[Mython]

  Just to offer some perspective on the high perceived price of Blu Mk 2 and it's M-scaler, here are some competing prices from other products:
 
Esoteric P-02X SACD/CD Transport (no DAC.  This is the latest model that was just released.  Yes, this also does SACD but it has no scaler) - $23,000  
 
dCS Upsampler 2.0 (upsamples to 2x DSD or 24/352.8.  Also has ethernet input and can stream Tidal or Spotify.  Does not include a CD transport at this price) - $22,000
 
Blu Mk 2 incorporates both with a scaler that performs well beyond the capability of the dCS scaler for under $10k USD.  I think Chord has produced a product that easily overachieves at its price point.  Considering MSB's DACs are at about 6,000 TAPS, I'm sure some MSB owners are looking at Blu Mk 2 as the steal of the century given that MSB charges $11,500 for their own Signature CD transport with power base which doesn't include a scaler.  Considering the step upgrade that I heard from DAVE to 1 million taps is greater than what I heard from Hugo to DAVE, I believe the price is very fair.

 
Even though Blu & DAVE are currently way out of my price bracket, I am nonetheless inclined to agree with you, Romaz; everything is, indeed, relative.

 

[m0nster]

@Jazz I remember that you live in Switzerland too. Do you know where I could get Chord stands for the Dave?
 
Since there may be a 2nd Chord product on its way in the foreseeable future the stand would look pretty nice... Help would be appreciated

 

[miko64]

Try k55.ch they are extremely helpful
 
Michael

 

[m0nster]

 
Try k55.ch they are extremely helpful
 
Michael

Even more people from Switzerland - sweet!
 
Thank you for your help. I haven't found in on the homepage of k55.ch, but I could ask them by e-mail.
 
Michael too

 

[JaZZ]

  Yes all file formats benefit from WTA filtering, and 44.1 to 384 kHz all have exactly the same amount of processing irrespective of sample rate. Of course, the filter is used differently with 48k and 96k - they run in different modes - but the coefficient bank is the same, and the DSP does exactly the same number of multiply and add, so the output resolution in terms of timing accuracy is identical. Later I will formally try to do some listening tests on different resolution sources and feedback on the results. My superficial impression is that older redbook recordings seem to benefit more - some of the problems from 1960's Decca for example are very much reduced.
 
Yes the 10A peak is at 1V so we are looking at only 10W maximum. What I have always done is to have the core regulator as close as possible to the FPGA, so it is normally mounted directly underneath the FPGA. This means that the PSU inductor and ground loop is kept as small as possible. This means the conversion of noisy signal correlated 10A/1V to clean 2A/5V is constrained to as small an area as possible, so we have the least possible signal correlated noise on the ground plane, which would then upset the pulse array elements. When I used to use linear regulators, I had measured problems due to this, and each place and route on the FPGA would measure and sound different. But since using switchers, and controlling the current loops, I think I have pretty much solved this issue - at least with Mojo/Hugo/Dave as I now see no measurement problems or differences in sound quality with differing place and route. Here is one example of supposedly nasty switching regulators sounding and measuring very much better than linear regulators.
 
The first version of Blu 2 had 6A regulators, and this was with the 512,000 taps. But using the first 1,000,000 taps design meant the regulator collapsed, so it would not work. Now I had expected this, and a 12A regulator version was already being built and arrived soon after the completion o the 1M design.
 
Rob

 
Now I am curious! In what way do 192 or 384 kHz recordings benefit from low-pass filtering with the sophisticated interpolation algorithm of the WTA filter? Or any low-pass filtering at all? I mean, the signal curve seems virtually perfect for the audio bandwidth even without it, apart from dispensable ultra-high-frequency noise from the sampling frequency which could be removed by simple analogue filters...

 

[JaZZ]

  @Jazz I remember that you live in Switzerland too. Do you know where I could get Chord stands for the Dave?
 
Since there may be a 2nd Chord product on its way in the foreseeable future the stand would look pretty nice... Help would be appreciated

 
Yes, as mentioned by Michael  K55 (my preferred audio dealer) would be the place, or the Swiss distributor directly.

 

[Rob Watts]

 

  Yes all file formats benefit from WTA filtering, and 44.1 to 384 kHz all have exactly the same amount of processing irrespective of sample rate. Of course, the filter is used differently with 48k and 96k - they run in different modes - but the coefficient bank is the same, and the DSP does exactly the same number of multiply and add, so the output resolution in terms of timing accuracy is identical. Later I will formally try to do some listening tests on different resolution sources and feedback on the results. My superficial impression is that older redbook recordings seem to benefit more - some of the problems from 1960's Decca for example are very much reduced.
 
Yes the 10A peak is at 1V so we are looking at only 10W maximum. What I have always done is to have the core regulator as close as possible to the FPGA, so it is normally mounted directly underneath the FPGA. This means that the PSU inductor and ground loop is kept as small as possible. This means the conversion of noisy signal correlated 10A/1V to clean 2A/5V is constrained to as small an area as possible, so we have the least possible signal correlated noise on the ground plane, which would then upset the pulse array elements. When I used to use linear regulators, I had measured problems due to this, and each place and route on the FPGA would measure and sound different. But since using switchers, and controlling the current loops, I think I have pretty much solved this issue - at least with Mojo/Hugo/Dave as I now see no measurement problems or differences in sound quality with differing place and route. Here is one example of supposedly nasty switching regulators sounding and measuring very much better than linear regulators.
 
The first version of Blu 2 had 6A regulators, and this was with the 512,000 taps. But using the first 1,000,000 taps design meant the regulator collapsed, so it would not work. Now I had expected this, and a 12A regulator version was already being built and arrived soon after the completion o the 1M design.
 
Rob

 
Now I am curious! In what way do 192 or 384 kHz recordings benefit from low-pass filtering with the sophisticated interpolation algorithm of the WTA filter? Or any low-pass filtering at all? I mean, the signal curve seems virtually perfect for the audio bandwidth even without it, apart from dispensable ultra-high-frequency noise from the sampling frequency which could be removed by simple analogue filters...

 
OK just to illustrate. Before Dave, I used to WTA filter to 16FS (768 kHz) and then use third order IIR filters to get me to 2048FS or 104 MHz. As part of the Dave program, I then partially replaced the the IIR filters with a 16FS to 256FS WTA FIR filter, and it sounded dramatically better - particularly in the ability to perceive the starting and stopping of notes. but if you look at the actual difference in timing accuracy between a 16>256FS WTA> 3rd order IIR>2048 FS path and a 16> 3rd order IIR>2048 FS path we are talking about very small differences indeed - but it is so very audible. I used to think 1uS errors would be inaudible, now I take the view that any timing error (where a transient is slightly earlier or later against the rest of the signal) of any size, no matter how small, is important. This sensitivity is very, very strange, but it clearly exists.
 
Now when people get their hands on the Hugo 2, you can try for yourself. As the white to orange filter, or the green to red filter settings selects the 256FS filter (white is 256FS WTA, orange is 16FS WTA), so people will be able to hear for themselves. And for me it is not a small change in SQ.
 
So in the case of 256FS we can say that to 88 nS resolution, we have close to ideal sinc recovery of the original timing. But a 384 kHz signal innately only has a poor 2.6 uS timing resolution, so you can see if 88 nS is important, then 2.6 us will be grossly audible. That's why 384 kHz recordings do indeed benefit from the WTA filters ability to reconstruct timing down to a resolution of 88 nS.
 
Remember that to perfectly reconstruct the missing bits from one sample to the next sample one must use a infinite oversampling sinc function FIR filter. If you do not do this, the output timing of transients will be incorrect - what the surprise is is how sensitive the ear brain is to these errors. I wonder too how much further we need to go in timing resolution and accuracy before the brain can no longer perceive the difference.
 
Hope this clarifies, but it is a complex subject.
 
Rob

 

[Crgreen]

You should work as a salesman for Chord. I feel bound to point out that just because other, possibly lesser, products cost more does not necessarily mean that any of them respeesent value for money.

 

[JaZZ]

 

Spoiler: Warning: Spoiler!

Quote:

 

  Yes all file formats benefit from WTA filtering, and 44.1 to 384 kHz all have exactly the same amount of processing irrespective of sample rate. Of course, the filter is used differently with 48k and 96k - they run in different modes - but the coefficient bank is the same, and the DSP does exactly the same number of multiply and add, so the output resolution in terms of timing accuracy is identical. Later I will formally try to do some listening tests on different resolution sources and feedback on the results. My superficial impression is that older redbook recordings seem to benefit more - some of the problems from 1960's Decca for example are very much reduced.
 
Yes the 10A peak is at 1V so we are looking at only 10W maximum. What I have always done is to have the core regulator as close as possible to the FPGA, so it is normally mounted directly underneath the FPGA. This means that the PSU inductor and ground loop is kept as small as possible. This means the conversion of noisy signal correlated 10A/1V to clean 2A/5V is constrained to as small an area as possible, so we have the least possible signal correlated noise on the ground plane, which would then upset the pulse array elements. When I used to use linear regulators, I had measured problems due to this, and each place and route on the FPGA would measure and sound different. But since using switchers, and controlling the current loops, I think I have pretty much solved this issue - at least with Mojo/Hugo/Dave as I now see no measurement problems or differences in sound quality with differing place and route. Here is one example of supposedly nasty switching regulators sounding and measuring very much better than linear regulators.
 
The first version of Blu 2 had 6A regulators, and this was with the 512,000 taps. But using the first 1,000,000 taps design meant the regulator collapsed, so it would not work. Now I had expected this, and a 12A regulator version was already being built and arrived soon after the completion o the 1M design.
 
Rob

 
Now I am curious! In what way do 192 or 384 kHz recordings benefit from low-pass filtering with the sophisticated interpolation algorithm of the WTA filter? Or any low-pass filtering at all? I mean, the signal curve seems virtually perfect for the audio bandwidth even without it, apart from dispensable ultra-high-frequency noise from the sampling frequency which could be removed by simple analogue filters...

 
OK just to illustrate. Before Dave, I used to WTA filter to 16FS (768 kHz) and then use third order IIR filters to get me to 2048FS or 104 MHz. As part of the Dave program, I then partially replaced the the IIR filters with a 16FS to 256FS WTA FIR filter, and it sounded dramatically better - particularly in the ability to perceive the starting and stopping of notes. but if you look at the actual difference in timing accuracy between a 16>256FS WTA> 3rd order IIR>2048 FS path and a 16> 3rd order IIR>2048 FS path we are talking about very small differences indeed - but it is so very audible. I used to think 1uS errors would be inaudible, now I take the view that any timing error (where a transient is slightly earlier or later against the rest of the signal) of any size, no matter how small, is important. This sensitivity is very, very strange, but it clearly exists.
 
Now when people get their hands on the Hugo 2, you can try for yourself. As the white to orange filter, or the green to red filter settings selects the 256FS filter (white is 256FS WTA, orange is 16FS WTA), so people will be able to hear for themselves. And for me it is not a small change in SQ.
 
So in the case of 256FS we can say that to 88 nS resolution, we have close to ideal sinc recovery of the original timing. But a 384 kHz signal innately only has a poor 2.6 uS timing resolution, so you can see if 88 nS is important, then 2.6 us will be grossly audible. That's why 384 kHz recordings do indeed benefit from the WTA filters ability to reconstruct timing down to a resolution of 88 nS.
 
Remember that to perfectly reconstruct the missing bits from one sample to the next sample one must use a infinite oversampling sinc function FIR filter. If you do not do this, the output timing of transients will be incorrect - what the surprise is is how sensitive the ear brain is to these errors. I wonder too how much further we need to go in timing resolution and accuracy before the brain can no longer perceive the difference.
 
Hope this clarifies, but it is a complex subject.
 
Rob

 
Thanks a lot, Rob! That explains it for me. But it's indeed incredible that even 384 kHz recordings benefit from a careful interpolation algorithm for filtering quantization noise and preserving timing (= transient) accuracy. However, I take it that low sampling rates will benefit clearly more – up to equal sound quality, if I interpret some of your previous statements correctly, at least with tap counts approaching infinite numbers enough for sensitive human ears. Do you think the million taps are already enough to make 44.1 kHz recordings virtually indistinguishable from 192 or 384 kHz recordings?

 

[Rob Watts]

No I don't think 384/24 is indistinguishable with 48/24 or  certainly not 44.1/16 - but my feeling is that the standard is much less important than the recording method, so in that sense M scaler is a great equalizer - although this was true with Dave too, certainly with the ability to enjoy older recordings. For example, Decca recordings from 1960 to pre dolby 1967 redbook recordings do things that modern 192/24 kHz recordings are incapable of doing - notably being able to record speed and impact of real instruments. Modern recordings are smooth and refined but seem incapable of reproducing timbre variations and raw impact like the classic 1960's recordings. So simple microphone and short signal paths with sound optimized custom built mixers and amps are more important than 44.1 or 192. That said, my older recosrdings benefit much more from M scaler - which is what one would expect.
 
Future recordings with Davina will allow us to evaluate exactly what the losses are in sample rate and reconstruction. With the tremendous change M scaler offers clearly the next question is how much further can we take this, and Davina will tell us exactly how close 1M taps to ideal actually is.
 
Rob

 

[romaz]

You should work as a salesman for Chord.

 
The price is indeed high and so this is how I had to look at it to justify this purchase for myself.

I feel bound to point out that just because other, possibly lesser, products cost more does not necessarily mean that any of them respeesent value for money.

 
You are absolutely correct but imagine what would happen if Blu Mk 2 was half the price. A much cheaper non-Chord DAC could potentially reach or exceed DAVE's performance for half the price of DAVE which would be quite upsetting for DAVE owners who have no plans to upgrade. I believe there is a balance here that must be respected for Chord not to upset its customers. As all things go, eventually, with time and advancement of technology, 1 million taps will likely one day trickle down to a Mojo-type device at Mojo-type pricing and Hugo 2 is a good example of that progression and so for those that are willing to wait, I'm sure that Chord will one day deliver. What is exciting for me is to see how quickly technology is advancing, not just for Chord but for other brands as well and so it is the consumer that wins here. Rob had previously shared with me that when he first idealized the concept of 1 million TAPS while at university years ago, FPGAs didn't even exist and so 1 million TAPS had always been more of an idealized concept rather than a practical thought. To actually have achieved 1 million TAPS is not only a momentous occasion for Rob, I'm sure, but really a momentous occasion for all of us who have embraced digital as our preferred medium. As the old guard who will continue to cling to their turntables for sentimental reasons die off, I can't see analog media as having any real reason to exist in the future.

 

[Silvertone4]

Rob, (or anyone that attended the CES demos)

Are there improvements or differences in the bass region of Blu 2 vs Dave that can be attributed to the 1M taps?

Thanks in advance

 

[Rob Watts]

Actually my position in the early 1980's was that we needed 16 bit sinc accuracy, and hence a million taps before digital would sound perfect, or close to perfect. And that it would be completely and utterly impossible to ever get to a million taps. So I was very, very wrong about the latter - but - and here is the fun part - given how much better the M scaler sounds - my intuition that 1M taps was essential for good digital sound has turned out right.
 
Rob

 

[romaz]

Rob, (or anyone that attended the CES demos)

Are there improvements or differences in the bass region of Blu 2 vs Dave that can be attributed to the 1M taps?

Thanks in advance

Yes, it was quite evident.  Going from no M-scaler to 1 million TAPS, the bass was noticeably more thunderous.  This was one of the things that was immediately identifiable.

 

[lovethatsound]

I don't have a problem with how much the blu mk2 costs,I don't have a problem with Rob Watts,but at the moment i have got a problem with Chord.That problem being i was told the original blu would be able to be upgraded to the new spec . I've looked at the blu mk2 on the chord website,and basically it looks more or less the same,even at the back,and i can see No reason why the original blu can not be upgraded.I really think Chord should look at this issue again and make good on what they have said.☺