[BackToAnalogue]

I agree this is extremely helpful. All of Purrin's observations support my understanding (or current theory, since it is not accepted by anyone apart from me at the moment) of how 'information' about music is transmitted to our auditory system and how this is affected by different formats and the various limitations of different approaches to DAC design. My thinking is summarized in an earlier post here.



http://www.head-fi.org/t/693798/thoughts-on-a-bunch-of-dacs-and-why-i-hate-chocolate-ice-cream/2025#post_11136795



I am glad that someone, who has clearly developed the necessary skills of highly trained hearing...............................................

I have spent many months thinking about this; it is an enormously complex subject; it is not fully understood by any of the people working at the real sharp end of DAC design; the information that is available online is riddled with misunderstandings, bad Maths, incorrect assumptions, and half understood techniques. The terminology and measurements used date from a different era, and a different domain (analogue).

Oh my goodness.

With respect, you are not trying to say that you know more about this subject than someone who has actually designed a DAC, are you?



It is one thing to study and read about DACs, it is quite another thing to actually design a DAC.

 


Er no, I don't think that I said that did I? If I implied it then I certainly didn't intend to.

All I said was that I have a far better understanding than I did 6 months ago and it has taken a great deal of time and effort to get there because of the factors I listed. I have also identified several widely held misconceptions.

It is very likely that the couple of dozen or so people in the world who do this for a living know everything that I do, and a lot more besides, but they don't write it down in blogs like this. And as far as I am aware none of them have stuck their heads above the parapet to correct the '24-bits makes no difference' argument because it is a good argument (I was convinced by it for several months) and hard to explain why it is wrong to a non Mathematician. It is easier just to wait for the people who care to use their own ears, because the vast majority of people don't care.

 

[purrin]

Taking hires original material (DSDx128, 24/192 PCM, etc.) and inspecting data in something like Adobe Audition to confirm actual hires content (there is quite a bit of fake hires content out there), and downconverting to various bit and sampling rates, my personal threshold of audibility is 18 bits. Sampling rate past 44.1 doesn't seem to matter*. That is, I cannot tell the difference between DSD/ hires PCM and the same material downconverted to 44.1kHz at 18 bits. At 16 bits, I start to hear a difference - 18 bits at 44.1kHz seems enough; with the caveat being good A-D or as pure a chain as possible.
 
*Some DACs sound different (not necessarily better, not necessarily worse) at different input sampling rates, real or non-existent hires data or not.
 
The quality of A-D conversion and final mastering is HUGE. That's the other half of the conversion process. Starting off the bat, there's no chance in hell for most recordings to sound like the tapes they came from regardless of how good the playback gear. This goes back to what I was saying about the downsides of better D-A.
 
Some pure chain DSD stuff is really good, but different. Unfortunately, the pure chain DSD recordings are rare and usually consists of craptastic "audiophile" music that I don't care for, i.e. mediocre soul-less girl vocals + guitar -or- mediocre soul-less girl vocals + lightweight piano. Bleh.

 

[En_R]

   
As good as the Theta Gen Va was, it was soundly beat by an mbl 1511D (which is a Delta Sigma design) in my setup. I highly recommend an audition of an mbl if you get a chance. It certainly changed my thoughts on Delta Sigma dacs. Not only that but it sadly destroyed even my vinyl setup. Just wondering, but what vinyl setup have you compared the SFD-1 to? What tubes are you using in it?  It didn't sound much like my vinyl rig at all no matter which tubes I tried (Amprex, RCA, Siemens, Tungsol, and Telefunken). I can't say Ultra Analog based designs sound like vinyl. I've owned three Ultra analog based dacs. A Levinson 35, Audio Research DAC1- 20, and a Sonic Frontiers SFD-1. Out of those three, the Levinson came closest to vinyl but was still a ways off.

 
MBL 1511/1531 are both very colored. Mids are warm and dark, with very noticeable roll-off in the top end. My biggest problem with the lower end MBL sources is that they lack layering throughout the spectrum (mushy bass, lack of instrument separation, etc), and congested soundstage. If the 1511 is your reference then perhaps you have different sonic priorities than most.

I'd like to point out that for the price of the MBL DAC you could have purchased all the prior DACs listed, or you could have gotten a real setup like a Levinson 30.6 + 31.5, instead of the inferior 35. In fact for a bit more you could have gotten a Wadia 7+9 =/

It would be good to say what you paired the levison/sfd-1/DAC1 with as well. They don't have USB, and transports matter. So does the coax/BNC cable.

 

[BirdManOfCT]

 
Some pure chain DSD stuff is really good, but different. Unfortunately, the pure chain DSD recordings are rare and usually consists of craptastic "audiophile" music that I don't care for, i.e. mediocre soul-less girl vocals + guitar -or- mediocre soul-less girl vocals + lightweight piano. Bleh.

 
LOL. So true. I have run into a few good songs in the process, but most of them I play and think "Really?".

 

[purrin]

Heh, glad somebody understands

 

[drez]

   
Theta Gen V (R2R) with Data III transport or OR5 USB-AES2 gizmo is more resolving than the best resolving S-D DACs that I have heard, including those from Bricasti, Berkeley, Auralic, etc. I would agree with you that most if not all modern R2R implementations are the suck in terms of resolution. This is what prompted me to go back in time (when the R2R implementations were better) to see what R2R can truly do. Not this current limp-dip PCM1704 garbage, which I only partially embraced.
 
Yggy (an R2R DAC about to be released) easily and significantly beats Gen V (and thereby all other modern S-D DACs I've heard) in terms of resolution. This is what happens when you have two or three guys crazy enough to stick military grade chips with < 1 LSB error and develop crazy firmware so the chip will eat audio data. DAC accuracy / linearity = resolution. Proof is in the pudding via listening and experimenting. Try building a DAC yourself. Start stacking chips for better linearity and hear the result. Yes, I've done this. Even mediocre DAC designers know this.
 
 
That is a very good point. My SFD-1 DAC sounds most like vinyl. It emulates vinyl. The UltraAnalog based implementations tend to sound that way. Over time, I felt that the SFD-1 imparted too much of a vinyl characteristic on everything. On the other hand, the PWD-2, M1, Vega (all superb SD DACs) tended to impart garbage that I know isn't there. Only the Gen V provided the best "what you get out is what your put in" aspect for me: analog mixed and mastered recordings tended to sound analog; recordings where digital (of varying quality) was part of the chain sounded digital (or varying quality). I use the Gen V to get the maximum from the best recordings I have, but there is a cost to poor recordings. This is why I keep the the SFD-1 DAC when I want something more forgiving that's always sounding "like vinyl" regardless of recording.
 
Once you hear enough stuff, you can decide for yourself where you want to sit. There is no clear win. I tend to lose a lot because I won't play bad recordings through the Gen V, which is what I use for playback the most. This tends to cause an unhealthy obsession with hunting down the best masters.
 
 
That's the $64,000 question. Rarely can you have everything you want. Best to develop your senses and determine where your priorities lie. It usually comes down to finding equipment which is a balance of attributes you like at a price you can afford. Jitter should not be an issue in 2014 if you have a decent budget. Computer to DAC data transmission, i.e. USB, is a more serious concern, but there has been a lot of progress in this area. Actually, the most serious concern is the proliferation of low cost SD DACs and how audiophiles are willing to take it up the butt with all the hashy treble garbage by paying thousands and thousands of dollars for DACs which are essentially tweaked manufacturer evaluation boards.

Yeah I'm pretty intrigued by the Ygg.  personally I am yet to be able to identify digital hash, possibly because the only non SD DAC I have used has been PCM1704, and well I would rather take the digital hash

.  Most of the treble issues I have detected have been able to be reduced through improved input signal (computer and transport, cables) and vibration management (isolate a component/rack/shelf too much and you start to hear it's own resonant characteristics…) but to be honest I have come to accept some brightness glare assuming it is part of the signal.  Mics, mixing consoles etc.  My general philosophy is that if something can sound less analytical/glare/splashy while sounding more resolving, progress has been made, however I am stubborn in not giving up a single drop of resolution.
 
I will probably give the Ygg a spin when that comes out, not having heard non-soft or grungy sounding multibit (read: PCM1704) I am skeptical but hoping to be surprised.
 
Computer/USB is definitely a big barrier in digital.  Many of my friends have said that even a mid-fi CD player can beat an average workstation, but I can't comment on this personally.  I have a pretty tin-hat music server, and recently built a general use workstation, and comparing the two, the tin hat server is much more resolving, but not any less bright.  I have tried disabling cores to emulate lower CPU power use (and lower performance) and this yields a more laid back sound, but some resolution is missing.  So unfortunately I am unable to determine if the reduced brightness is from lower electrical noise or lower computer performance.  Things like the JCAT USB card and disabling hardware/software/OS threads have however yielded more of the no-compromise improvements.  Well of course there is a compromise though - in my case usability is horrible.

 

[BackToAnalogue]

I think that there may well be a problem with some USB Asynch implementations given the number of posts I have seen from people reporting this. ASynch USB should perform better than SPDIF with no jitter being introduced by the comms clocking or disc reads, but that relies on both ends of the USB connection having high performance USB interface chips and high CPU performance. Windows is not designed to help a DAC maintain a steady stream of data, it is designed to do whatever Microsoft deem to be the most important activity first. Your DAC comes last.

The first thing to do is a whole list of small tweeks and settings to ensure that Windows behaves itself. Also a fast PC, USB 3 ports (much faster if you can put up with possibly having to fix compatibility problems) .and anything else to help PC performance.

All current DAC chips (correct me if I'm wrong) were designed based on a synchronous data flow, SPDIF. To then have to bolt on a very fast asynchronous data flow is quite a big ask and I would expect to have to increase processor power by between 5 and 10 times to ensure no buffer overflows or underflows. If you then also take into account Windows uneven data flow it still may sometimes have problems. The easiest way to fix this is to introduce another (intermediate) buffer between the PC and the DAC. A $25 USB 3.0 4 port hub will do the job fine. This should smooth the data flow to the DAC and will introduce an additional time delay which it uses to do this.

 

[tonykaz]

Mr. Purrin & fellow searchers of DAC,
 
Is there any point in discovering the DAC used in the recording process? and then finally the Mastering process?  
 
Somehow, it seems to me, the DAC ( I have ) can do no more than reverse the work done by the two DACs proceeding it.  
 
In the few Mastering Studios I've visited the engineers have an Atomic Clock ( 10M ) they point to, where is our DAC's Clock?, a little Quartz crystal which we use to control our FM Radio Transmitter's Frequency?, accurate and stable enough to satisfy the FCC, is it enough to keep our music in focus?, it does seem good enought for consumer use but is this the best we can expect for a proper Musical Instrument that we hope our Home Audiophile DAC to be?  
 
Tony in Michigan

 

[computerparts]

   
MBL 1511/1531 are both very colored. Mids are warm and dark, with very noticeable roll-off in the top end. My biggest problem with the lower end MBL sources is that they lack layering throughout the spectrum (mushy bass, lack of instrument separation, etc), and congested soundstage. If the 1511 is your reference then perhaps you have different sonic priorities than most.

I'd like to point out that for the price of the MBL DAC you could have purchased all the prior DACs listed, or you could have gotten a real setup like a Levinson 30.6 + 31.5, instead of the inferior 35. In fact for a bit more you could have gotten a Wadia 7+9 =/

It would be good to say what you paired the levison/sfd-1/DAC1 with as well. They don't have USB, and transports matter. So does the coax/BNC cable.

 
Can't agree at all. I find it interesting how people who prefer the Levinson sound find the mbl to sound colored. Your comments seriously make me wonder if you had something setup wrong or if you had a faulty unit. You'll not find one review of anything from mbl reflecting those comments. In my setup the mbl 1511D paired with a Theta Jade had excellent transient response, extension, and separation. The dynamics were second to none. I had many compliments about my system from visitors back then. The most common complement referred to it such as live music which I agree with. We all have our preferences and I found I don't particularly care for the Levinson sound. The 35 may be inferior to the 30.6 but it still retains the same flavor since the designs are very similar. As someone who plays guitar, I find that flavor to be rather artificial sounding.
 
As far as price, I didn't pay anything near retail for the mbl since I got it used.

 

[Sorrodje]

  Heh, glad somebody understands

 
I guess he's not alone to understand.

 

[computerparts]

 
Yeah I'm pretty intrigued by the Ygg.  personally I am yet to be able to identify digital hash, possibly because the only non SD DAC I have used has been PCM1704, and well I would rather take the digital hash

.  Most of the treble issues I have detected have been able to be reduced through improved input signal (computer and transport, cables) and vibration management (isolate a component/rack/shelf too much and you start to hear it's own resonant characteristics…) but to be honest I have come to accept some brightness glare assuming it is part of the signal.  Mics, mixing consoles etc.  My general philosophy is that if something can sound less analytical/glare/splashy while sounding more resolving, progress has been made, however I am stubborn in not giving up a single drop of resolution.
 
I will probably give the Ygg a spin when that comes out, not having heard non-soft or grungy sounding multibit (read: PCM1704) I am skeptical but hoping to be surprised.
 
Computer/USB is definitely a big barrier in digital.  Many of my friends have said that even a mid-fi CD player can beat an average workstation, but I can't comment on this personally.  I have a pretty tin-hat music server, and recently built a general use workstation, and comparing the two, the tin hat server is much more resolving, but not any less bright.  I have tried disabling cores to emulate lower CPU power use (and lower performance) and this yields a more laid back sound, but some resolution is missing.  So unfortunately I am unable to determine if the reduced brightness is from lower electrical noise or lower computer performance.  Things like the JCAT USB card and disabling hardware/software/OS threads have however yielded more of the no-compromise improvements.  Well of course there is a compromise though - in my case usability is horrible.

 
IMO a computer is a horrible source or even transport period. It's just not made for that purpose in mind (laptops are even worse yet). In a typical pc, you've got the cpu and gpu emitting frequencies in the ghz with no shielding in sight. The wiring going to the front panel usb port is horrible with minimal shielding. If you're using the onboard audio digital output then that's even worse since that's on the same exact board that everything else emitting noise is on (the motherboard). The power supply is inadequate for audio as the power for everything in the computer comes from one source, again no shielding in sight. Music servers are a bit better but still share some of the same problems regarding shielding and frequency emissions. In a cd player, there is no cpu or gpu contaminating everything in sight nor are there any components for wi-fi or lan. So it's fairly easy to see why a mid-fi cd player will destroy a typical pc, workstation, or even a music server. I think one should only expect so much when using a typical pc/workstation/music server for audio.

 

[ciphercomplete]

 So it's fairly easy to see why a mid-fi cd player will destroy a typical pc, workstation, or even a music server.

 
Except that this is not the rule.  I've owned quite a few transports and cd players and my music server beat them out.  Two of my friends own PWT with similar DACS to mine (Master 7 and SA2) and by comparison you don't hear all the dreaded nasties of a computer transport in my system vs theirs.  Saying that any old mid fi cd player would beat any computer is false.  Computers, despite their issues, can be done right provided you know what you are doing.  Where people screw up imo is attempting to use their personal computers as a media pcs or using over the top processors and unnecessary GBs of memory that need lots of power.  Dedicate a low power consuming pc to music only and get a good linear pc power supply and get a good dac (or usb to spdif/I2s converter) that effectively deals with usb power issues.  The only caveat there really is is expense, but you get awesome convenience in return.   

 

[computerparts]

   
Except that this is not the rule.  I've owned quite a few transports and cd players and my music server beat them out.  Two of my friends own PWT with similar DACS to mine (Master 7 and SA2) and by comparison you don't hear all the dreaded nasties of a computer transport in my system vs theirs.  Saying that any old mid fi cd player would beat any computer is false.  Computers, despite their issues, can be done right provided you know what you are doing.  Where people screw up imo is attempting to use their personal computers as a media pcs or using over the top processors and unnecessary GBs of memory that need lots of power.  Dedicate a low power consuming pc to music only and get a good linear pc power supply and get a good dac (or usb to spdif/I2s converter) that effectively deals with usb power issues.  The only caveat there really is is expense, but you get awesome convenience in return.   

You're right it's not the rule. There are always exceptions. For example, a $10k music server should beat out a $1k cd player. I can't remember many people raving about the PWT and I've never heard a setup with one so I can't comment. However, you stated "you don't hear all the dreaded nasties of a computer transport in my system vs theirs" which tells me it's not a valid comparison unless I'm misunderstanding. Have you put the music server in their system to see how it sounded? Have you put their PWT and dac in your system to see how it sounded?
 
Also, when you were comparing your music server to other dacs and cd players, were they all in the same price range as the music server? Did the rest of your system stay the same or were their any other changes after you bought the music server? What I'm getting at is it's easy to say X beats Y but if other components/ancillaries in the system have changed, then it is not a valid comparison. I'm sure you can obtain good results from a music server, but a pc is still a pc and it will have its limitations compared to a dedicated dac and transport.

 

[cizx]

I hear that if you keep adding zeroes, the sound keeps getting better.

 

[Chris J]

I hear that if you keep adding zeroes, the sound keeps getting better.

Ones and Zeroes.
Go from 16 bit to 24 bit and the sound improves by 50%.