Multibit (16bit) Versus BitStream (1bit)... Which is better to you?

Jul 10, 2004 at 6:37 PM Thread Starter Post #1 of 29

Duncan

Headphoneus Supremus
Joined
Jun 26, 2001
Posts
13,510
Likes
1,848
Now, this is probably going to end up as quite a heated debate... flames aren't really necessary as its all a matter of opinion... but just curious to know which you think is better, and why?

I, myself - am starting to think that Multibit is the way to go...
 
Jul 10, 2004 at 6:42 PM Post #2 of 29
DSD is boss to me!
eggosmile.gif
 
Jul 10, 2004 at 7:05 PM Post #3 of 29
Quote:

Originally Posted by richx
DSD is boss to me!
eggosmile.gif



lol... no cheating
wink.gif


Out of just 16bit, or 1bit... which do you think is best
smily_headphones1.gif


(16bit, I myself am now convinced, adds more life, and more zest to recordings...)
 
Jul 10, 2004 at 8:01 PM Post #5 of 29
You are asking a question that is all but unanswerable. You are assuming:

1. There are two players that besides the use of multi-bit DACs vs. bitstream DACs are completely identical, ready for quick and easy A/B...

2. That there are people here who have heard both extensively

IMO, depending on build quality and vintage, you could easily find examples of each that out=perform the other.
 
Jul 10, 2004 at 8:15 PM Post #6 of 29
For the data format on the storage medium, I prefer multibit PCM - but for the DAC itself, I'm ok with both multi- and singlebitters. The actual implementation of the DAC and output stage seems more important than the basic priciple, anyway...

Greetings from Hannover!

Manfred / lini
 
Jul 10, 2004 at 8:22 PM Post #7 of 29
Quote:

Originally Posted by lini
For the data format on the storage medium, I prefer multibit PCM - but for the DAC itself, I'm ok with both multi- and singlebitters. The actual implementation of the DAC and output stage seems more important than the basic principle, anyway...


I'll take it beyond that, my friend. It's the sound...it's the earbones. Bits and DACs mean very little to me.

What I watch is my toe...if it's tapping, then the music is happening.

smily_headphones1.gif
 
Jul 10, 2004 at 9:39 PM Post #8 of 29
I strongly believe that multibit is a vastly superior concept from a mathematical standpoint. As to which sounds better, it's all about how it's implemented.
 
Jul 10, 2004 at 9:56 PM Post #9 of 29
Quote:

Originally Posted by jefemeister
I strongly believe that multibit is a vastly superior concept from a mathematical standpoint. As to which sounds better, it's all about how it's implemented.


Without getting overly technical, researching this, is exactly why i've ended up getting a multibit player...

according to something I read (if I can find the link, i'll post it) 1bit x 256 sampling only provides enough data for 12bit representation... where did the other 4 bits go?
eek.gif
 
Jul 10, 2004 at 11:44 PM Post #10 of 29
what Jeff said .

But I will take it a step beyond.

My personal dacs are all except for one zero oversampling multibit dacs with a current output to a passive IV stage and 1:1 transformer coupling.

The other is a 4X oversampling dac but the output section is as above.

I have and have had many dac chips (Burr-Brown,Analog Devices,Cirrus Logic,but no AKM or Phillips) through my doors and in direct comparison with the same type of output sections I prefer the R2R multibit chips.

The more advanced the chips get it seems the less good they sound and my personal theory is they are crammed with too many features and where you could once design for a specific section you now can only tweek around the chip itself.

Input receiver,digital filter,oversampling,dac,output amp = all included in the chip !

what does that leave the designer other than the power supply, layout scheme and I/O interfacing ?

Maybe work on the master clock but again not much else beyond interfacing the "black box" that is the modern dac chip.Kinda makes everyone able to whip out a design and maybe that is why we see so many cheap DVD players.
It's easy !

It also comes down to personal taste and what your idea of what sounds good is .
If you come from an analog background of vinyl and magnetic tape you will most often prefer the relaxed sound and apparent low level detail of the R2R chips but if you come from the digital age and that is the only sound you know then you like the apparent gobs of sharply etched detail that is thrown at you by the newer chips.They have more "zing" and are more in your face.

So individual taste and what you think sounds like the real thingdetermines what type of chip you think is the more accurate.

I will still take viny first though but that is a fight for another day
icon10.gif
 
Jul 11, 2004 at 12:38 AM Post #11 of 29
If your question is about which storage medium is preferrable, multibit (aka. PCM , used in Redbook and DVD-A) is technically better but bitstream (aka. sigma-delta, aka. DSD, used in SACD) isn't that far behind.

If you question is about DAC implementation, no pure bitstream DACs have been designed since the early 1990s. They were popular in the late '80s and fizzled out by the early '90s. All modern DAC chips are either some hybrid of the two techniques or (less popular) pure multibit. The reasoning behind the hybrids is both technical and practical, because they combine the advantages of multibit (dynamics, linearity) and bitstream (low-level accuracy) and because the drive towards SACD makes it more convenient/economical to have one chip that can handle both PCM and DSD. Hybrids can do this easily, with only the bitstream half of the DAC being used in DSD conversion.
 
Jul 11, 2004 at 6:22 PM Post #12 of 29
adding to my above post :

This is my preference for the DA Converter only , the reproduction of the music where my dacs are all from 16-20 bits.
For Ad Conversion i tend to go in the opposite direction and like the 24 bit/192 khz multi-bit sigma-delta converters.
With the DAC i am interested in the ultimate sound and not how it specs out.
But for recording the requirements are different.Here the extra bits can be considered "headroom" like the old "0db with headroom of 6db or 9db " point of analog recording but in the case of digital it is noise floor and bit accuracy. A 16 bit ADC more often than not only gets you 14 tru bits so 24 bits is in reality closer to 20 when all things are considered-noise floor/headroom/overload margins.

Overload a digital recorder and you will know what bad sound is !


So it is not about sound but capturing the input signal . The input buffers are selected not for sound but for AC and DC accuracy ,settling time and noise floor.It makes no sense to have a 24 bit ADC and mate it to an input stage capable of only 16 bits accuracy due to noise and settling time limitations.
Clock stability is another important factor as is the ability to sync adcs together for multichannel recording and editing.And finally the ADC for the conversion ,noise shaping,dither,etc.

I personally have no problem enjoying 24 bit recordings played back at 16 or 18 bits because in reality the actual bit loss is very small.

Just like with most pro or semi pro equipment where you will see "inferior" active and passive parts the requirements are different .It is easier to record a performance than it is to reproduce it at the other end.
 
Jul 11, 2004 at 6:42 PM Post #13 of 29
Quote:

Originally Posted by Wodgy
If your question is about which storage medium is preferrable, multibit (aka. PCM , used in Redbook and DVD-A) is technically better but bitstream (aka. sigma-delta, aka. DSD, used in SACD) isn't that far behind.


Let’s rephrase this question and ask rather that for the purposes of music encoding, PCM and SDM, which is better? I think that compared to SDM, PCM is a simple but crude encoding algorithm compared to SDM. The main thing I think that 'straight' PCM has going for it is its simplicity. Compared to PCM, SDM looks quite complex however as a predictive coding algorithm, I think it has some keys advantages over non-predictive coding algorithms such as PCM for applications with fairly defined bandwidth such as music.
 
Jul 11, 2004 at 6:55 PM Post #14 of 29
Give me a high end multi-bit converter, a high end bitstream converter - and the master tape (preferably analog)! Then we'll have a blind test. In any other case all we have is preferences.


Regards,

L.
 

Users who are viewing this thread

Back
Top