[Yoga]

Thanks @romaz.
 
Currently pondering which route to go; mono or d1-dual + Egoista.
 
(For Abyss.)

 

[a1uc]

I also came from the Vega to the D1 Dual , I think the Vega is the best DAC in that price range . The only reason I moved on was poor customer service , Vincent for sure is on the ball with customer service .

 

[romaz]

My gut feel is that for headphones, money is better spent on the source (reclocker / server what not) 

 
Completely agree.

 

[Yoga]

Regarding transport, has anyone used any Aurender products? The W10/W20 get fantastic reviews, and their X100 offers their sonic benefits at a much cheaper price ($3500).
 
I have the SMS100 but find it rather lacking.

 

[romaz]

  I have been reading a lot of great comments in regards to tonality, staging, backgound blackness etc. My main concern with the TotalDAC DACs are the resolution figures. I have only read speculation in regards to the ENOB and that was from a while ago. Did anyone speak with Vincent meanwhile, or has some more information?

 
This is a good question because we all like to look at hard numbers when we assess equipment but there is also a lot of misunderstanding about what these numbers mean.  There is absolute resolution and some will use ENOB (effective number of bits) as a measure of this and then there are the things that muddy resolution (ie distortion, noise floor, time smearing due to jitter, skew and drift in the sampling rate, etc).  The better DACs handle the latter well and many will argue that overcoming these timing and noise issues are what defines the best DACs.  As for bit depth, I'll let you decide how important this number is.  
 
When it comes to bit depth, MSB is supposed to be the best and they tout this frequently.  The TotalDac utilizes Vishay Foil resistors that many consider among the best resistor there is and Vincent uses the more expensive variety with a minimal tolerance of 0.01%.  While his R2R is a 24 bit ladder, the Vishay Foil 0.01% resistor is limited to only 14 bits.  Many vendors are not honest when they report this but this is the ENOB of the TotalDac, only 14 bits.  The Schiit Yggy reports 21 bit capability.  Most delta sigma DACs including the $150 AQ Dragonfly and my Bricasti M1 claim to be 24 bit DACs.  The MSB Diamond DAC is supposed to be the best at 27 bits of resolution.  Based on this figure, you would think the TotalDac is not a very resolving DAC.  In fact, if you use this as your guide, the $150 Audioquest Dragonfly which uses a 24 bit ESS Sabre chip should resolve better than even the $30,000 TotalDac d1-twelve but our ears will tell us there is more to resolution than bit depth.
 
I will try my best to explain although I also shared this on the Nagra thread. Resolution is important only as far as what a human brain/ear can discern and this is quantified in a variety of ways and not just with bit depth.  One way is a measure called time resolution.  The human brain/ear has the ability to discern variations in sound if it occurs over a time span of 7 microseconds or more.  What this means is if variations of sound (such as 2 instruments playing) occur over a time span <7 microseconds, our ear will not be able to discern or resolve it.  At >7 microseconds, we will know if something is missing.  With Redbook (PCM), sampling occurs at 44kHz which equates to an auditory time resolution of 20.8 microseconds according to studies performed by Meridian, meaning that our ears can clearly tell that information is missing when you listen to a CD.  Those of us who have compared CD to vinyl, for example, know very well this is true.  At 96 kHz sampling, the auditory time resolution improves to 10.4 seconds which is much better but compared to vinyl, our ears will still be able to tell that vinyl is superior. It is only when you boost the sampling rate to 192 kHz that you bring down this auditory time resolution to 5.2 microseconds, finally below the 7 microsecond threshold.    At this sampling rate, your ear should be UNABLE to discern between vinyl and digital when it comes to time resolution.
 
As for bit depth, for music this actually relates more to dynamic range and not true resolution but it will have some bearing on what we perceive as low level resolution.  Dynamic range is the range in volume between the noise floor and maximum volume.  Each bit of data represents 6 dB of dynamic range.  Redbook is recorded in 16 bits = 96 dB.  Analog (vinyl) is recorded at 20 bits = 120 dB.  This is not a linear scale and 16 bits to 20 bits represents a 16x increase in dynamic range.  Again, this is why vinyl can sound more dynamic than CD.  At 24 bits, however, you get a dynamic range of 144 dB and so finally, at this bit level, digital should surpass vinyl when it comes to dynamic range.  
 
Well, if you look at published dynamic range figures, you will see just how well bit depth correlates with actual dynamic range.  The most practical way to measure and compare dynamic range figures is the S/N (signal to noise) ratio even though DR and SNR are not exactly the same thing.  Some companies will report dynamic range numbers (absolute peak values) but more companies will report S/N ratio (average peak values) and most believe this to be the more helpful number.  The $150 AQ Dragonfly DAC which has a 24-bit Sabre chip has a measured S/N ratio of 113 dB while playing a 24 bit digital file.  Remember, at 24 bits you should have a S/N ratio of 144 dB.  The Schiit Yggy with its 21 bit DAC reports a higher S/N ratio of >117 dB.  The MSB Analog reports an impressive S/N ratio of 140 dB.  The much more expensive MSB Diamond DAC V at 27 bits somehow resolves no better than the Analog with a S/N ratio of 140 dB.  Clearly, something is preventing these high-bit DACs from reaching their theoretical potential.  So how does the TotalDac measure?  From the d1-single all the way up to the d1-twelve, the quality of the resistors are the same and so each TotalDac has a lowly bit depth of only 14 bits yet Vincent's most basic d1-single has a S/N ratio of close to 150 dB meaning even the entry level TotalDac d1-single has greater dynamic measurements than the MSB Diamond DAC V.  The TotalDac monoblocs have a S/N ratio that approaches 160 dB.  With the d1-twelve, the noise floor is so low it is literally unmeasurable by Vincent's equipment and so a S/N ratio cannot be calculated.  Read another way, the d1-twelve is almost pure signal with no noise.  If this is how you wish to measure resolution as MSB would have you believe, then the d1-twelve may well be the highest resolving DAC in the world and the entire TotalDac line outresolves all of the MSB DACs.
 
What is the practical meaning of all of this?  Well, it's not what most people think.  No one really needs to hear any sound that is more than 120dB.  My comfortable listening levels don't often go beyond 90dB because I value my hearing.   In fact, sound levels of >160dB are known to be lethal.  What is important are the dynamic contrasts of an undistorted signal compared to the noise floor.  Because the noise floor of the TotalDac is so black, the color contrasts are deeper and sharper.  The tone is richer and fuller.  With the TotalDac, you feel like you get punched in the gut from a sound that comes from nowhere.  This to me is the sound signature of all TotalDacs.  From the d1-single to the d1-twelve, it becomes a matter of degree.

 

[romaz]

  Regarding transport, has anyone used any Aurender products? The W10/W20 get fantastic reviews, and their X100 offers their sonic benefits at a much cheaper price ($3500).
 
I have the SMS100 but find it rather lacking.

Yes.  I have heard both the N10 and W20 and they are indeed fantastic and it was my experience with the N10 that first opened my eyes to the importance of the source.  While MSB and DCS can do it, the TotalDac cannot take advantage of the W20's external clock option or the dual AES/EBU option and so I have not focused on the more expensive W20.  Connecting via USB, I found no difference between the two in terms of SQ meaning they were both equally superb.  
 
As for the X100/N100, I have heard these as well and they are not in the same league as the N10/W20, not really even close but that shouldn't come as a surprise as the N10 is more than twice the price of the X100.  I believe the difference is the OCXO clock in the more expensive units but also better shielding.  The X100 is more in line with your SMS100 but also the Moon Mind, Auralic Aries, Wyred 4 Sound MS-2 and your typical CAPS server which is to say these units are a bit above a typical Mac Mini with Amarra/Audirvana but not by much.  I am speaking from experience as I have tried each of these units and also personally own a Mac Mini, purpose-built Linux-based server and Auralic Aries with linear PSU.  These units have either been re-purposed or are up for sale.  Unfortunately, my experience has been that unless you spend at least $5,000 USD, you will not find a source that is much better than a Mac Mini.
 
If you are going to roll DACs at the level of any of the TotalDacs, any MSB DAC, a Nagra, Light Harmonic, EMM Labs, Berkeley, Bricasti, DCS, etc., you really owe it to yourself to get a source that befits it.  This has been the most misunderstood and underrated part of the audio chain as most believe when it comes to the source, bits are bits.  The TotalDac is so revealing of its source that you will do it a grave injustice if you pair it with something not on the same level.  I will go so far to say that I would rather own a d1-single with a d1-server (otherwise known as the d1-integral server) than the d1-monobloc paired with something like a Mac Mini or even an Auralic Aries.  In fact, I am testing this as I write this and the d1-server/reclocker is making a bigger difference compared to going from the d1-dual to the monobloc.

 

[nepherte]

  In fact, I am testing this as I write this and the d1-server/reclocker is making a bigger difference compared to going from the d1-dual to the monobloc.

 
This is really valuable information. I've been wondering what route to take and my gut always tells me to take the server/reclocker route. While it's not in the cards for me right away, it probably will be in the next few months or so.

 

[Yoga]

  Yes.  I have heard both the N10 and W20 and they are indeed fantastic and it was my experience with the N10 that first opened my eyes to the importance of the source.  While MSB and DCS can do it, the TotalDac cannot take advantage of the W20's external clock option or the dual AES/EBU option and so I have not focused on the more expensive W20.  Connecting via USB, I found no difference between the two in terms of SQ meaning they were both equally superb.  
 
As for the X100/N100, I have heard these as well and they are not in the same league as the N10/W20, not really even close but that shouldn't come as a surprise as the N10 is more than twice the price of the X100.  I believe the difference is the OCXO clock in the more expensive units but also better shielding.  The X100 is more in line with your SMS100 but also the Moon Mind, Auralic Aries, Wyred 4 Sound MS-2 and your typical CAPS server which is to say these units are a bit above a typical Mac Mini with Amarra/Audirvana but not by much.  I am speaking from experience as I have tried each of these units and also personally own a Mac Mini, purpose-built Linux-based server and Auralic Aries with linear PSU.  These units have either been re-purposed or are up for sale.  Unfortunately, my experience has been that unless you spend at least $5,000 USD, you will not find a source that is much better than a Mac Mini.
 
If you are going to roll DACs at the level of any of the TotalDacs, any MSB DAC, a Nagra, Light Harmonic, EMM Labs, Berkeley, Bricasti, DCS, etc., you really owe it to yourself to get a source that befits it.  This has been the most misunderstood and underrated part of the audio chain as most believe when it comes to the source, bits are bits.  The TotalDac is so revealing of its source that you will do it a grave injustice if you pair it with something not on the same level.  I will go so far to say that I would rather own a d1-single with a d1-server (otherwise known as the d1-integral server) than the d1-monobloc paired with something like a Mac Mini or even an Auralic Aries.  In fact, I am testing this as I write this and the d1-server/reclocker is making a bigger difference compared to going from the d1-dual to the monobloc.

As per usual, your wisdom is invaluable. Thanks Roy.
 
@isquirrel speaks very highly of the CAD server. CAD server + DAC vs TotalDAC mono seems to be the option here then. The mono not requiring a headphone amplifier...

 

[romaz]

 
As per usual, your wisdom is invaluable. Thanks Roy.
 
@isquirrel speaks very highly of the CAD server. CAD server + DAC vs TotalDAC mono seems to be the option here then. The mono not requiring a headphone amplifier...

Sure.  If you have the means and you want the ultimate, the TotalDac mono + CAD could certainly be it.  I have a CAD coming Wednesday so I will be able to compare it directly against the TotalDac Server and Aurender N10.  For those who already own a d1-dual or are looking to save money somewhere so you can direct it to getting a better source, my feeling right now is that you should stay with the d1-dual as I am starting to believe it gets you 75% of what the monobloc provides and that 75% is awfully good.  Also, with the d1-dual, you are able to go direct to headphone without a dedicated headphone amplifier so don't feel you have to get the monoblocs to derive this benefit.

 

[Yoga]

  Sure.  If you have the means and you want the ultimate, the TotalDac mono + CAD could certainly be it.  I have a CAD coming Wednesday so I will be able to compare it directly against the TotalDac Server and Aurender N10.  For those who already own a d1-dual or are looking to save money somewhere so you can direct it to getting a better source, my feeling right now is that you should stay with the d1-dual as I am starting to believe it gets you 75% of what the monobloc provides and that 75% is awfully good.  Also, with the d1-dual, you are also able to go without a dedicated headphone amplifier so don't feel you have to get the monoblocs to derive this benefit.

d1-dual powering cans - I wasn't aware of that, thanks. Do you think it would drive the Abyss well enough?
 
I await your CAD/server findings with anticipation :¬)

 

[romaz]

 
d1-dual powering cans - I wasn't aware of that, thanks. Do you think it would drive the Abyss well enough?
 
I await your CAD/server findings with anticipation :¬)

Yes, another head-fier has reported that his d1-dual powers his Abyss satisfactorily via XLR outputs.

 

[negura]

 

 
I have been looking for the correlation between resolution and DR. Based on the below indeed it appears the resolution is actually directly related to both DR and noise.
 
This is for an ADC circuit: 
Effective resolution = log2 [full-scale input voltage range/ADC RMS noise]
(https://www.maximintegrated.com/en/app-notes/index.mvp/id/5384)
 
If we extrapolate to a DAC then the variables are full-output voltage range (DR) and noise. It is at least interesting how TotalDAC gets those DR figures as shouldn't then the ENOB figure be flying through the roof? This is where things become quite unclear to me in regards to TotalDACs: how is it possible to have 155-160dBfs DR and 14 bits at the same time? Is there something else we should know.
 
As in your example: Analog (vinyl) is recorded at 20 bits = 120 dB (the theoretical/ideal maximum). That is the same formula, which gives: 1 bit = 6dB. The D1 should have 26 bits of resolution. Why is the ENOB so low then (or how can the DR be so high)?
 
Regardless. At budgets exceeding 10K, I would expect excellent performance in all respects (for the respective budget): both great objectively measured performance AND to sound really great. As I couldn't more useful info than that one graph on the Total DAC website - have any reviewers (like Stereophile) or independent ones actually measured their DACs?

 

[Beolab]

I have been looking for the correlation between resolution and DR. Based on the below indeed it appears the resolution is actually directly related to both DR and noise.

This is for an ADC circuit: 
[COLOR=333333]Effective resolution = log2
 [full-scale input voltage range/ADC RMS noise][/COLOR]

If we extrapolate to a DAC then the variables are full-output voltage range (DR) and noise. It is at least interesting how TotalDAC gets those DR figures as shouldn't then the ENOB figure should fly through the roof? This is where things become quite unclear to me in regards to TotalDACs: how is it possible to have 155-160dBfs DR and 14 bits at the same time? Is there something else we should know.

As in your example: Analog (vinyl) is recorded at 20 bits = 120 dB (the theoretical/ideal maximum). That is the same formula, which gives: 1 bit = 6dB. The D1 should have 26 bits of resolution. Why is the ENOB so low then (or how can the DR be so high)?

Regardless. At budgets exceeding 10K, I would expect excellent performance in all respects (for the respective budget): both great objectively measured performance AND to sound really great. As I couldn't more useful info than that one graph on the Total DAC website - have any reviewers (like Stereophile) or independent ones actually measured their DACs?

You do not need to have degree at Harward to figure ot that something is funky with the Total DACś specs. It should have almost 27bit if every bit is 6db.

And it is also interesting what thing is preventing the MSB V Diamond to max out ?

Can someone with intel describe this for us?

 

[arnaud]

You do not need to have degree at Harward to figure ot that something is funky with the Total DACś specs. It should have almost 27bit if every bit is 6db.

And it is also interesting what thing is preventing the MSB V Diamond to max out ?

Can someone with intel describe this for us?

I am no EE engineer but I think this is all about what figures you're looking at. These SNR plots are time averaged, aren't they? If so, with a 24 bit ladder array and proper noise shaping / dithering, I don't see what is so funky about the figures Vincent is getting.
 
That is unless this degree in Harward I am missing is preventing me to understand the facts 

 

 

[negura]

 
I am no EE engineer but I think this is all about what figures you're looking at. These SNR plots are time averaged, aren't they? If so, with a 24 bit ladder array and proper noise shaping / dithering, I don't see what is so funky about the figures Vincent is getting.
 
That is unless this degree in Harward I am missing is preventing me to understand the facts 

 

 
So just to understand: Are you saying they are not linked to effective resolution/ENOB? I am genuinely trying to get some clarity over the discrepancies.