[JaZZ]

  Watts has also programmed his WTA filter to ignore test impulse patterns so it’s impossible to ‘see’ its pre/post ringing behaviour in the time domain...

 
@Rob Watts: What's the reason for this?

 

[ecwl]

I realize sometimes there's understanding something and then there's really understanding something. I've always known from what Rob Watts has said in this column that in order to allow the S/PDIF coaxial/BNC input of DAVE to accept 384kHz/24-bit signals, they are not galvanically isolated. I have a USB source and Toslink source. But I also needed room correction/parametric EQ for my video system so I use a miniDSP product to implement the DSP and connect it's coaxial output to Chord DAVE. I said to myself, even if there is some RF/ground noise getting into the DAVE when I'm playing music, I should just ignore it because I'm not going to hook and unhook the BNC cable every time I want to listen to music. Besides the miniDSP product already has isolation transformer on the output.

My dealer and I were talking about cables and so I changed some in my video system but I wasn't ready to test them yet so I went back to listening to music. I couldn't understand why there's improvement to the sound. So I put back the original cables and then just plug in and unplug the BNC connection to the DAVE, something I told myself I would never bother to do for convenience. Yup, that was what made the difference.

I thought I would share in case people are still using suboptimal BNC inputs into DAVE. If you can, switch over to Toslink. I'm guessing if you have good sources like Chord CD players, it probably doesn't matter. But if like me, your other sources aren't the greatest, you may want to not use the BNC inputs. Or unplug them for optimal musical enjoyment.

 

[paulchiu]

I realize sometimes there's understanding something and then there's really understanding something. I've always known from what Rob Watts has said in this column that in order to allow the S/PDIF coaxial/BNC input of DAVE to accept 384kHz/24-bit signals, they are not galvanically isolated. I have a USB source and Toslink source. But I also needed room correction/parametric EQ for my video system so I use a miniDSP product to implement the DSP and connect it's coaxial output to Chord DAVE. I said to myself, even if there is some RF/ground noise getting into the DAVE when I'm playing music, I should just ignore it because I'm not going to hook and unhook the BNC cable every time I want to listen to music. Besides the miniDSP product already has isolation transformer on the output.

My dealer and I were talking about cables and so I changed some in my video system but I wasn't ready to test them yet so I went back to listening to music. I couldn't understand why there's improvement to the sound. So I put back the original cables and then just plug in and unplug the BNC connection to the DAVE, something I told myself I would never bother to do for convenience. Yup, that was what made the difference.

I thought I would share in case people are still using suboptimal BNC inputs into DAVE. If you can, switch over to Toslink. I'm guessing if you have good sources like Chord CD players, it probably doesn't matter. But if like me, your other sources aren't the greatest, you may want to not use the BNC inputs. Or unplug them for optimal musical enjoyment.

 
Can you use USB input to Dave within the realm of your video/audio setup?  Shouldn't that be better sounding than optical?

 

[ecwl]

Can you use USB input to Dave within the realm of your video/audio setup?  Shouldn't that be better sounding than optical?

Audio is always through USB from my computer.
Video at night with headphones is from Oppo BDP-103 Toslink to the Chord DAVE and really the difference between USB and Toslink is really, really small. Or maybe it's really big and I just didn't notice because of the noise from the BNC input...

Video in general is through the Oppo BDP-103 (or PVR to Oppo) and its coaxial output to the miniDSP for parametric EQ and then to BNC into DAVE. That's presumably where the "noise" comes from. I know the miniDSP may be a noisier source than I had expected...

My computer was built to be under-powered to reduce noise output that would affect USB playback because every DAC I've played around with is affected by the USB source. DAVE obviously changed everything.

Yes, theoretically, I could build a completely new desktop to accept video and audio input from Oppo and then use the computer to output the video and audio to DAVE via USB. But that's even a more complex system. You're talking to someone who didn't want to unplug a BNC cable when listening to audio.

Ultimately, I also want to reassure people, the difference is clearly audible but I've lived with the DAVE for over a month and I have always been super ecastitic with its performance. Never did I say to myself, I wonder if DAVE could even sound better by unplugging the BNC input when I'm only listening to music. But now that I've tried it out, I thought I should let people know in case their BNC sources are also noisy. Obviously your mileage may vary...

 

[Jawed]

 
DAVE HiFi News Review! http://bit.ly/1U0mG1d

 

"Especially with archive recordings, the switch from PCM+ to DSD+ mode pays dividends in the sharpening-up of the sonic picture and instrumental timbres alike, and freeing just a shade more of that considerable dynamic ability."

Hmmm... well, that's controversial, if nothing else!

 

I believe he is merely stating that DSD material sounds best when in DSD+ mode, as opposed to PCM+ mode. I don't think it's meant as a comparison of PCM and DSD, per se.

 

[Rob Watts]

Quote:

  Watts has also programmed his WTA filter to ignore test impulse patterns so it’s impossible to ‘see’ its pre/post ringing behaviour in the time domain...

 
 

 

   
@Rob Watts: What's the reason for this?

The WTA algorithm actually uses my own windowing function of the ideal sinc impulse response. I took an awful lot of time (man years) with this, both in listening tests and in trying to understand what was going on - the understanding being used to allow me to try listening to different things. At the end of the day, the algorithm is fine tuned by listening tests, but you need understanding in order to change the critical parameters - in short knowing what those parameters are. Anyway, with the blu CD player, using an impulse response you will get all the coefficients to a 24 bit accuracy, so it would be easy to reverse engineer the WTA algorithm.
 
Frankly I should not have worried. This industry has its collective head in the sand. I have been publically talking about the importance of very long tap length filters for 17 years and still nobody else bothers about it.  
 
Rob

 

[ismewor]

Hi Rob.
 
Just purchased the DAVE, awesome device. and i would like to know which BNC can take the I2S input if possible?
 
 
Thanks,
Y

 

[Rob Watts]

Sorry but I2S is not supported as an input with Dave.
 
Rob

 

[rkt31]

I don't know much but does that mean that there is no pre or post ringing in Dave due to high tap length and wta algorithm ?

 

[JaZZ]

  Quote:

  Watts has also programmed his WTA filter to ignore test impulse patterns so it’s impossible to ‘see’ its pre/post ringing behaviour in the time domain...

 

  @Rob Watts: What's the reason for this?

The WTA algorithm actually uses my own windowing function of the ideal sinc impulse response. I took an awful lot of time (man years) with this, both in listening tests and in trying to understand what was going on - the understanding being used to allow me to try listening to different things. At the end of the day, the algorithm is fine tuned by listening tests, but you need understanding in order to change the critical parameters - in short knowing what those parameters are. Anyway, with the blu CD player, using an impulse response you will get all the coefficients to a 24 bit accuracy, so it would be easy to reverse engineer the WTA algorithm.
 
Frankly I should not have worried. This industry has its collective head in the sand. I have been publically talking about the importance of very long tap length filters for 17 years and still nobody else bothers about it.  
 
Rob

 
Don't get me wrong: I'm not worried in the least, since DAVE sounds perfect to me. But does that (the windowing function?) mean a single pulse doesn't appear on the oscillogram? How about rectangles?

 

[ecwl]

   
The WTA algorithm actually uses my own windowing function of the ideal sinc impulse response. I took an awful lot of time (man years) with this, both in listening tests and in trying to understand what was going on - the understanding being used to allow me to try listening to different things. At the end of the day, the algorithm is fine tuned by listening tests, but you need understanding in order to change the critical parameters - in short knowing what those parameters are. Anyway, with the blu CD player, using an impulse response you will get all the coefficients to a 24 bit accuracy, so it would be easy to reverse engineer the WTA algorithm.
 
Frankly I should not have worried. This industry has its collective head in the sand. I have been publically talking about the importance of very long tap length filters for 17 years and still nobody else bothers about it.  
 
Rob

 
Don't be discouraged. I do see some online people who use software upsampling like HQPlayer, XXHighend or Bug Head Emperor caring about tap length as they are very well aware of your work. So maybe it is good to protect your IP and not let them steal the WTA algorithm. It always sucks to be a great visionary. It takes a long time for regular folks to appreciate the greatness of the vision. I strongly suspect Chord Mojo is now truly bringing in a lot more converts and we lucky Chord DAVE owners are truly blessed by your latest achievement.

 

[Mython]

I realize sometimes there's understanding something and then there's really understanding something. I've always known from what Rob Watts has said in this column that in order to allow the S/PDIF coaxial/BNC input of DAVE to accept 384kHz/24-bit signals, they are not galvanically isolated. I have a USB source and Toslink source. But I also needed room correction/parametric EQ for my video system so I use a miniDSP product to implement the DSP and connect it's coaxial output to Chord DAVE. I said to myself, even if there is some RF/ground noise getting into the DAVE when I'm playing music, I should just ignore it because I'm not going to hook and unhook the BNC cable every time I want to listen to music. Besides the miniDSP product already has isolation transformer on the output.

My dealer and I were talking about cables and so I changed some in my video system but I wasn't ready to test them yet so I went back to listening to music. I couldn't understand why there's improvement to the sound. So I put back the original cables and then just plug in and unplug the BNC connection to the DAVE, something I told myself I would never bother to do for convenience. Yup, that was what made the difference.

I thought I would share in case people are still using suboptimal BNC inputs into DAVE. If you can, switch over to Toslink. I'm guessing if you have good sources like Chord CD players, it probably doesn't matter. But if like me, your other sources aren't the greatest, you may want to not use the BNC inputs. Or unplug them for optimal musical enjoyment.

 
I am well aware of the miniDSP product (and similar products), although I admit I have not personally heard them.
 
My reason for being aware of the miniDSP is because of my interest in Siegfried Linkwitz's LX521.3 dipole loudspeaker (now LX521.4 in digital-only X-over form)
 
However, as you will note from my earlier post (linked above), I would prefer to build the LX521.3 with 'analogue active crossover', rather than employing miniDSP for crossover purposes.
 
 
My concern is that the miniDSP, and others of the same ilk, introduce an ADC->DAC stage into the playback chain, over and above the existing ADC (recording of the original performance) and hi-fi DAC.
 
I can't imagine shelling-out for a high-end DAC (particularly one as accomplished as DAVE) and then introducing an ADC->DAC stage, with vanilla Cirrus DAC chips, or similar, (generally running at 24/96 in this application). On a theoretical level, at least, it seems to me to be unnecessarily undermining the purity of the end-result.
 
Am I overlooking something?

 

[Rob Watts]

I don't know much but does that mean that there is no pre or post ringing in Dave due to high tap length and wta algorithm ?

I actually posed this issue as a question on another thread, but did not get an answer.
 
The ideal response is a sinc impulse response, this means it will have an infinite amount of pre and post ringing. But a filter that has this response will perfectly reconstruct a bandwidth limited signal absolutely perfectly with no difference whatsoever, its just displaced in time.
 
So we have a paradox - the filter that has the most ringing will re-create the signal perfectly with no added ringing whatsoever - but clearly having an infinite amount of ringing is not reproducing the input signal (an impulse) perfectly. How do we explain this contradiction?
 
This is an incredibly important question as virtually the whole audio industry talks about the importance of no pre-ringing - but they have all got it completely and utterly wrong.
 
The answer to the conundrum is that an impulse response is an illegal signal  - it is not bandwidth limited as it has the same energy at FS/2 as at DC, being a completely flat frequency response - that's why the signal is used for frequency response measurements. But sampling theory absolutely requires bandwidth limited signals - that means at exactly FS/2 the signal level is zero. Indeed, in a properly designed ADC, there will be negligible output at FS/2, so an impulse will never be presented to a DAC using a music file.
 
So if you use an conventional illegal impulse response signal then the best filter will have the worst ringing; but using music, or a bandwidth limited impulse response, it will actually have the least possible difference from the original continuous analogue signal that was in the ADC - and of course will sound very much closer to the original analogue signal before it was sampled.
 
Rob

 

[ecwl]

   
I am well aware of the miniDSP product (and similar products), although I admit I have not personally heard them.
 
My reason for being aware of the miniDSP is because of my interest in Siegfried Linkwitz's LX521.3 dipole loudspeaker (now LX521.4 in digital-only X-over form)
 
However, as you will note from my earlier post (linked above), I would prefer to build the LX521.3 with 'analogue active crossover', rather than employing miniDSP for crossover purposes.
 
 
My concern is that the miniDSP, and others of the same ilk, introduce an ADC->DAC stage into the playback chain, over and above the existing ADC (recording of the original performance) and hi-fi DAC.
 
I can't imagine shelling-out for a high-end DAC (particularly one as accomplished as DAVE) and then introducing an ADC->DAC stage, with vanilla Cirrus DAC chips, or similar, (generally running at 24/96 in this application). On a theoretical level, at least, it seems to me to be unnecessarily undermining the purity of the end-result.
 
Am I overlooking something?

 
I don't use ADC. That would be crazy/idiotic as you pointed out. miniDSP nanoDigi 2x8 B takes a digital coaxial or Toslink signal, ASRC it to 24-bit/96kHz (suboptimal, I know, for DAVE) and then applies digital parametric EQ and then outputs it via coaxial. I also run parametric EQ off my music server/PC in JRiver before sending the USB signal to the DAVE. Unfortunately, I have some room acoustic issues that I did not want to correct physically for aesthetic reasons so I have to correct the acoustic issues using subtle parametric EQs.
 
I am also suspecting that most of the RF/ground noise is coming from my cable company's cable as the DAVE sound is improved if I simply disconnect the cable box HDMI cable from the rest of my system. I'm testing out various scenario to figure out where the major source of noise is coming from and addressing it shortly. In the mean time, listening to music means the BNC input would be unplugged from Chord DAVE.
 
As a total aside, I'm not familiar with Siegfried Linkwitz's LX521.3. However, if you really want to optimize everything and have unlimited resources, you should consider the following. Don't use analog active crossover. Output your digital coaxial/Toslink music source signal into the miniDSP nanoDigi 2x8 B or some other digital crossover device and then split the signal 3-way with crossover and EQ as appropriate and you'll get 3 digital S/PDIF coaxial output that you can send to 3 Chord DAVEs as your DACs and then you can hook up 3 amplifiers to the 3 Chord DAVEs to playback on the speakers.

 

[AndrewOld]

I actually posed this issue as a question on another thread, but did not get an answer.

The ideal response is a sinc impulse response, this means it will have an infinite amount of pre and post ringing. But a filter that has this response will perfectly reconstruct a bandwidth limited signal absolutely perfectly with no difference whatsoever, its just displaced in time.

So we have a paradox - the filter that has the most ringing will re-create the signal perfectly with no added ringing whatsoever - but clearly having an infinite amount of ringing is not reproducing the input signal (an impulse) perfectly. How do we explain this contradiction?

This is an incredibly important question as virtually the whole audio industry talks about the importance of no pre-ringing - but they have all got it completely and utterly wrong.

The answer to the conundrum is that an impulse response is an illegal signal  - it is not bandwidth limited as it has the same energy at FS/2 as at DC, being a completely flat frequency response - that's why the signal is used for frequency response measurements. But sampling theory absolutely requires bandwidth limited signals - that means at exactly FS/2 the signal level is zero. Indeed, in a properly designed ADC, there will be negligible output at FS/2, so an impulse will never be presented to a DAC using a music file.

So if you use an conventional illegal impulse response signal then the best filter will have the worst ringing; but using music, or a bandwidth limited impulse response, it will actually have the least possible difference from the original continuous analogue signal that was in the ADC - and of course will sound very much closer to the original analogue signal before it was sampled.

Rob

Thank you very much for this post Rob, it gives me some insight as to why your DACs are so special - I ordered a DAVE a couple of weeks ago, and am so looking forward to getting it. My Hugo and then my Hugo TT have given so much pleasure I just decided that some things are really worth spending money on. All the best.