Maybe it IS all in the DAC chips


I’ve often pushed back against the idea of purchasing any piece of audio gear based on the parts contained, especially DAC’s.  While the Digital to Analog Converter (DAC) itself is the key component, so much goes into making a good sounding product,  like stream buffers/jittter removal, power supplies and analog output stage that I thought this could lead buyers astray.

Still, having done that, and still thinking that I correctly warned others to avoid buying just on DAC brands alone, I have found myself after many years, returning again to vintage Burr Brown based DACs and home theater processors.  

Color me changed with experience. 

erik_squires

@mikhailark 

It is not in the chip, it is in subsequent analog filters that are necessary and are specific to each chip type. DAC output requires brickwall filter

Really?  The steepness of the filter needed depends on the sampling frequency, and it is designed to remove artefacts beyond human hearing.  So a dac designed for CD playback with quadruple oversampling requires a much less steep filter than one operating at 44-kHz.

DSD hardly needs any filter - it operates at a minimum of 2.8-MHz.  No brickwall filter required there

@faustuss 

SACD is a multi-format medium from single layer SACDs that will only work in players capable of decoding them along with Super Audio CD that contain multiple formats on a single disc and are designed to be backward compatible with all CD players and not requiring any "down conversion" whatsoever. Just look at the labeling on your SACD's packaging.

I really don't know why you bother putting finger to keyboard laugh.

Some Super Audio Compact Disk (SACD) releases have an optional, separate RedBook CD layer so they be used on any CD player or transport to play 2-channel 16-bit 44,100 sample-per-second PCM audio, Exactly the same as a CD because that is what that layer is EXACTLY.

The much higher density SACD layer normally contains multi-channel 5.0 or 5.1 DSD at about 2.8-Mbps per channel.  Often there is a stereo (2-channel) DSD mix as well.  To play this layer you need a SACD capable player or transport.

If you have a SACD player that has to down-sample high resolution DSD to Redbook CD quality, you would probably be miffed and feel cheated on.  That is my Reavon experience and the reduction is sound quality was almost immediately apparent - within 20 seconds.  No A / B required.

the vast majority of the DAC chips available for the last couple of decades have incorporated PCM up to 192 Khz as well as DSD capability all the way up to 512 when it became available

Let's be clear - DSD64 samples at 64 times the rate of Redbook CDs.  DSD512 is eight times faster again.  I would not make that generalisation because it is not true!

Probably just the particular chip's data sheet you looked at because TI would not let it's pants down and not be competitive in that market segment

Why don't you read the spec sheets for the 24-bit and 32-bit Burr Brown (TI) chips Reavon use.  You will see that TI can provide an external chip that can decode DSD - the dacs themselves cannot.  My suspicion is that the external chip down-converts to CD quality PCM.  PCM is all these TI dacs can handle.

Now please give your fingers a rest

@richardbrand 

SACD is a multi-format medium from single layer SACDs that will only work in players capable of decoding them along with Super Audio CD that contain multiple formats on a single disc and are designed to be backward compatible with all CD players and not requiring any "down conversion" whatsoever. Just look at the labeling on your SACD's packaging.

"I really don't know why you bother putting finger to keyboard laugh.

Some Super Audio Compact Disk (SACD) releases have an optional, separate RedBook CD layer so they be used on any CD player or transport to play 2-channel 16-bit 44,100 sample-per-second PCM audio, Exactly the same as a CD because that is what that layer is EXACTLY.

The much higher density SACD layer normally contains multi-channel 5.0 or 5.1 DSD at about 2.8-Mbps per channel.  Often there is a stereo (2-channel) DSD mix as well.  To play this layer you need a SACD capable player or transport.

If you have a SACD player that has to down-sample high resolution DSD to Redbook CD quality, you would probably be miffed and feel cheated on.  That is my Reavon experience and the reduction is sound quality was almost immediately apparent - within 20 seconds.  No A / B required."

the vast majority of the DAC chips available for the last couple of decades have incorporated PCM up to 192 Khz as well as DSD capability all the way up to 512 when it became available

"Let's be clear - DSD64 samples at 64 times the rate of Redbook CDs.  DSD512 is eight times faster again.  I would not make that generalisation because it is not true!"

Probably just the particular chip's data sheet you looked at because TI would not let it's pants down and not be competitive in that market segment

"Why don't you read the spec sheets for the 24-bit and 32-bit Burr Brown (TI) chips Reavon use.  You will see that TI can provide an external chip that can decode DSD - the dacs themselves cannot.  My suspicion is that the external chip down-converts to CD quality PCM.  PCM is all these TI dacs can handle.

Now please give your fingers a rest"

Why do you continuously go on these endless dissertations reiterating what I just said but without all the filler words. Yes DSD 512 is incorporated in many off the shelf chips just as I said when it became available. You should also check TI's catalog before spouting off because yes, they do incorporate Hi-res PCM and DSD on their chips. Why wouldn't they, they've been the preeminent semiconductor developer and manufacturer forever.

You might try being less condescending and insulting when you respond to my posts. I'm only giving it back to you and it's quite obvious to anyone capable of doing their own research, you're not an authority. I obviously intimidate the daylights out of you.smiley

@richardbrand - oversampling does not increase amount of information in the stream (per information theory) so although it helps somewhat, it does not eliminate need of the proper filter and there is always compromise between slope and phase linearity and quality of parts. My point was that analog part is much easier to mess up. DSD still requires low pass albeit less complex. I can see from GR Research YT videos how often crossover filters in $$$$ speakers are messed up. So there.

@mikhailark 

oversampling does not increase amount of information in the stream (per information theory) so although it helps somewhat, it does not eliminate need of the proper filter

As you say, repeating a sample four, eight or more times does not add any information (not even helping somewhat).  What it does do is change the proper filter characteristics needed, from steep slope to shallow slope with far less phase shift, etc.. Philips knew this from the get go.

My beef was with your earlier statement that "DAC output requires brickwall filter" which is untrue for oversampling dacs and especially untrue for dacs that decode DSD natively.