CHORD ELECTRONICS DAVE

[romaz]

Still in addition to the $690, there's the HQPlayer cost...

Between the mR and HQ Player, IMO, with the DAVE, the mR is easily the more important of the two. At $150, HQ Player is worthy of its asking price but if I had to buy a separate PC just to run it, personally, I would just stay with the mR.

I have no intention of using Roon so I would have to figure out how to make Tidal and JRiver work with HQPlayer...

If you decide you want an mR, you can use JRiver without having to use Roon but as far as I'm aware, you have to use Roon to stream Tidal.

...I would now have to leave (a computer) on all the time and then figure out how to remote control the desktop so that it can send the correct signals from HQPlayer to microRendu...Sounds like too much work as I still have thousands of hours of music I just want to listen to instead of fiddling around with my computer audio setup for now.

Using JRiver, you would still need to have some computer running somewhere. As for remote control, because of the integration between Roon and HQP, you just need a remote for Roon (iPad or Android) and Roon will seamlessly control HQP.

 

[romaz]

Can someone explain why applying dither is able to leave the music "bit perfect"?

I have wondered this as well. The best that I can surmise is that dithering, as a noiseshaper, isn't truncatung or changing any bits. I don't know.

 

[brightonjel]

This question is a big one and I'm not sure I can remember enough to do it justice, but here goes anyway!
 
Quantization errors occur in a digital system when reducing the number of bits being used to represent something; this is of course particularly true in going from an analog signal (continuous) to a digital one (represented by a number with an accuracy expressed by some finite number of bits used to store it, let's say 16).  In practice, these rounding errors are a) likely correlated with the signal being processed, which is bad, and b) of a power that's more noticeable the smaller the signal being considered is, which is also bad.
 
Adding dither - i.e. a computer-generated number, at a level of + or - 1/2 LSB, say, randomizes those errors, improving the perceived quality of the resulting signal.  In addition, the mathematical function for generating the dither signal can itself be tuned in order to shape the noise profile of the combined signal such that, to the human ear at least, what noise there now is should be less noticeable. Is the resulting stream "bit perfect" at source?  No, but from the analogue perspective then no digital representation can be completely perfect over time because quantization always introduces errors, albeit vanishingly small as the bit depth used increases (adding one more bit of resolution reduces noise power by -6db, i.e. 1/4x).  In addition, even the best quality D/A converters only manage about 20 bits of resolution in any case.
 
The term "bit perfect" can be applied to subsets of the processing chain, e.g. going from Roon via Ethernet to uRendu to a DAC. This can either be a bit perfect chain, where the same bits in always produce the same bits out, or, if you resample, change the volume, or otherwise process things along the way, then not.  For example, applying a digital volume correction, perhaps for volume leveling, somewhere along the way will change the bits even if it isn't itself a dithered operation, so now it won't be!
 
Hope I haven't just confused things further ... it really is a fascinating - but complex - topic.

 

[Rob Watts]

  Quote:
 
 
I have just spent last night and this morning listening and comparing Roon direct to DAVE vs Roon to HQ Player to DAVE and have been listening to a variety of 16/44 PCM, 24/192 PCM, DXD, DSD128 and DSD256 files that are very familiar to me.  I have also tested Tidal streaming through HQ Player, something I did not realize was possible.  This is my first time to experience HQ Player with my DAVE, Esau, and so this experience is brand new to me.  It's been a long time since I've used Audirvana and so I cannot make any comments on how it might directly compare.
 
Here is my schematic that another user suggested and that Andrew Gillis has endorsed.  Obviously, in my case, substitute my DAVE for the PS Audio DirectStream.  HQ Player is running on my Windows machine, Roon Server is running on my sonicTransporter and the microRendu is running in Roon/HQ Player/NAA mode:
 

 
Here are my early listening observations:
 
PCM files:
 
With no upsampling/filtering and no dithering, Roon alone vs Roon/HQP sound identical.  Even in my system that I consider to be very resolving with all of my HF cables, I cannot tell a difference.
 
With no upsampling/filtering but with TPDF dithering enabled, there is a noticeable difference in favor of Roon/HQP.  Both sound very good but with TPDF dithering, my 16/44 files come to life more.  Transitions are more authoritative and there is better layering.  There appears to be more air around voices and instruments also.  I am hearing no downside with HQP in the chain and the signal being fed to the DAVE remains bit-perfect.  The partnering between Roon/HQP/microRendu is seamless.  Obviously, much testing and configuring have been done by Roon, HQP and Sonore to enable this seamless compatibility.  The delta is not huge but there is no need to blind test as the difference is fairly clear.
 
With upsampling/filtering turned on, it sounds good but not as good as with it off.  There is a definite loss of depth.  DAVE clearly upsamples better than HQP even when I upsample to 768kHz PCM or DSD7 256+FS.
 
DSD files:
 
With upsampling/filtering off and dithering off, once again, I hear no difference.
 
With upsampling/filtering off and dithering set to TPDF, once again, Roon/HQP sounds a bit better.
 
With upsampling/filtering on, once again, it sounds slightly worse. I even transcoded to PCM 768kHz since DAVE is a superior PCM DAC and native DSD sounds better.  Definitely, leave your file in its native format. 
 
There are other dither modes I will need to test but TPDF seems to be a good way to go.
 
Based on these early observations, it would appear to me that HQ Player used as a player without any oversampling or filtering but with TPDF dithering is superior to Roon by itself (to my ears) with the DAVE.  I would agree that upsampling should be left to the DAVE and I find no advantage to transcoding from one format to another.  HQ Player without Roon suffers from a clunky and unattractive interface and would be much less enjoyable from a usability standpoint without Roon.  Roon + HQ Player + mR in NAA mode seems to result in a definite improvement.  Unfortunately for me, another expense.
 
As a side note, if you wish to use HQP with Roon and your mR, you cannot use ASIO or Wasapi as a back end, you will need to use "Network Audio Adapter".  This means you can't utilize Chord's native DSD option but DoP seems to work just fine.
 

 
 

 
"With no upsampling/filtering but with TPDF dithering enabled, there is a noticeable difference in favor of Roon/HQP.  Both sound very good but with TPDF dithering, my 16/44 files come to life more.  Transitions are more authoritative and there is better layering.  There appears to be more air around voices and instruments also.  I am hearing no downside with HQP in the chain and the signal being fed to the DAVE remains bit-perfect."
 
I am somewhat surprised by this, as this sound quality is exactly what I used to hear with dither being added to data and doing nothing else. I used to do this with the PDM1024 way back in 1992 (probably the first time newly invented FPGA's were being employed in audio). The reason dither works in this application is that it randomises the data, and in the case of the PDM1024 I used multiple DSD256 noise shapers, each fed with its own different random noise. So each DSD sequence was randomised, and you got better small signal accuracy by summing in the analogue domain these different sequences. Technically, it improves small signal accuracy, so it sounds smoother, with better depth and layering. 
 
But with the invention of pulse array, and with 5 bit noise shaping at 2048 FS, I completely solved the problem of small signal accuracy (at least as far as measurements were concerned) and employing extra dither no longer made an improvement in sound quality. So I dropped the procedure, as it did not make a difference to the SQ.
 
On Chord's Blu CD transport, I still keep dither as an option - but this is merely intended for poorer DAC's that have small signal linearity problems.
 
Of course with Dave, I have even better small signal performance, so it makes it less likely that adding dither will improve this aspect. But perhaps other things are going on, that is not small signal related. I can add dither to Dave and see if it improves SQ at all. If it does improve SQ significantly we will do a code update for Dave.
 
Note that adding dither means the data is changed and is now no longer bit perfect. Not that I have any problems with doing that so long as its a change with only SQ benefits.
 
 
"DSD files:
 
With upsampling/filtering off and dithering off, once again, I hear no difference.
 
With upsampling/filtering off and dithering set to TPDF, once again, Roon/HQP sounds a bit better."
 
This is perplexing, as you can't add dither to a DSD file. You would need to convert it to PCM, then add dither, then re-noise shape it to get back to DSD. That process would sound poor, as a DSD noise shaping function is certainly not transparent. I suspect that something else is going on, as dithering set to TPDF does not mean dithering is actually working all the time - it may be only on truncation process. Since no actual processing is going on, no truncation, and so perhaps no dither would then be added. It's certainly impossible to add dither to DSD directly without re-noise shaping.
 
Rob

 

[x RELIC x]

...

Of course with Dave, I have even better small signal performance, so it makes it less likely that adding dither will improve this aspect. But perhaps other things are going on, that is not small signal related. I can add dither to Dave and see if it improves SQ at all. If it does improve SQ significantly we will do a code update for Dave.

...

Presuming if the DAVE is sent back to Chord, right?

Rob, did you get a chance to compare a 'fresh' DAVE for burn-in comparison to the one you've been using as you mentioned earlier?

 

[Rob Watts]

Code updates (if needed) will need a return to a trained distributor or Chord. I won't do it for a minor change.
 
The new Dave turned up on Friday, but I am away on holiday for the school half term this week. Should be able to listen to it next week.
 
Rob

 

[x RELIC x]

Code updates (if needed) will need a return to a trained distributor or Chord. I won't do it for a minor change.

The new Dave turned up on Friday, but I am away on holiday for the school half term this week. Should be able to listen to it next week.

Rob

Excellent, enjoy your holiday.

 

[ecwl]

I use parametric EQ in JRiver and JRiver is set to TPDF dithering so JRiver always dithers with TPDF. But I believe if you're playing a 16/44 file without DSP and you set your JRiver audio driver output bit depth to 24-bit or 32-bit, JRiver will automatically apply the dither you chose as the dithering method at 24/32-bit. In fact, now I wonder if I should change my output bit depth from 24-bit to 32-bit to see if it works and if there's a sonic difference.

 

[bigfatpaulie]

  I'd love to hear what you think with the DAVE and your DNA Stratus, Paul.

 
You and me both!  I'm going to let the DAVE settle in and get to know that on its own first.  I suspect they will pair very nicely. 

 

[bigfatpaulie]

  Code updates (if needed) will need a return to a trained distributor or Chord. I won't do it for a minor change.
 
The new Dave turned up on Friday, but I am away on holiday for the school half term this week. Should be able to listen to it next week.
 
Rob

 
The new DAVE?
 
I'm sorry - what?

 

[lovethatsound]

The new DAVE?

I'm sorry - what?

it's just a brand new Dave from chord,it's the same Dave has everyone has go;Rob just wanted a new one.

 

[ecwl]

The new DAVE?

I'm sorry - what?

it's just a brand new Dave from chord,it's the same Dave has everyone has go;Rob just wanted a new one.

It's not a new model. It's a new set of DAVEs off the manufacturing line. Rob Watts was going to compare this newly manufactured DAVE with his older one to see how long burn-in takes. The innards and codes are completely identical.


Sent from my iPhone using Tapatalk

 

[bigfatpaulie]

That's good to know - thank you.
 
It's official, I hate DSD.  The DAVE is the second DAC I've had a nearly impossible time getting to work consistently.  I'm I alone in my issues with it?  For the record I'm using Foobar2000.

 

[JaZZ]

  That's good to know - thank you.
 
It's official, I hate DSD.  The DAVE is the second DAC I've had a nearly impossible time getting to work consistently.  I'm I alone in my issues with it?  For the record I'm using Foobar2000.

 
Hi Paul
 
I have the same issue with the Hugo, but not with DAVE... or so I thought! But it turns out that also with DAVE I don't reliably get native DSD from .dff files, whereas .dsf files are o.k. Sometimes foobar takes the selected DFF decoder frequency for playing back PCM.
 
You may want to have a look at the recent posts in the Hugo thread addressing DSD problems with foobar and Chord drivers.

 

[bigfatpaulie]

   
Hi Paul
 
I have the same issue with the Hugo, but not with DAVE... or so I thought! But it turns out that also with DAVE I don't reliably get native DSD from .dff files, whereas .dsf files are o.k. Sometimes foobar takes the selected DFF decoder frequency for playing back PCM.
 
You may want to have a look at the recent posts in the Hugo thread addressing DSD problems with foobar and Chord drivers.

 
 
Thanks.  I went through that (and 3 other similar 'how toos') to no avail.  I, like you, can't double click on the ASIO drivers.  Also, if I select foo_dsd_asio I get "Unrecoverable playback error: The process receiving or sending data has terminated" when trying to play anything.  I did download a file from that thread and now it seems I can play DSF files if I select "ASIO: ASIO Chord 1.05", not the suggested foobar device. 
 
Oddly enough, my DAVE says 352.8kHz with a DSD64 file on a dark blue background - should it not be pinky/purple?