Newbie question on DAC

[Crosstan]

Hi all, pardon me if the question sounds stupid but I am trying to get my head round it.

How does higher bit rate of DAC helps with the resolution of the music? I read about all the hype with ESS 32bit DAC and I'm sure it is a good chip. But what makes a 32bit chip better than a 24bit chip? I mean the source material is already encoded in 16bit (red book CD) right? So the extra bits on the DAC are wasted right?

Also how does having more DAC chips help in the analogue output? I see some high end DACs quoting having 8 DAC chips for stereo output.

Appreciate all insights as I am trying to make a decision on which diy DAC kit to get as a first project on my diy journey

[SY]

With dithered sources (99.99% of what you can get), the difference isn't resolution, it's signal to noise and dynamic range. In home situations, it's not likely that you'll have an audible noise advantage beyond 16 bits.

I'd base my decision on what sort of source material you intend to listen to- if very little or none is 32 bit, why pay for that? Now, if you're after test signals and having best performance no matter whether audible or not, that's a different story, but if (like me), your primary interest is the music, you don't have to chase numbers.

[Pano]

On the other hand...

If your recordings are 24 bit, or you wish to use the volume control onboard the DAC chip, then 24/32 bits can be a good thing. A lot of music player software processes in 32 bit, tho I have no idea how to get that out of the computer and in into the DAC. 24 bit yes.

Multiple chips are supposed to provide lower noise and more drive current. More drive current often sounds good. I agree with SY, for home the 16 bit S/N ratio is plenty.

[fb]

Having just searched for some 'no fuss'/high value DACs for my stereo 3-way active system, have a look at:

- Twisted Pear Opus: Has balanced out. Controller available (Volumite) to use volume control on DAC chip. (I just ordered 3 of these; apparently they don't pop at turn on/off - great for connecting straight to amp)


- New Small DIY Gigawork Dac?

[DF96]

You ask about bit rate, then talk about word length, so I assume it is the latter you are interested in. A 16-bit recording played directly on a true 16-bit DAC is fine, but not all '16-bit' DACs are accurate in the last bit position. If you want to fiddle with the sound (e.g. digital volume control) then more bits are needed to maintain resolution.

Parallel DACs can improve sound, if the DACs have random errors in the last bit position or two. The Central Limit Theorem says that the errors will tend to cancel out. Four or eight '16-bit' DACs which are really only good to, say, 14 bits can be paralleled to get something like 15.5 bits. If the DACs are already perfect, or if the errors are correlated, then no improvement results. Similarly, second-order distortion can be cancelled by using push-pull DACs, but third-order is left unchanged.

[SoNic_real_one]

The 32 bit it is just bull... For Digital Processing, yes you might need more bits, but for a DAC that cannot output more that the 20-22 bit equivalent quality, that is just marketing. Even if you use as source some 24 bit audio (yes there is such thing, not everybody records in the antique CD/redbook format).
PS: TI just "reworked" some of their DAC's to accept 32 bit format data. Of course the final audio quality remained the same.

[Crosstan]

Thanks for all the advise thus far. Looks like I am really quite clueless with DAC. While I do some parallel reading of some undergraduate text on electronics, I hope that the more experience and knowledgeable folks out here could help me get up to speed faster.

To DF96:
I have assumed that the bit refers to the word length (or the number of available quantized states) while the frequency refers to the sampling rate of the analogue signal. So I can understand if a source material has higher bit rate, it would sound better (in theory anyway). I can understand how the extra bits in DAC are used for digital filtering or volume control. However, you seem to suggest that the bits quoted on DAC refers to another thing. Could you enlighten me further or point me to some good reading material for DAC?

From reading the rest of the replies, am I right to say that I should have no audible advantage of building a DAC using a 32bitDAC as opposed using a 24bitDAC? So I could just save myself some money and start the project using a 24bit DAC chip?

Thanks fb for the recommendation. Will definitely check out the OPUS.

[DF96]

If you have a 16-bit recording, then a 24-bit DAC does not change the bit rate as you just use zeroes for the lowest 8 bits.

Roughly, bit rate = sampling rate x word length (for a lossless system). However, bit rate is a term which is mainly used for comparing lossy encoders rather than lossless systems like CD. For lossy encoders, with other things being equal (which they are not), higher bit rate means higher quality.

Given issues like dither, there seems little point in using a DAC which is higher in word length than the data sent to it. A properly dithered signal can achieve resolution beyond the raw DAC accuracy.