One DAC To Rule Them All.... - Introducing the Kassandra from Aries Cerat

Believe High Fidelity

[Industry Expert]
Nov 19, 2015
1,665
318
355
Hutto TX
ibelieveinhifi.com
I am sure Trinity, Light Harmonic, dCS and MSB (elect) will disagree…such is life, choice is good.

Of course. In the brief time I have introduced both the product and brand I would like to point out that all things have been revealed.

You know the Kassandra doesn't use the defacto Delta Sigma and in fact uses the most precise ladder dacs available ....32 times.

You know the output is single ended, not with an off the shelf tube most would use, but a super tube with a circuit design to match it

You have seen the power supplies and every other non off the shelf part of the internals without even having to ask for it.

You will find even the proprietary technology that is used is explained in this thread

Nothing is hidden, which is something I cannot say for the others mentioned. Full disclosure, not taboo about how great our design is for reasons unexplained....
 
  • Like
Reactions: Zappadaddy

wisnon

Well-Known Member
Dec 12, 2011
3,525
635
1,200
Of course. In the brief time I have introduced both the product and brand I would like to point out that all things have been revealed.

You know the Kassandra doesn't use the defacto Delta Sigma and in fact uses the most precise ladder dacs available ....32 times.

You know the output is single ended, not with an off the shelf tube most would use, but a super tube with a circuit design to match it

You have seen the power supplies and every other non off the shelf part of the internals without even having to ask for it.

You will find even the proprietary technology that is used is explained in this thread

Nothing is hidden, which is something I cannot say for the others mentioned. Full disclosure, not taboo about how great our design is for reasons unexplained....

AD1865 is multi-bit (chip), but not resistor ladder AFAIK. Not Delta-sigma but not resistor ladder either.

MSB, Trinity, dCS dont use delta sigma. The first and last are proprietary Ladder or RING technology. For DSD, Lampizator does not use a Dac chip at all with ladder tech for PCM and Chord DAVE uses FPGA tech with 164K tap length in its proprietary WTA filter and now claims not to decimate DSD when playing back. Rob claims unblelievable noise floor and its noise shaper with performance of -350db!!! MSB select claims to have no output section, but drives the amp directly from the many banks of resistor PCBs in parallel (summation). Trinity goes to great length to select the best BB1704 multibit chips and then does analog oversampling with very expensive clocks (lianotec).

Rob's own words from Head-fi (Chord Dave):

"The perception of depth is a weird phenomena and something we take for granted. I am on holiday in Catalan, Spain and yesterday visited the monastery at Montserrat. We went for a walk, and was about a mile out viewing the monastery; the bells peeled out. The perception of depth was stunning, shut ones eyes and you can hear the bells a mile away with amazing accuracy.

Then we were lucky enough to hear the choir in the basilica. I was 150 feet away, and again, shut ones eyes and the sound was 150 feet away. It was glorious.

But the amazing thing is how the brain manages to compute depth from very tiny cues and get it to such accuracy and we take it entirely for granted.

So far early versions of Dave has been shown at a few shows and listeners have reported back about how unusual the depth perception is with Dave. Now this is due to the DAC resolving accuracy of very small signals - for some reason any small non-linearity of small signals upsets the brains ability to determine depth. What is curious is that there seems to be no limit to how accurate the linearity needs to be; Dave's noise shapers are accurate to -350 dB and this was the performance required by depth perception. Indeed, the brain may be sensitive to even smaller levels, but 350 dB is the best I can do with current FPGA's. But if you had said 2 years ago that one would need 350 dB performance from a noise shaper to get proper depth perception I would have said you were completely mad, as this is ridiculously small levels. But I have done thousands of depth listening tests, and always came to the same conclusion - very very small errors are significant. No doubt the sound science brigade will be on my back about this; but sound science is about observation not pet theories; and the observations are saying that something very weird is going on about depth perception (something which our understanding of how the brain achieves this level of accuracy is very limited).

Getting back to Dave and classical music. Sure classical music is not close miked, and so perceiving depth would be beneficial to that genre. But depth is often added in recordings by adding reverb. Also, its about small signal linearity which is useful for detail resolution as well as depth perception. But because Dave's depth perception is so much deeper than other DAC's its easy to latch onto that aspect of performance - its about four times deeper with recordings that have good depth than Hugo for example. But there is a lot more to Dave than just depth.

On the design of Dave, I was at one point improving smoothness and warmth. It got richer and darker, almost to the point where it sounded too dark and smooth - transients were starting to sound soft. Now this aspect was based on solid engineering, that of improving noise floor modulation, so I knew it was more transparent for certain. But it was sounding too rich and dark. Now that's OK - nobody knows what a perfect DAC sounds like (neutral just means average really), so I was prepared to live with it. But then I discovered what Hugo's ability to resolve the timing of notes - the perception of instruments starting and stopping - came from, and once found it I could maximise it due to the size available on the FPGA.

By maximising the timing aspect, I got Dave to sound much faster, tighter and leaner - the perfect remedy to something sounding too soft and smooth. Now again this performance is engineering based and is fundamentally more accurate and transparent.

So what does this mean? Well against Hugo, Dave has much more depth, which you can easily perceive and you don't need to do an AB test. If you hear depth that is deeper than anything ever heard from an audio system then you know something is better. But with Dave being at the same time faster tighter and more dynamic and also richer and darker than Hugo, then you need to do an AB test to perceive that aspect.



And another point - depth is really useful for AV, a lot of depth is recorded onto films and it really makes a big difference to enjoyment having a huge sound stage."
================

Feedback from a fan who posted at CA:

Mid-way through the evening there was a short presentation by Rob Watt, the independent designer Chord use for their digital devices. Dave employs a 28nm Xilinx FPGA that gives RW 166 DSP cores with which to do interpolation, filtering, etc. Alas, I don't have any copies of the slides, but I do remember that 17th order noise shaping was used and there was a very interesting configuration on the output amp that I hadn't seen before. RW is in the "it's most important to get the transient timing right" camp, also following the path other UK manufacturers hold to which is that the ear can discern timing differences equivalent to 250khz sampling rates, and it's this that's important in hi-fi reproduction, more so than frequency range. With Dave, Rob has produced a device that has some of the lowest noise figures I've ever seen, with a basic noise floor down at -150 db and an extremely clean-looking -127db A weighted result with a 1 kHz signal at 2.5 v output (from memory, and again wish I had the slides). RW was adamant that removing music modulations appearing in the sound floor is crucial, as it's this that greatly contributes to opening up a 3D level of imaging. RW is also working on an A to D box for Chord that runs at 768khz and so Dave also offers that sampling rate too, the hope being that, a bit like MQA, there's the chance of really pulling together the entire recording-to-reproduction chain, offering the listener a chance to hear what the performer (or recording engineer, at least!) really wanted to have conveyed.
============================

Rob Watts answer to me about DAVE DSD:

The DSD filtering is much more complex that with Hugo, simply because I have more space on the FPGA. There is almost no limit to how much filtering you need to do to remove the HF distortion and noise that the DSD noise shaper generates. Out of band noise creates intermodulation distortion with analogue electronics, if the noise is random then the distortion is random, so you then get noise floor modulation. The brain is very sensitive to noise floor modulation, and perceives it as a brightness to the sound, so reducing it makes it sound smoother and darker. There is virtually no limit to how small noise floor modulation needs to be, so any trace of HF and RF noise is audible.

Just to give you an idea how sensitive RF/HF noise is, I have a HF filter option with Dave. This is a 60 kHz filter, and is used to remove the ADC noise shaper noise from 192k, DXD and 384k recordings. Now with DXD recordings it works a treat, its much smoother and darker. But oddly, you can hear the benefit with 44.1k - even CD sounds smoother. Now that was very unexpected, as the WTA filter has at worst 120 dB of filtering, and typically (for music sources) is 140dB. So HF noise is below -140 dB for CD, but an extra stage of filtering gives a subjective benefit - and its down to the reduction of HF/RF noise. Just to illustrate the problem DSD64 is -20dB down at 100kHz.

With Dave, I had the freedom to make much more elaborate filters - and ended up with a non decimating design that used 64 DSP cores alone. But it would not fit with the large PCM WTA filters, so Dave has two modes - PCM+ and DSD+. PCM+ is 164k taps, and has Hugo DSD 64/128/256 filter. DSD+ is 82k taps PCM and has the better DSD filter, and also supports DSD512. You can use DSD+ with video, as the group delay is smaller, although most projectors have enough delay to match PCM+ delay.

Rob

===================
More Tech:

Sound quality has been an interesting story. Now my intention is to make things as transparent as possible, as the gap from listening to a live orchestra to hearing it in the home is vast. And if every link in the chain was perfectly transparent, it would not be a vast difference when listening to live unamplified sound.

Problem is nobody has heard a perfectly transparent DAC. Neutral really means average tonal balance and peoples idea of transparent is often etched and over emphasised details, giving a hyper sound quality but no musicality. And its musicality (defined as having emotional and involving music) that is the purpose of this - not something that sounds impressive for which only one recording actually works.

Half way through the development, I had made substantial improvements from Hugo. But it was getting smoother, and darker in tonal balance - to the point where I knew people would say its far too smooth. I felt it was too dark - I like the way Hugo's presentation can sound very fast and sharp but at the same time smooth. But Dave's extreme smoothness was not a colouration or error. Some designers add distortion (2nd harmonic) to the bass to make it sound phat, warm and soft. Or they use capacitors that create low frequency errors that fatten up the bass too. Dave's refinement was not this; it was based on reduction of distortion (most notably noise floor modulation - Dave's noise floor is at -180dB and is completely unchanged whether there is a signal or not - no other non Chord DAC has come even remotely close to this level of performance). So I knew for sure that Dave's smoothness was true transparency, not some tuning artefact or distortion. And I would have been prepared to leave it at that, as I mentioned earlier nobody has a clue as to what a perfect DAC would sound like.

But then another extremely important milestone was reached. One of the puzzles I needed to find out was the mystery of why Hugo had the timing performance it has. By timing I mean Hugo has this ability to enable one to perceive the starting and stopping of notes with remarkable ease, previous DAC's I designed sounded a bit soft, muddled and confused by comparison. Also, you could follow rhythm with ease, and perceive the interplay of different instruments - the way that musicians "talk" to one another. It also had the ability to paint a wide range of timbre and perception of timbre is down to timing too. All of these qualities is down to the perception of timing, so from an engineering POV where was this performance coming from as it was entirely unexpected. Now Hugo was the first project that had a major upgrade in all of the code that goes into the FPGA - this was a 6 year development. So there was something I had done that unexpectedly improved the perception of timing - and I know it was not the WTA filter, as I had previously designed 32,000 tap filters and they did not sound like Hugo.

What I had done was to improve noise floor modulation by using much better 2048 FS interpolation filters. And these proved to be the magic sauce of Hugo's ability to reproduce the starting and stopping of notes. In short it was the time domain accuracy that was important. Before Dave, my target was time domain accuracy of less than 1uS when reconstructing the missing bits from one sample to another. This number is based on the inter aural delay that we know is accurate to about 4uS, but its not based on how the brain processes ear data - science has no idea how the brain separates instruments out and calculates placement information. Things we completely take for granted. But the reality is that there is no real limit to how accurate the time domain needs to be - instead of targeting uS it needs to be nS. Once that was appreciated, I could radically redesign the interpolation filters to more accurately reconstruct the continuous time analogue signal that was present at the ADC.

Boy did this make a big difference. The perception of timing was much faster and brighter than before; instruments were more real with more sense of individual power. Now when it becomes easier to hear the starting and stopping of notes, things become much brighter and more immediate. Its like adjusting the focus on a lens; poor focus means the edges are blurred and soft and rounded; putting it back into focus gives edges much more immediacy and become more vivid. Now like the situation with Dave's refinement, the perception of speed and timing is not something artificial; a WTA filter running at 16 FS (1.4uS resolution) is fundamentally less time domain accurate than a filter running at 256 FS (88nS resolution).

What was nice was that the balance was restored (and I stress I am not trying to create a particular sound) in that Dave went from sounding really rich and dark and impossibly smooth to something much faster and immediate. Now Dave still sounds rich and smooth; but it is also capable (when the occasion demands) of sounding very fast and sharp.

The next aspect that had a lot of attention was depth perception. Now this is an area with audio I have had a considerable interest over; listen to a organ and a choir in a cathedral. If its 100 feet away, it sounds exactly 100 feet away. But play it back on your audio and its at best only a few feet back. Why is audio so bad at depth perception? The problem is down to being able to accurately reproduce very small signals. As soon as you get small errors in amplitude for small signals, the brain can no longer get a handle on depth and so does not give you an impression of depth - you just get an organ sounding very ambient but with no sensation of depth - flat as a pancake. With DAC's there is a fundamental problem of small signal amplitude accuracy. R2R DAC's have enormous problems with this, as you simply can't get the resistors to match - and it shows with very large measurable errors. Delta sigma have better accuracy, but they still suffer. We can get a handle on a noise shapers performance by running simulations of the noise shaper. Simulations aren't a pretence, in case of Verilog simulation its actually the real performance of the noise shaper. The best noise shapers in high end silicon DAC's are at best 140dB devices - that is, if you run a Verilog simulation with say a 0dB 1kHz signal you will get -140dB THD and noise. This means the noise shapers noise floor is at best at -160 or -170dB. Now this means that a signal at -170 dB is completely lost - any signals below the noise shaper noise floor is eliminated. So small signals are attenuated or altered by the noise shaper - and this affects depth perception.

Now with Hugo, the noise shapers are 200 dB performance (that's about a thousand times more resolution than conventional high end DAC's), and this is one of the reasons why Hugo has the reputation for reproducing good depth. Now I thought that 200 dB performance was good enough - but you can never make assumptions as to whether something will make a difference unless you do rigorous carefully controlled listening tests. With Dave I had enough space to do more advanced noise shaping. So I increased it to 220 dB - and depth got better. Then 240 dB - again you could easily hear an improvement in depth. This process continued over several months with radical re-designs. At 330dB it was still sounding deeper - eventually I got 350 dB performance and you could still hear an improvement from 330dB. Now this is completely extraordinary; and if somebody told me that 350dB performance was necessary a few years ago I would have said they were nuts as it implies that there is no limit to how accurate small signals need to be - that the brain can detect any size of error no matter how small. Utterly amazing - but I can only report what I can easily and consistently hear.

At the end of this process, I could hear depth that is around four times deeper than with Hugo. Now that is only if depth is recorded, close miked sounds with no reverb still sound flat as it is supposed to.

So Dave has been a very exciting and rewarding project for me. I got to find discover more about DAC's in the last year than the previous decade. Most importantly I understand how digital degrades depth perception and upsets the perception of timing - and these things are absolutely key for trying to re-create the musicality of live unamplified music. In short I have vastly underestimated how sensitive the ear/brain is to extraordinarily small timing and amplitude errors.

So how does Dave sound? I am (quite rightly) not supposed to talk about the sound of my own projects but I guess I can hint at it. Its like a Hugo but with a lot more depth - that's the first thing one notices. For the first time ever, I am now starting to get depth to sound more like real life - some recordings can sound really spooky as to how deep they can go. And this aspect is (mostly) down to the noise shaper performance. But to get 350 dB on the noise shaper was pretty involved - I ended up with 17th order, with 46 integrators. Its so huge the noise shaper alone would not fit on Hugo's FPGA. But the improvements in noise floor modulation, and the timing accuracy, also have a very big influence, but the first thing you should notice is the uncanny depth.

Rob
 

FrantzM

Member Sponsor & WBF Founding Member
Apr 20, 2010
6,455
29
405
I sincerely wonder what is left to discuss if this product is so great..
I hope it can be upgraded ... You should hear what some $500 DAC can do today to 10 K DAC of only two years ago ...
 

dallasjustice

Member Sponsor
Apr 12, 2011
2,067
8
0
Dallas, Texas

wisnon

Well-Known Member
Dec 12, 2011
3,525
635
1,200
BelieveHiFi:

I do believe the Cerat will sound wonderful as its extremely overbuilt and that normally comes out in SQ, but the competitors are not resting on their laurels or using off the shelf tech. Dont fall into that trap. People are really innovating or trying new implementations of the tried and true.

Having heard the DAVE at a show and the MCB Select and spending lots of time with the Trinity, as well as owning the top Lampizator Dac, I can tell you that none of them are "ordinary". At this level it does come down to taste and price.

Again, choice is good.
 
Last edited:

dallasjustice

Member Sponsor
Apr 12, 2011
2,067
8
0
Dallas, Texas
It plays it truncated to 18 bits. The best ladder plays at 21 bits mx and most at about 19, AFAIK.

I am not talking about resolution. Maybe the DAC can deliver incredible measurements. I am talking about the actual data.

I believe you are correct that the AD1865 will truncate the 24 bit datastream down to 18bits. Do the gawdy s/n ratio measurements even matter beyond 108db for this DAC? There are many DACs which can deliver up to 21 bits of resolution (126db) while properly handling 24 bit datastream.
 

Believe High Fidelity

[Industry Expert]
Nov 19, 2015
1,665
318
355
Hutto TX
ibelieveinhifi.com
Blelieve HiFi:

I do believe the Cerat will sound wonderful as its extremely overbuilt and that normally comes out in SQ, but the competitors are not resting on their laurels or using off the shelf tech. Dont fall into that trap. People are really innovating or trying new implementations of the tried and true.

Having heard the DAVE at a show and the MCB Select and spending lots of time with the Trinity, as well as owning the top Lampizator Dac, I can tell you that none of them are "ordinary". At this level it does come down to taste and price.

Again, choice is good.

Agreed and my commentary is focused to affirm our commitment to be transparent both in sound and in business practices. Having heard most of the good stuff out there I can attest that after 3 years of searching Aries Cerat is going change what can be expected from conventional designs. I do hope that a lot of folks make it to CES to hold me accountable to that..
 

DaveyF

Well-Known Member
Jul 31, 2010
6,129
181
458
La Jolla, Calif USA
Having just come from hearing the new MSB Diamond Select DAC ( $90K), I do think that there are some great new DAC's coming onto the market. Pricing structure needs to be much higher though. $300-$500K for a top line DAC would be more appropriate. So, we can see how this new DAC is really a TOTAL steal at only $35K. Even the reference model at $85K is actually given away!!
 

wisnon

Well-Known Member
Dec 12, 2011
3,525
635
1,200
Having just come from hearing the new MSB Diamond Select DAC ( $90K), I do think that there are some great new DAC's coming onto the market. Pricing structure needs to be much higher though. $300-$500K for a top line DAC would be more appropriate. So, we can see how this new DAC is really a TOTAL steal at only $35K. Even the reference model at $85K is actually given away!!
LoL

The $13K Chord Dave sound GREAT on PCM, but I didnt hear it on DSD (which so far hs not been great on ANY FPGA Dac I have heard). Honestly, from the brief dem of the Dave and later the Select, i cant see where the price difference is in PCM SQ. In terms of size yes, but not sound.
 

esldude

New Member
Having just come from hearing the new MSB Diamond Select DAC ( $90K), I do think that there are some great new DAC's coming onto the market. Pricing structure needs to be much higher though. $300-$500K for a top line DAC would be more appropriate. So, we can see how this new DAC is really a TOTAL steal at only $35K. Even the reference model at $85K is actually given away!!

I agree. Until we get in the upper half of six figure pricing we can't expect a truly high quality listening experience to emerge. If someone were to be able to do that for $85 k it would be big news. I however being the skeptical type just find it a bridge too far. $85k just doesn't provide enough resources for a top end result even with extremely clever engineering.
 

Johnny Vinyl

Member Sponsor & WBF Founding Member
May 16, 2010
8,570
51
38
Calgary, AB
I agree. Until we get in the upper half of six figure pricing we can't expect a truly high quality listening experience to emerge. If someone were to be able to do that for $85 k it would be big news. I however being the skeptical type just find it a bridge too far. $85k just doesn't provide enough resources for a top end result even with extremely clever engineering.
I'm fukced! LOL!
 

Believe High Fidelity

[Industry Expert]
Nov 19, 2015
1,665
318
355
Hutto TX
ibelieveinhifi.com
I agree. Until we get in the upper half of six figure pricing we can't expect a truly high quality listening experience to emerge. If someone were to be able to do that for $85 k it would be big news. I however being the skeptical type just find it a bridge too far. $85k just doesn't provide enough resources for a top end result even with extremely clever engineering.

You're absolutely right. However nothing a little sparing use of gold can't cure :)

500k here we come
 

FrantzM

Member Sponsor & WBF Founding Member
Apr 20, 2010
6,455
29
405
Having just come from hearing the new MSB Diamond Select DAC ( $90K), I do think that there are some great new DAC's coming onto the market. Pricing structure needs to be much higher though. $300-$500K for a top line DAC would be more appropriate. So, we can see how this new DAC is really a TOTAL steal at only $35K. Even the reference model at $85K is actually given away!!

Frankly I have to wonder how $85K for a HiFi component is a steal ... I know we , audiophiles live in an alternate universe where hyperbole reigns and words have little meaning but man! I guess it's because no one aside from the OP has heard it .. Once we do then we will know .. just like porn ( you know what it is once you see it) or "natural" ...
 

Believe High Fidelity

[Industry Expert]
Nov 19, 2015
1,665
318
355
Hutto TX
ibelieveinhifi.com
Frankly I have to wonder how $85K for a HiFi component is a steal ... I know we , audiophiles live in an alternate universe where hyperbole reigns and words have little meaning but man! I guess it's because no one aside from the OP has heard it .. Once we do then we will know .. just like porn ( you know what it is once you see it) or "natural" ...

It's not a steal. Poor choice of words, but even for 85k it is remarkable. That you can quote me on.

Whether the sonic improvements merit that much money spent is relative
 
  • Like
Reactions: Zappadaddy

bonzo75

Member Sponsor
Feb 26, 2014
22,435
13,467
2,710
London
And uranium, would cost a bomb
 

Audiophile Bill

Well-Known Member
Mar 23, 2015
4,293
4,093
675
We demand the whole thing wired in Graphene with platinum output transformers :p
 

microstrip

VIP/Donor
May 30, 2010
20,806
4,698
2,790
Portugal
(...) Rob's own words from Head-fi (Chord Dave):

"The perception of depth is a weird phenomena and something we take for granted. I am on holiday in Catalan, Spain and yesterday visited the monastery at Montserrat. We went for a walk, and was about a mile out viewing the monastery; the bells peeled out. The perception of depth was stunning, shut ones eyes and you can hear the bells a mile away with amazing accuracy.

Then we were lucky enough to hear the choir in the basilica. I was 150 feet away, and again, shut ones eyes and the sound was 150 feet away. It was glorious.

But the amazing thing is how the brain manages to compute depth from very tiny cues and get it to such accuracy and we take it entirely for granted.
(...)

Time for re-listen to an old favorite. I have both the LP and the CD, and it is one of those cases when the LP version sounds much better than the CD. I have to listen to it in the Chord Dave ...
 

Attachments

  • a1.jpg
    a1.jpg
    145.2 KB · Views: 422

About us

  • What’s Best Forum is THE forum for high end audio, product reviews, advice and sharing experiences on the best of everything else. This is THE place where audiophiles and audio companies discuss vintage, contemporary and new audio products, music servers, music streamers, computer audio, digital-to-analog converters, turntables, phono stages, cartridges, reel-to-reel tape machines, speakers, headphones and tube and solid-state amplification. Founded in 2010 What’s Best Forum invites intelligent and courteous people of all interests and backgrounds to describe and discuss the best of everything. From beginners to life-long hobbyists to industry professionals, we enjoy learning about new things and meeting new people, and participating in spirited debates.

Quick Navigation

User Menu

Steve Williams
Site Founder | Site Owner | Administrator
Ron Resnick
Site Co-Owner | Administrator
Julian (The Fixer)
Website Build | Marketing Managersing