches the desired value at the top. This all happens much, much faster than the word rate of a multi-bit converter. Because it's moving in small increments it has to move much faster.
So how in the world does a 16 bit word get turned into a bunch of instructions that say "up" or "down" thousands of times? Through math that converts one to another. Math that is NOT REVERSIBLE. Which means that when you convert your 16 bit words into Delta/Sigma, you lose something. Because you can't reverse the process and go back to your 16 bit words. Something was lost in translation from MB to D/S.
I'm told that modern D/S DACs use more than 1 bit internally. 4 or 5 bits perhaps. Again, this transformation from mutli-bit to D/S is not reversible.
In summary: In a multi-bit DAC, the input words (16 bits for example) are fed directly to the electronics of the DAC and are used to derive the output voltage and waveform. In a D/S DAC, the words are "sliced and diced" into another form to make a signal of "add one" or "subtract one", and this signal is used to construct the output waveform. They are not equivalent.
There's another huge consideration here though: The Schiit megacombo-burrito filter. It's responsible for another conversion that's important in an audio DAC, and is separate from the above discussion. There *are* other multi-bit DACs on the market. But none of them have the Schiit filter.
Brian.
rtaylor76
100+ Head-Fier
This page I found explained it the best for me without going into digital math. Here is a copy/paste shortened version:
"...a R2R [multibit] converter is a static device, in that it is able to generate a clean voltage (with 16 bits resolution there are 65536 different voltages) and hold that voltage until another number is converted, or if numbers repeat - for an infinitely long time, while maintaining an extremely low noise level.
In order to achieve a certain level of performance, the resistors (R's) inside the converter chip must be precisely trimmed, which can make a device very expensive to manufacture, especially when true 24-bit resolution is required
In order to avoid the required precision in the manufacture of R2R converters (and save manufacturing costs),...a single switch replaced the precision resistors of the R2R DAC, and it was believed (and shown to a certain extent) that if only the switch was toggled fast enough, one could also achieve many different voltages."
It goes on to explain multiple voltages produced, the limitations of this high-switching, the added noise, and of course because they are high switched, they are more susceptible to jitter. Essentially, the D-S chips interpret the incoming signal to a 1-bit output. 1-bit only has two outcomes (0 or 1), and these two bits are from a continuous signal, higher or lower than the last bit (the essence of DSD). And with a simple switch, they can build all of that in one integrated circuit. It's why we have all the nice small devices we have.
Also note that Schiit uses 2 DAC chips per channel in Yggy and Gumby, and only one total for Bimby. I don't know why they need 2 per channel, but I do know that you better phase correlation with one chip per channel.
I am still waiting on the Moby - Modi sized multibit. I doubt they can get the megaburrito combo in such a small package along with USB, power, optical & coax input. However, they were able to build a very accurate RIAA and variable impedance and variable gain phono preamp in that small box that sounds spectacular.
Does anyone know if the architecture of the system would allow Schiit to make the upsampling filter defeatable so that it can run in pure NOS mode?
aamer23
100+ Head-Fier
- Joined
- Jul 7, 2014
- Posts
- 296
- Likes
- 36
To have my cake and eat it too, I guess. Being new to multibit and NOS DA's (unless my older CD players from the late 90's contained them), I would really appreciate a product that was multibit with the ability to defeat upsampling (even if it is preserving the original samples). Maybe NOS is incompatible with the architecture they've built around the DA chip.
Just curious. I saw some employee from Schiit posting on here, so I'm hoping someone might know.
Just curious. I saw some employee from Schiit posting on here, so I'm hoping someone might know.
aamer23
100+ Head-Fier
- Joined
- Jul 7, 2014
- Posts
- 296
- Likes
- 36
I was just wondering what area's of the bimby could still be improved and realistically how much more can it be improved
disastermouse
Headphoneus Supremus
That was helpful! Are you also able to explain what the megaburrito filter does?
aamer23
100+ Head-Fier
- Joined
- Jul 7, 2014
- Posts
- 296
- Likes
- 36
from what I understand yggy has a better / newer analog output stage was wondering if that would improve bimby
Baldr
Sponsor: Schiit Audio
- Joined
- May 14, 2011
- Posts
- 943
- Likes
- 7,517
Baldr
Sponsor: Schiit Audio
- Joined
- May 14, 2011
- Posts
- 943
- Likes
- 7,517
Pure NOS is not pure. It requires a very complex analog reconstruction filter to get rid of all aliasing over half of the sampling rate. These filters have horrendous time domain anomalies. Other than historical interest, there is little value in them.
Users who are viewing this thread
