[Stefano74]

Sorry for the delay. I can help with an R-1. I should be able to get some listening time Sunday when I get home.

Any word from Magna?

Have sent you PM.

 

[Chopin75]

Got the same answer here; additionally, he very much doubts this is only with R2R 1, and the same applies to all Audio-GD range (R28, R8, R7, etc.).

According to him the difference can only be the source (with active filters) or the DSD files used.

If I apply filtering in HQPlayer (e.g. native DSD 256 to "upsampled" DSD 256) the noise is gone. According to him this is the way to go, and SQ shall remain unchanged.

Yes, it makes the sound cleaner, but a bit more sterile, less air & atmosphere. The transparency is less, e.g. without filter you feel like your head is inside the piano sound board, with filter applied, there is a slight veil, the minute detail is gone, the timber of strings are less apparent. The whole point of DSD is to get these details with all the naturalness. PCM can be detailed but it is less natural. SO I guess resampling makes it more like PCM. I tried both embedded in Euphony Audio (linux) as well as via Mac OS. Mac is worse, quite easily tell the difference. It is more subtle with linux and one may not mind the slight change esp the background is now blacker with filter. There is annoying click/pop once resampling is applied as the track is changed even in the same album . I should ask hQplayer about that

 

[ItsAllInMyHead]

Quick update -

@Stefano74 was kind enough to point me to the files he was using. I'm still fiddling, but a few points seem clear so far:

• I hear what I think he's hearing with the tracks he's provided through both an R-1 and the updated R2R-11. I saw no point in moving the R-8 to test.
• It's different than the original noise I was experiencing with the R2R-11, which is now eliminated after the update and was never present on my R-1 or R-8.
• The same noise is not present through either an iFi Micro DSD BL (any filter setting) or a Sony PHA-2A.
• It does seem to increase in severity from DSD 64 => DSD 256 in the files Stefano provided.
• It is not present on other tracks I have. However, I have an extremely limited DSD library. If I can still find out how to download the tracks I have freely, I'll PM anyone links. They're all DSD 64.
• It IS present using DoP 1.0. This confirms my understanding of DoP. If anyone has experience with DoP eliminating this issue, please post.
• It is not present when converting on the fly to PCM using Roon.
• It is present whether passed straight from computer to DAC directly with USB or through a Singxer SU-1 (I2S over HDMI) either native or DoP. This essentially eliminates the Amanero card(s) as the root cause or any USB firmware related issues.

I don't think it affects anything related to this, but I'm using V3S firmware on the R-1 with the latest ASIO drivers for the Amanero cards.

I downloaded a few more files for fiddling. If anyone else has files that either exhibit or don't exhibit this issue that are freely downloadable, let me know.

I'll send Kingwa links to the test files and see if he's got any input. If anyone else has already contacted Audio-Gd about this, let me know. There's no need to bombard the inbox.

Thanks to Stefano again!!!

[Edited to add - much later] This noise is also present using a Mac/Audirvana. I originally hooked everything through PC to most closely emulate others' situations. So, it seems to be independent of platform, USB driver, USB firmware, DoP, A-GD DAC model, DAC firmware etc. It is a combination of some files and the A-GD DACs. An important emphasis to some may be that I have the updated R2R-11, and this particular issue is still present. So, if DSD is critical to your music enjoyment, and you don't use a software filter, or use files that might exhibit this behavior, I'd recommend caution.

 

[Chopin75]

Quick update -

@Stefano74 was kind enough to point me to the files he was using. I'm still fiddling, but a few points seem clear so far:

• I hear what I think he's hearing with the tracks he's provided through both an R-1 and the updated R2R-11. I saw no point in moving the R-8 to test.
• It's different than the original noise I was experiencing with the R2R-11, which is now eliminated after the update and was never present on my R-1 or R-8.
• The same noise is not present through either an iFi Micro DSD BL (any filter setting) or a Sony PHA-2A.
• It does seem to increase in severity from DSD 64 => DSD 256 in the files Stefano provided.
• It is not present on other tracks I have. However, I have an extremely limited DSD library. If I can still find out how to download the tracks I have freely, I'll PM anyone links. They're all DSD 64.
• It IS present using DoP 1.0. This confirms my understanding of DoP. If anyone has experience with DoP eliminating this issue, please post.
• It is not present when converting on the fly to PCM using Roon.
• It is present whether passed straight from computer to DAC directly with USB or through a Singxer SU-1 (I2S over HDMI) either native or DoP. This essentially eliminates the Amanero card(s) as the root cause or any USB firmware related issues.

I don't think it affects anything related to this, but I'm using V3S firmware on the R-1 with the latest ASIO drivers for the Amanero cards.

I downloaded a few more files for fiddling. If anyone else has files that either exhibit or don't exhibit this issue that are freely downloadable, let me know.

I'll send Kingwa links to the test files and see if he's got any input. If anyone else has already contacted Audio-Gd about this, let me know. There's no need to bombard the inbox.

Thanks to Stefano again!!!

[Edited to add - much later] This noise is also present using a Mac/Audirvana. I originally hooked everything through PC to most closely emulate others' situations. So, it seems to be independent of platform, USB driver, USB firmware, DoP, A-GD DAC model, DAC firmware etc. It is a combination of some files and the A-GD DACs. An important emphasis to some may be that I have the updated R2R-11, and this particular issue is still present. So, if DSD is critical to your music enjoyment, and you don't use a software filter, or use files that might exhibit this behavior, I'd recommend caution.

Thanks for sharing your experience. I gather the noise level of the updated R2R-11 and R2R1 are about the same now? I doubt the latest R-8, or R-7 have better performance with DSD then. My experience is similar to yours. Software digital filter can help - using HQplayer, resampling DSD without up sampling can cut down the noise to almost inaudible level. There is perhaps slight loss in detail and "air" and ambience, it is a bit like digital remastering of analogue recording where the original analogue is more organic but noisier but the remastered is cleaner but more sterile. I don't find this a great solution. The audio-gd's chapels DSD processing has more simple digital filter, according to Magna (I think that is what they meant??!) The signa delta chips have more aggressive filtering, hence cleaner sound. I would like to try the latest ESS 9038 chop DAC to compare (if I can borrow one somewhere!). I think DSD DAC needs excellent filtering and in the Audio-gd it is the digital filter that is defective. Hopefully Kingwa can upgrade this in future, even if this means a higher cost.

 

[gLer]

Just remember in all of this that R2R dacs were never designed or intended to be used with DSD, they are PCM native. That means DSD dacs need to have dedicated native DSD decoders installed in the processing path to support native DSD, and I imagine this is where Audio-gd’s DSD quality in their R2R dacs takes a hit. Delta sigma chips like the Sabre 9038 were designed to decode DSD natively, with all the necessary filters in place, so will naturally be flawless with DSD content. Just something to keep in mind.

 

[Mr.Mantas]

Just remember in all of this that R2R dacs were never designed or intended to be used with DSD, they are PCM native. That means DSD dacs need to have dedicated native DSD decoders installed in the processing path to support native DSD, and I imagine this is where Audio-gd’s DSD quality in their R2R dacs takes a hit. Delta sigma chips like the Sabre 9038 were designed to decode DSD natively, with all the necessary filters in place, so will naturally be flawless with DSD content. Just something to keep in mind.

What are your thoughts about new HOLO AUDIO Spring II R2R dac? They claim that up to DSD1024 is top notch on it:
"DSD is natively supported this R2R Discrete DAC. In fact there are two separate DAC's, one does DSD and one PCM. Completely separate to achieve highest results. Holo Audio supports DSD natively on R2R DAC"


https://www.magnahifi.com/en/webshop/product/holo-audio-spring2-l1

I wonder does Kingwa has to make some significant hardware changes in order to to his dacs flawless with DSD or its enough to make firmware update?

 

[ItsAllInMyHead]

Thanks for sharing your experience. I gather the noise level of the updated R2R-11 and R2R1 are about the same now?

[SNIP]

I doubt the latest R-8, or R-7 have better performance with DSD then. Hopefully Kingwa can upgrade this in future...

Any time. This is a great community. I am still very curious about all of this, and I'm learning as I go.

I have an R-1. Stefano has the R2R-1. Yes, the performance for DSD seems similar if not identical now between the R-1 and the R2R-11 for DSD for the purposes of this specific "noise" with a few specific tracks.

I sent Kingwa an email with an explanation and a link to the test files Stefano sent me. I am not sure of forum rules re: posting to sites, but the links are from a pretty common test bench file site. He's just released another DAC it seems (the AS-1 "Ultra Distortion") so he's probably pretty busy. He may want to change the name.

I don't think it is against the rules to advise a Google Search for "Free DSD Test Bench", and you'll find the site. If anyone can shed some light on if it's within policy to name the site, I'd happily post so everyone can hear what we're talking about.

Caution to all others out there - don't use my experience as the only guideline. Before Stefano sent me some links to new tracks, I had a grand total of 3 DSD tracks in my library. All three of those played (and still play) "as intended" to what I can hear. So, I never understood what everyone was talking about until this morning. I had tried converting on the computer from PCM to DSD to see what all the fuss was about, and never experienced the issue. So it's not all DSD. However, I've only tried the PCM to DSD with Audirvana, and that automatically applies filtering. I listen to almost all PCM (and mostly Red Book)... so this is more of an experiment for me than anything practical.

I've also reached out to a friend with some expertise to see if he can shed light on the D2D conversion and/or the digital information remaining in the files from this site. I'd like to personally understand how that could contribute. As mentioned, my grand total of (now 4) other tracks from other sources play as intended to my ears. This is a pretty noticeable thing vs. what I was originally experiencing with the R2R-11. I don't really want to buy more DSD music to try out.. I'm not really all that interested.... so if folks have more free file sites for downloads that show this problem (or don't) I'm happy to check them out. So far, the only files I have that have the issue are from one site, so that doesn't provide any definitive results.

I've been screwing with this all morning now... and here's what's really interesting (to me).

The site we're using has songs in various formats both PCM and DSD. If I take their PCM files and play them PCM, they play what I'd assume to be "as intended". If I software convert the exact same song in files ranging from 16/44.1 up to 24/352.8 to DSD from 64 to 256 on the fly using Audirvana and the 7th order filters (A, B, or C), it sounds like the DSD file downloaded (with the noise). If I software convert other PCM files that I have in my library to DSD... no noise. If I take the DSD file and convert it to PCM... no noise. So there is something about the recordings / files also. I just don't know what it is. Note, not making excuses for the A-GD DACs. This noise is not present with at least two other DACs.

• Song 1 - Downloaded format(s) PCM - Playback in PCM => No Noise. Software convert to DSD => Noise
• Song 1 - Downloaded format(s) DSD - Playback in DSD => Noise. Software convert to PCM => No Noise
  
  
• Song 2 (from my library) - Downloaded format(s) PCM - Playback in PCM => No Noise. Software convert to DSD => No Noise
• Song 2 (from my library) - Downloaded format(s) DSD - Playback in DSD => No Noise. Software convert to PCM => No Noise

Screenshot of some of the songs / formats.



Separate thoughts -

I'll see if I can get my hands on the NFB 11.38 to compare to the R2R-11. My guess is that it will be a non-issue, but who knows. It should also be clearly noted that the A-GD R2R DACs have a different physical section on the boards and electrical routing for DSD and PCM, so while the statement @gLer made is technically accurate re: R2R vs. D/S, I have no idea how it applies to the DSD DAC sections in the A-GD DACs or the architecture for the DSD section of the A-GD DACs. If someone can confirm that information for the DAC section for DSD that would seal it up. However, I don't want anyone thinking that the DSD signal goes through the same DAC section or that it's even technically the same process or section for the Digital to Analog Conversion "DAC" - it's not. You have the DSD DAC and the PCM DAC in the same box. What I am reasonably confident of is that the Audio-GD DACs do not internally convert DSD to PCM and feed the signal to the PCM section of the DAC. I will double confirm with Kingwa, I don't want to leave that to chance.

After moving some files, I also tried this fiddling with the R-8. Same behavior, but to my ears a lesser extent. I have to REALLY turn up the volume to hear it on DSD 64 or 128, it's audible at normal volumes w/256. So it's clearly there, but something seems different. Same sound, it's not changing in character, only attenuation. Not enough to go on. I'd say it has the same issue. If I listened to DSD music, and all my DSD music had this issue... I would not be pleased with the performance.

Happy Listening!


Cheers.

Edited to add - I feel like I should say something nice... I absolutely love the sound of all these DACs with all PCM and some DSD.... Clearly b/c I own a few. This whole DSD troubleshooting is just a fun side project.

2nd Edit to add - I need to type more slowly. Added second song scenario in red.

 

[Chopin75]

Ok it is too long and complicated to discuss DSD recording, playback Digital to analogue conversion here. and may need new thread. I try to summarize briefly as best as I can and I may not be entirely correct.
BUt this may help in understanding what we are dealing with regards to R2R DAC and DSD

Long time ago...... far far away there was on analogue recording, then someone decided to do digital recording and all hell break loose----> 16 bit recording, back then there were only R2R to decode (I think also to record)
Then there is the birth of Signa Delta 1 bit and somewhere along the line they increased to 20 bit recording and eventually 24 bit, and so there is this war between 1 bit and multibit recording.
Sony promoted 1 bit DSD recording using Signa Delta chip AD converter. and converted all analogue to digital to 1 bit DSD64 ------> led to SACD
1 bit DSD recording while very simple and most analogue, cannot be edited, and so it has to be converted back to PCM:
1 bit/DSD -----> PCM ------> CD recording or -___> DSD ----> SACD
PCM eventually still dominated the universe, despite Sony's effort to promote DSD/SACD and eventually killed SACD (almost) by forbidding native DSD to be sent out of SACD players... ----> hence sluggish to no development of external DSD DACs

In the meantime Sigma Delta chips rule, earlier ones were 1 bit, where even PCM where initially 1 bit before being processed
Then Multi-bit SD ----> e,g ESS9018 or the AKM chips ----> really multitasking chips that can do everything DSD, PCM DTS, M-ch

Then in 2011 - DSD native recordings are released as Sony no longer restricts master recordings from releasing (they are not producing any DSD recordings anyway) but the restriction on SACD still exist. Then the DSD DACs begin to appear

So there are few basic ways to do DSD decoding:


Signa Delta 1 bit --- this was designed for mainly DSD, first found in early SACD players, not so good for PCM

Signa Delta Multi-bit --- this is for everything, but probably not best for any. Though it may be very efficient in M-ch PCM so heavily used for cinema/DVD/blu-ray that use M-ch/DTS etc ......super dynamics > 130 db for ESS9038
Is this best for DSD, I don't know, it may be the only chip to do DSD M-ch well. Some theorize that the 1 bit signal is temporarily packaged into 5-6 bit like a pseudo-PCM and then decoded that way to make it more stable. Would that compromise the sound, ? Not sure.

Chipless discrete pathways - Holospring, audio-gd, I think total DAC ?? ... Lampizator..etc f these DACs use chipless, as Signa Delta has theoretical flaws (claimed by certain makers of chipless DAC) in decoding DSD due to its more complex way of decoding 1 bit (but I cannot remember exactly how, or if that is really proven or just a theory.....who knows

So the DSD signal does not get processed in the same way as the PCM inside the R2R. It does not get converted to PCM at all, it seems to run some sort of special path within R2R circuit or something there but not processed as multibit.
Here is what Holospring creator said:

I just explain how Spring can do 'discrete

DSD conversion'. Actually it is like DCS and CHORD. If you looking to DCS

and CHORD's pcb, you will find 'discrete DSD conversion' is also using

switches and precision resistors, like resistor ladder dacs. So, no matter

it is 'discrete DSD conversion' or 'discrete PCM conversion', they all use

the same discrete devices. Just work in different architecture. That makes

possible to combine these two modes in one device.

I also heard that apparently for DSD to work well the filters have to be first class. It is possible the audio-gd chipless design does not have adequate digital/analogue filters.

I have not seen any review that really compares signa Delta vs a chipless design yet for DSD.

 

[ItsAllInMyHead]

Snips and points to may websites that explain this... I like the story approach though. Well done

So the DSD siganl does not get processed in the same way as the PCM inside the R2R. It does not get converted to PCM at all, it seems to run some sort of special path within R2R circuit or something there but not processed as multibit.

Audio-GD have confirmed that the DSD does not get converted to PCM in their R2R DACs. It's converted natively. I've posted that I'd confirm my thought b/c it's too important to not be sure. It's confirmed. See also previous post. I did not ask about the architecture or methodology. It's probably too complex for me. If others are interested, I'd suggest e-mailing them.

I've sent Audio-GD links to test files, and they're working on it.

We may have a real answer soon. Thanks to @Stefano74 once again for passing along the files and the suggestion to use common files for troubleshooting and to @Chopin75 and probably a few others that I may be missing for bringing this issue to light. This issue does not occur with all DSD (the "other issue" with the R2R-11 was across all DSD). That much is clear to me, and now also to Kingwa.

I don't think Kingwa would mind me posting...

"I test other DSD files fron 64 to 512 but without this issue, only the file I downloaded from the link you send me has this noise.
Now I am no idea why cause this, maybe need more time."

I've sent him a few more files that I've found (with and without issue) and a few more that Stefano was kind enough to send (with issue)

I'll wait patiently.

(crosses fingers)

 

[Chopin75]

Snips and points to may websites that explain this... I like the story approach though. Well done



Audio-GD have confirmed that the DSD does not get converted to PCM in their R2R DACs. It's converted natively. I've posted that I'd confirm my thought b/c it's too important to not be sure. It's confirmed. See also previous post. I did not ask about the architecture or methodology. It's probably too complex for me. If others are interested, I'd suggest e-mailing them.

I've sent Audio-GD links to test files, and they're working on it.

We may have a real answer soon. Thanks to @Stefano74 once again for passing along the files and the suggestion to use common files for troubleshooting and to @Chopin75 and probably a few others that I may be missing for bringing this issue to light. This issue does not occur with all DSD (the "other issue" with the R2R-11 was across all DSD). That much is clear to me, and now also to Kingwa.

I don't think Kingwa would mind me posting...

"I test other DSD files fron 64 to 512 but without this issue, only the file I downloaded from the link you send me has this noise.
Now I am no idea why cause this, maybe need more time."

I've sent him a few more files that I've found (with and without issue) and a few more that Stefano was kind enough to send (with issue)

I'll wait patiently.

(crosses fingers)

Thanks for all your hard work in dealing witt seh this! I am glad Kingwa has finally realized the issues. I did not send him files so I don;t know what files he used. It definitely depends on the recordings as well. I read somewhere that for 1,0 bit signal DSD recording, there is no way to code for silence or reduced volume easily. So during editing in order to creat some sort of silence or quiet passage, noise is introduced. It is really complicated and I am not sure if I even interpret these science paper properly. Probably it depends on the editing process. PCM-->DSD has less issue. Analogue tape to DSD also has less issues. Native DSD recordings (not from PCM) has more issues. Eudora has quite a bit of noise but recordings are beautiful.

this site is where I get most DSD, all original DSD recordings (or DXD), there are free samples to try but I am not sure if they have noise issue.
https://www.nativedsd.com

 

[gLer]

Snips and points to may websites that explain this... I like the story approach though. Well done



Audio-GD have confirmed that the DSD does not get converted to PCM in their R2R DACs. It's converted natively. I've posted that I'd confirm my thought b/c it's too important to not be sure. It's confirmed. See also previous post. I did not ask about the architecture or methodology. It's probably too complex for me. If others are interested, I'd suggest e-mailing them.

I've sent Audio-GD links to test files, and they're working on it.

We may have a real answer soon. Thanks to @Stefano74 once again for passing along the files and the suggestion to use common files for troubleshooting and to @Chopin75 and probably a few others that I may be missing for bringing this issue to light. This issue does not occur with all DSD (the "other issue" with the R2R-11 was across all DSD). That much is clear to me, and now also to Kingwa.

I don't think Kingwa would mind me posting...

"I test other DSD files fron 64 to 512 but without this issue, only the file I downloaded from the link you send me has this noise.
Now I am no idea why cause this, maybe need more time."

I've sent him a few more files that I've found (with and without issue) and a few more that Stefano was kind enough to send (with issue)

I'll wait patiently.

(crosses fingers)

Fantastic post and feedback. Glad that you have Kingwa’s ear. Here’s hoping the solution is a simple firmware update for the FPGA dacs.

 

[Chopin75]

Fantastic post and feedback. Glad that you have Kingwa’s ear. Here’s hoping the solution is a simple firmware update for the FPGA dacs.

Hopefully just Software update. Kingwa had to put in hardware filter for the R2R-11 but then it had much more serious issues.

 

[Stefano74]

Thanks for all your hard work in dealing witt seh this! I am glad Kingwa has finally realized the issues. I did not send him files so I don;t know what files he used. It definitely depends on the recordings as well. I read somewhere that for 1,0 bit signal DSD recording, there is no way to code for silence or reduced volume easily. So during editing in order to creat some sort of silence or quiet passage, noise is introduced. It is really complicated and I am not sure if I even interpret these science paper properly. Probably it depends on the editing process. PCM-->DSD has less issue. Analogue tape to DSD also has less issues. Native DSD recordings (not from PCM) has more issues. Eudora has quite a bit of noise but recordings are beautiful.

this site is where I get most DSD, all original DSD recordings (or DXD), there are free samples to try but I am not sure if they have noise issue.
https://www.nativedsd.com

Have tried their free samples, and they also have noise.

 

[FredA]

Hopefully just Software update. Kingwa had to put in hardware filter for the R2R-11 but then it had much more serious issues.

It should be normally just a firmware change. One thing with recent dsd dacs is none does the decoding natively, stricky speaking, cause the one-bit signal is converted to a mutlibit one, with fewer bits than an actual pcm dac circuit would use. The idea is still to ouput pulses, but with more ouput values than just 0 and 1. The conversion to multibit is possibly the cullprit here. One possible error is going beyond the actual range allowed by the dac circuit. Audio-gd uses 8 bits i think in its dsd conversion circuit.

 

[Chopin75]

I did read that so,e DAC do process 1 bit as 5-6 bit with multibit Signa delta chips as it is more stable this way apparently wih less error but I believe that the aud gd is not doing any PCM conversion nor converting 1 bit to 5-8 bit at least not according to kingwa. And hence any firmware update would be separate from thr R2R ladder. It appears to be a simple processing of 1 bit with simple digital filter and decoded separately then sent to the analogue filers. The issue is likely inadequate digital filtering as the noise is reduced with hQ player resampling. Holospring and total dac probably use similar technology