CHORD ELECTRONICS DAVE

[JaZZ]

 

So Chord DAVE uses 164000 taps to upsample from 44.1kHz to 704kHz and then from 704kHz to 11.2MHz...

 
I think that's a misconception. After pondering back and forth what these infamous taps actually mean, I'm back to my original understanding (Rob, correct me if I'm wrong):
 
Taps define the complexity of the low-pass filter necessary for preventing aliasing. The high upsampling rate is necessary to enable such a digital filter at all (so upsampling doesn't happen by means of the taps). The basis for the high tap rate approach is the Nyquist-Shannon theorem that sais for the perfect reconstruction of a band-passed (= low-pass filtered) signal you just need double the sampling rate of the filter frequency. The crux of this formula is that it implies an infinite steepness of the indispensable anti-aliasing filter – at 22.05 kHz at the latest for a sampling rate of 44.1 kHz. The Blu 2 seems to get quite close to an ideal result:
 
Frequency response:               0 Hz (DC) – 20 kHz ± 0.0000001 dB
In-band ripple 0 Hz – 20 kHz:    ± 0.0000002 dB
 
That's quite an achievement! It hints to an extremely sharp filter at around 22 kHz.
 
And why is the sharp filter so important? Because of the ringing (= resonance) it implies. A low-pass filter imperatively causes delayed decay – the sharper, the more it looks like a ringing. This means bad transient reproduction – abrupt starts and stops aren't possible anymore, because sharp edges would require unlimited bandwidth. So the sharper and steeper the filter, the more pronounced/longer the ringing, but at the same time its frequency content will more and more be limited to the filter frequency. So an infinite sharpness and steepness guarantees that the (now infinite) ringing exclusively consists of the filter frequency, which is in the ultrasonic range in the case of the redbook format – whereas audible signal contents will end abruptly, so to speak. That's the timing accuracy Rob speaks of. (At least in my interpretation.)
 
Thought to the end, it's clear that higher sampling rates, such as 88.2 kHz and above, aren't that much dependent on the high tap count if at all, because the anti-aliasing filters are relatively far away from the audio band. That's why the greatest advantage of Chord's new-generation DACs lies in the «low-res» formats 44.1 and 48 kHz. A high enough tap count and its careful implementation could indeed make hi-res formats dispensable – also in the real world, not just theoretically.

 

[ecwl]

Good news about the external upsampler - how do you know that?!

It's in the slides of that link I provided. Rob Watts also talked about it in the audio.

 

[analogmusic]

10 AMPS current to run Blu 2. that's a crazy amount of processing power !!!!

 

[AndrewOld]

It's in the slides of that link I provided. Rob Watts also talked about it in the audio.

Oh yes, found it now, thanks!
 
"A separate WTM scaler for Dave is under development for other sources".
 
Excellent, really looking forward to that. Is it too much to hope it will incorporate some streaming technology pinched from the Poly? 

 

[romaz]

http://www.computeraudiophile.com/content/739-ces-2017-nice-surprises/

So computer Audiophile website talked about a few new products at CES but also took some photos of Rob Watts PowerPoint slides on Blu and Hugo 2. Basically, 1 million taps mean 10A to run on the FPGA chip which means it is too noisy to be part of DAVE. And there is a non-CD external upsampler in the pipeline for other sources.

This will be Davina. It will be both an ADC anf DDC and will have the same digital inputs as DAVE.

 

[Christer]

   
I think that's a misconception. After pondering back and forth what these infamous taps actually mean, I'm back to my original understanding (Rob, correct me if I'm wrong):
 
Taps define the complexity of the low-pass filter necessary for preventing aliasing. The high upsampling rate is necessary to enable such a digital filter at all (so upsampling doesn't happen by means of the taps). The basis for the high tap rate approach is the Nyquist-Shannon theorem that sais for the perfect reconstruction of a band-passed (= low-pass filtered) signal you just need double the sampling rate of the filter frequency. The crux of this formula is that it implies an infinite steepness of the indispensable anti-aliasing filter – at 22.05 kHz at the latest for a sampling rate of 44.1 kHz. The Blu 2 seems to get quite close to an ideal result:
 
Frequency response:               0 Hz (DC) – 20 kHz ± 0.0000001 dB
In-band ripple 0 Hz – 20 kHz:    ± 0.0000002 dB
 
That's quite an achievement! It hints to an extremely sharp filter at around 22 kHz.
 
And why is the sharp filter so important? Because of the ringing (= resonance) it implies. A low-pass filter imperatively causes delayed decay – the sharper, the more it looks like a ringing. This means bad transient reproduction – abrupt starts and stops aren't possible anymore, because sharp edges would require unlimited bandwidth. So the sharper and steeper the filter, the more pronounced/longer the ringing, but at the same time its frequency content will more and more be limited to the filter frequency. So an infinite sharpness and steepness guarantees that the (now infinite) ringing exclusively consists of the filter frequency, which is in the ultrasonic range in the case of the redbook format – whereas audible signal contents will end abruptly, so to speak. That's the timing accuracy Rob speaks of. (At least in my interpretation.)
 
Thought to the end, it's clear that higher sampling rates, such as 88.2 kHz and above, aren't that much dependent on the high tap count if at all, because the anti-aliasing filters are relatively far away from the audio band. That's why the greatest advantage of Chord's new-generation DACs lies in the «low-res» formats 44.1 and 48 kHz. A high enough tap count and its careful implementation could indeed make hi-res formats dispensable – also in the real world, not just theoretically.

Hello Jazz,
your final paragraph made me think of what some classical musicians/ listeners at What´s Best Forum reportedly had to say about DAVE. It sounded very good with 16/44.1 to them.  But  according to them, somewhat at the expense of higher res PCM formats and less good with DSD than native DSD  Dacs.And according to Computer Audiophile´s report from Las Vegas Rob Watts now claims that 16/44.1 is just as good as any hi res format.I have yet to hear that from any DAC with complex large scale symphonic or operatic music.
That is quite a claim considering nobody, absolutely nobody in the recording business records at anything lower than 24/48 in the classical genre which is the only one of any real interest to me at least.Moreover  His own future product DAVINA will supposedly record at 24/768 or similar rate. Why if 16/44,1 is  just as good ?
Most  classical labels record at 24/96 or higher PCM and some DSD 64 and 128 or even 256.
There are instruments in a symphony orchestra that have energy well into 30khz and higher. Some percussion reaches 100khz.
Do the new  Chord Dacs shave off everything above brickwall at 22khz?
Does the high f filter option on DAVE do that?
One of the surprises on my  own audition of DAVE was that some masterfile classical  material  I know very well both live in the halls and from the ADCs used at the sessions ,sounded slightly softer  and more "rounded" than the  real thing and raw playback at sessions.
 
 
I am a bit confused and surprised by the insistance on 44,1  and CDs by Chord when basically everyone else has long abandonded both.
But then again I know very little about digital theory.
And if HUGO 2 proves to sound better more resolved and more realistic on large  classical acoustic music than both Hugo  and the best newish portables from the competition I will upgrade to HUGO 2 asap,provided there is no cheaper competitor doing the same or even better for less  of my money.
I am after all, Mr Cheap  here.
 
I  also still suspect that one of the  few advantages of analogue done right as with direct cut LPs and the best tape recorders is that there is no similar bandwidth limiting or complex filtering interpolation tricks involved as with 16/44.1 or  timing issues as with low res pcm.

 

[Brushane]

Christer: I think you have to separate what is good for the studio producer is not the same as for the end consumer. There is a reason to record in hi res formats, since this gives you a whole lot of headroom later in the production process. Even for red book standard, it would be less than ideal to also record at 44.1 kHz.
 
I am happy there is a producer who acknowledges how good 44.1 kHz really is on a consumer level. I don't mind hi-res formats really, but I think most of the releases coming out are just BS to be honest. What really has an impact on sound quality is not hi-res formats, but how it was recorded to begin with. I'd take a well recorded 256 kbps VBR MP3 file over dynamically compressed hi-res 24 bit 96 kHz FLAC any day. The file compression is nothing compared to the problems we have today with dynamic compression.

 

[JaZZ]

 

Spoiler: Warning: Spoiler!

Quote:

  I think that's a misconception. After pondering back and forth what these infamous taps actually mean, I'm back to my original understanding (Rob, correct me if I'm wrong):
 
Taps define the complexity of the low-pass filter necessary for preventing aliasing. The high upsampling rate is necessary to enable such a digital filter at all (so upsampling doesn't happen by means of the taps). The basis for the high tap rate approach is the Nyquist-Shannon theorem that sais for the perfect reconstruction of a band-passed (= low-pass filtered) signal you just need double the sampling rate of the filter frequency. The crux of this formula is that it implies an infinite steepness of the indispensable anti-aliasing filter – at 22.05 kHz at the latest for a sampling rate of 44.1 kHz. The Blu 2 seems to get quite close to an ideal result:
 
Frequency response:               0 Hz (DC) – 20 kHz ± 0.0000001 dB
In-band ripple 0 Hz – 20 kHz:    ± 0.0000002 dB
 
That's quite an achievement! It hints to an extremely sharp filter at around 22 kHz.
 
And why is the sharp filter so important? Because of the ringing (= resonance) it implies. A low-pass filter imperatively causes delayed decay – the sharper, the more it looks like a ringing. This means bad transient reproduction – abrupt starts and stops aren't possible anymore, because sharp edges would require unlimited bandwidth. So the sharper and steeper the filter, the more pronounced/longer the ringing, but at the same time its frequency content will more and more be limited to the filter frequency. So an infinite sharpness and steepness guarantees that the (now infinite) ringing exclusively consists of the filter frequency, which is in the ultrasonic range in the case of the redbook format – whereas audible signal contents will end abruptly, so to speak. That's the timing accuracy Rob speaks of. (At least in my interpretation.)
 
Thought to the end, it's clear that higher sampling rates, such as 88.2 kHz and above, aren't that much dependent on the high tap count if at all, because the anti-aliasing filters are relatively far away from the audio band. That's why the greatest advantage of Chord's new-generation DACs lies in the «low-res» formats 44.1 and 48 kHz. A high enough tap count and its careful implementation could indeed make hi-res formats dispensable – also in the real world, not just theoretically.

Hello Jazz,
your final paragraph made me think of what some classical musicians/ listeners at What's Best Forum reportedly had to say about DAVE. It sounded very good with 16/44.1 to them.  But  according to them, somewhat at the expense of higher res PCM formats and less good with DSD than native DSD Dacs. And according to Computer Audiophile's report from Las Vegas Rob Watts now claims that 16/44.1 is just as good as any hi res format. I have yet to hear that from any DAC with complex large scale symphonic or operatic music.
That is quite a claim considering nobody, absolutely nobody in the recording business records at anything lower than 24/48 in the classical genre which is the only one of any real interest to me at least.Moreover  His own future product DAVINA will supposedly record at 24/768 or similar rate. Why if 16/44,1 is  just as good ?
Most  classical labels record at 24/96 or higher PCM and some DSD 64 and 128 or even 256.
There are instruments in a symphony orchestra that have energy well into 30khz and higher. Some percussion reaches 100khz.
Do the new  Chord Dacs shave off everything above brickwall at 22khz?
Does the high f filter option on DAVE do that?
One of the surprises on my  own audition of DAVE was that some masterfile classical  material  I know very well both live in the halls and from the ADCs used at the sessions ,sounded slightly softer  and more "rounded" than the  real thing and raw playback at sessions.

 
As to the bolded sentence: Maybe Rob's statement was addressing the Blu 2? I don't think he would have gone to one million taps if it wasn't audibly better than the DAVE's 164,000 taps.
 
Why record in hi-res nonetheless? Because worldwide music lovers with a DAVE or a DAVE plus a Blu 2 are an absolute minority. Moreover, for sound editing purposes it's always good to have some headroom. Moreover, «low-res» recordings require low-pass filtering before A/D conversion, which is just as problematic as it is on the playback side. But first of all because the equation of hi-res and low-res is far from being established and as mentioned above apparently limited to Chord's TOTL playback gear in the real world.
 
Yes, some musical instruments reach high into the ultrasonic range, but the ultrasonic content itself most likely isn't audible, at best the byproducts in the form of interferences, or it's even just about timing within the audio band, which is guaranteed by a sharp enough anti-aliasing filter (as described in my previous post).
 
No, the DAVE's HF filter is just an add-on primarily meant for DSD, but (to Rob's own surprise) it also works for low-res PCM in the form of subtle smoothing (not necessarily higher accuracy, though). I can reproduce your impression about the DAVE sounding slightly smoother than reality – at least it sounds smoother than my other DACs. But I like it, and it sounds fairly realistic to my ears. Now there's certainly no electronics component that's absolutely neutral, but after all the DAVE offers a non-digital sound that emulates live music passably well, better than any other source I know.

 

[rkt31]

10A at 12v is only 120w but at 220v is huge 2kw . so the exact power consumption power figure of blue mk2 is needed. now it is understandable why so many taps and hence so much power is kept separate . it is now clear that we may see a non CD upsampler too in future .

 

[JaZZ]

10A at 12v is only 120w but at 220v is huge 2kw . so the exact power consumption power figure of blue mk2 is needed. now it is understandable why so many taps and hence so much power is kept separate . it is now clear that we may see a non CD upsampler too in future .

 
That would be great, but 2 kW is a lot nonetheless. Imagine every second household consuming that much additional power! Also, the price would be high even without the CD transport function, so it's questionable if it would be a worthwile upgrade for DAVE owners.

 

[Christer]

  Christer: I think you have to separate what is good for the studio producer is not the same as for the end consumer. There is a reason to record in hi res formats, since this gives you a whole lot of headroom later in the production process. Even for red book standard, it would be less than ideal to also record at 44.1 kHz.
 
I am happy there is a producer who acknowledges how good 44.1 kHz really is on a consumer level. I don't mind hi-res formats really, but I think most of the releases coming out are just BS to be honest. What really has an impact on sound quality is not hi-res formats, but how it was recorded to begin with. I'd take a well recorded 256 kbps VBR MP3 file over dynamically compressed hi-res 24 bit 96 kHz FLAC any day. The file compression is nothing compared to the problems we have today with dynamic compression.

Hello Brushane,
I am  not interested in consumer level. I am  interested in the best possible SQ combined with the best possible simple miking recording technique and best played best  large scale and chamber art music /classical acoustic music performed in the best venues by the best artists reproduced as realistically and as close  to how it sounded live in the hall, as possible. And in my experience there should be as little  tampering or compression or post production involved as possible.
We live in an age when it both is and should be possible to hear recordings of live music without  unnecessary sacrifices of decimation and compression and mixing pult creations instead of how live music actually sounds in good hall. In my experience the lower the bits and res EVERYTHING ELSE EQUAL OF COURSE, the lower the realism.
Even with very good simple  miking I begin to quite  reliably hear the  typical  digital compromises from  24/48 and down.I have on quite a few occasions had the privilige to hear comparisons from DXD down to 24/48, 24/44,1 compared with live in the hall and mic feed.  And  on each occasion it sounds more digital the lower the res.
There should in other words be as  little  difference between what was heard in the hall and what is being repoduced on what you call consumer level as possible .
Hence  masterquality  should be the goal all the way imho.
Consumer level compromises should only apply for the Mp3 kids and  synthetic studio pop/rock crap  imho.
I am completely uninterested in what they do or don´t do.
There is virtually no low level detail reverb tails acoustic or other very  sofisticated timbral  and tonal and harmonic complexities  and finer nuances in those genres that hi res could reveal anyway. But there are tons of those things,in complex  densely scored large scale classical music.
Not only Mozart and Tchaikovsky or basically any other of the classical composers benefit from hi res but also many contemporary composers  like The Danish composer Norgard whose symphonies have been recorded both at 24/48 and 24/88.2 by DACAPO.
The 24/88.2recordings capture his complex music and dense textures audibly better and more realistically than the lower res ones.
And playing his music decimated  to 16/44.1 makes it sound strained and lifeless compared to the  24/88.2 masterfiles  via high quality electrostatic speakers as  just one  example of what I am talking about.
Both computers and hardrives and DACs are capable of handling and storing/playing back  hi res files with ease compared to when 16/44.1 was introduced and unfortunately became the standard for much too long imho.

 

[Brushane]

With all due respect, I think you misread what I meant with "consumer level". Consumer level is where you are at, unless you work with doing recordings yourself. Consumer level can be whatever format it is you are consuming, that is all: Be it whatever kids are using, or what you prefer to use.
 
My reply was in reference to this paragraph:

"That is quite a claim considering nobody, absolutely nobody in the recording business records at anything lower than 24/48 in the classical genre which is the only one of any real interest to me at least.Moreover  His own future product DAVINA will supposedly record at 24/768 or similar rate. Why if 16/44,1 is  just as good ?"

 
It looked to me as you misunderstood what has been said about 16/44.1 and it's greatness. Maybe you didn't, but this is what generated my reply.
 
All the best

 

[Sonic77]

http://www.computeraudiophile.com/content/739-ces-2017-nice-surprises/
 
So the Blu mk 2  is approximately $9,852.00
The Hugo 2 approximately $2,218.00
The Poly approximately $615.00
 
Using current conversion today.

 

[AndrewOld]

This will be Davina. It will be both an ADC anf DDC and will have the same digital inputs as DAVE.

Are you sure the Davina and the "separate WTM scaler for Dave for other sources" referred to above are the same thing?

Why would most people with a DAVE want an ADC any more than they want a cd spinner?

I do hope that that WTM scaler is separate from Davina.

 

[JaZZ]

 

Spoiler: Warning: Spoiler!

  Christer: I think you have to separate what is good for the studio producer is not the same as for the end consumer. There is a reason to record in hi res formats, since this gives you a whole lot of headroom later in the production process. Even for red book standard, it would be less than ideal to also record at 44.1 kHz.
 
I am happy there is a producer who acknowledges how good 44.1 kHz really is on a consumer level. I don't mind hi-res formats really, but I think most of the releases coming out are just BS to be honest. What really has an impact on sound quality is not hi-res formats, but how it was recorded to begin with. I'd take a well recorded 256 kbps VBR MP3 file over dynamically compressed hi-res 24 bit 96 kHz FLAC any day. The file compression is nothing compared to the problems we have today with dynamic compression.

Hello Brushane,
I am  not interested in consumer level. I am  interested in the best possible SQ combined with the best possible simple miking recording technique and best played best  large scale and chamber art music /classical acoustic music performed in the best venues by the best artists reproduced as realistically and as close  to how it sounded live in the hall, as possible. And in my experience there should be as little  tampering or compression or post production involved as possible.
We live in an age when it both is and should be possible to hear recordings of live music without  unnecessary sacrifices of decimation and compression and mixing pult creations instead of how live music actually sounds in good hall. In my experience the lower the bits and res EVERYTHING ELSE EQUAL OF COURSE, the lower the realism.
Even with very good simple  miking I begin to quite  reliably hear the  typical  digital compromises from  24/48 and down.I have on quite a few occasions had the privilige to hear comparisons from DXD down to 24/48, 24/44,1 compared with live in the hall and mic feed.  And  on each occasion it sounds more digital the lower the res.
There should in other words be as  little  difference between what was heard in the hall and what is being repoduced on what you call consumer level as possible .
Hence  masterquality  should be the goal all the way imho.
Consumer level compromises should only apply for the Mp3 kids and  synthetic studio pop/rock crap  imho.
I am completely uninterested in what they do or don´t do.
There is virtually no low level detail reverb tails acoustic or other very  sofisticated timbral  and tonal and harmonic complexities  and finer nuances in those genres that hi res could reveal anyway. But there are tons of those things,in complex  densely scored large scale classical music.
Not only Mozart and Tchaikovsky or basically any other of the classical composers benefit from hi res but also many contemporary composers  like The Danish composer Norgard whose symphonies have been recorded both at 24/48 and 24/88.2 by DACAPO.
The 24/88.2recordings capture his complex music and dense textures audibly better and more realistically than the lower res ones.
And playing his music decimated  to 16/44.1 makes it sound strained and lifeless compared to the  24/88.2 masterfiles  via high quality electrostatic speakers as  just one  example of what I am talking about.
Both computers and hardrives and DACs are capable of handling and storing/playing back  hi res files with ease compared to when 16/44.1 was introduced and unfortunately became the standard for much too long imho.

 
Christer, what I meant in my previous posts is that all your concerns are possibly addressed by a low-pass filter with one million taps (maybe you could even live with DAVE's «low-res» performance, as I do). So on «consumer level» (as explained by Brushane) redbook and the like will do.