GUSTARD DAC-R26 Balanced Decoder R2R+1Bit Dual Native Decoding Music Bridge
Jun 9, 2023 at 10:34 PM Post #6,406 of 8,833
Out of curiosity - why 15 bit?
Oh man
I went into it pretty deep into bit depth about a 100pages ago lol. But to summarize
ENOB of the R26 is 15.65 bits.
Setting the correct bit depth ensures that noise shaping is operating within the linear range of the DAC and therefore will be more effective.
When you set the bit depth too high, you get a poorer result due to the noise shaping not working fully. And so it's better to go a little too low than it is to go even 1 bit too high.

No DAC of any kind is linear to beyond 22 bits currently.
 
Last edited:
Jun 9, 2023 at 11:26 PM Post #6,407 of 8,833
Or just upscale PCM to DSD which I prefer by a lot on my setup. I think it's more important than what port you use in this DAC as well.
 
Jun 10, 2023 at 3:50 AM Post #6,409 of 8,833
Out of curiosity - why 15 bit?
It has something to do how HQPlayer operates internally. This software adds extensive noise shaping and the amount is handled internally by the settings in a dithering section. These two operations are technically different, but in HQPlayer there are linked together for some reason. For those who use HQPlayer, noise shaping and dithering depth become the same things, it is why you receive weird explanations mistakenly refering to noise shaping, instead of dithering.

I was always wondering why for the best SQ with HQPlayer you need to reduce word size of a dithering, "matching a ladder linearity". There is noise on all bits, including less significant bits that provide suficient randomisation of linearity errors. It is why I never accepted such explanation, you can see my responses a way back.

There is a question: Why there are benefits on some brands like Gustard or Holo Audio DACs and very little or none when using audiophile brands like Audio GD?

I had no real answer till finding opamps in the R26 audio path. Now it comes to a conclusion that answer is indeed not in dithering, but noise shaping. Noise shaped source is much easier for opamps to handle large signal fast transients. This is a big problem with opamps. If you look at any datasheet there are two figures for the opamp pulse response. One for a small signal ~50mV and another for a large signal ~2V. I will attach pictures when coming home. The former carries small overshots with oscilations. This is normal behaviour of a closed loop operation. On the large signal response plot there are no oscillations. It indicates that during a small period of time some internal amplification stages are saturated, output do not follow small changes of the input.
 
Last edited:
Jun 10, 2023 at 7:22 AM Post #6,410 of 8,833
I can't remember what the maximum input rate of the Ref 7.1 is, but I'd suggest trying HQPlayer upsampling to it. From my own experiences, I think the real limitation to Audio-gd DACs has been the digital filter. Paradoxically, I didn't feel that HQPlayer improved the R26 at all.
Regarding the maximum input rate, it is 96kHz, but the standard DSP setting (which I use) upsamples whatever it receives 8 times.

And yes, the initial digital processing in the Ref 7.1 was a limiting factor, but then Kingwa made a new DSP (version 7, I think it was called) which was offered to existing owners (and done via Team Viewer). I got it, and the upgrade in resolution was easily noticeable. It was literally like getting a new DAC. From memory, I would say the difference between the Ref 7.1 (w. DSP v7) and R26 is less than half, and more likely about a quarter or eighth of the difference before / after the DSP upgrade of the Ref 7.1.

After the DSP v7 upgrade on the Ref 7.1, I could hear the extra resolution and details on every single track I played.
Regarding the R26 vs. Ref 7.1, I only hear the extra resolution on particularly revealing tracks.
 
Jun 10, 2023 at 7:37 AM Post #6,411 of 8,833
Regarding the maximum input rate, it is 96kHz, but the standard DSP setting (which I use) upsamples whatever it receives 8 times.

And yes, the initial digital processing in the Ref 7.1 was a limiting factor, but then Kingwa made a new DSP (version 7, I think it was called) which was offered to existing owners (and done via Team Viewer). I got it, and the upgrade in resolution was easily noticeable. It was literally like getting a new DAC. From memory, I would say the difference between the Ref 7.1 (w. DSP v7) and R26 is less than half, and more likely about a quarter or eighth of the difference before / after the DSP upgrade of the Ref 7.1.

After the DSP v7 upgrade on the Ref 7.1, I could hear the extra resolution and details on every single track I played.
Regarding the R26 vs. Ref 7.1, I only hear the extra resolution on particularly revealing tracks.
Appreciate you have limited appetite for the additional cost and hassle of further tweaking. Fair enough, I hear ya. Though as a straightforward and pretty affordable first step to let the R26 internal renderer perform a little closer to its potential I’d recommend the iFi LAN iPurifier. Simpler, cheaper and cleaner than inserting a new switch or pair of FMCs with power supplies etc. It’d give you taste of the character of improvements a few of us have been enthusiastically trying to convey.

Edit… I see the iFi has already garnered quite a few glowing reviews on Amazon even though it’s only been out for a couple of months. Great thing with gizmos like this is you can use em with other DACs/servers/streamers too, or (as I do) in combination with additional ethernet filtration like switches or FMCs should one go down that route.

https://www.amazon.com/iFi-LAN-iSilencer-Electrical-Ethernet/dp/B0BV72SW8V#customerReviews
 
Last edited:
Jun 10, 2023 at 9:55 AM Post #6,412 of 8,833
Appreciate you have limited appetite for the additional cost and hassle of further tweaking. Fair enough, I hear ya. Though as a straightforward and pretty affordable first step to let the R26 internal renderer perform a little closer to its potential I’d recommend the iFi LAN iPurifier. Simpler, cheaper and cleaner than inserting a new switch or pair of FMCs with power supplies etc. It’d give you taste of the character of improvements a few of us have been enthusiastically trying to convey.

Edit… I see the iFi has already garnered quite a few glowing reviews on Amazon even though it’s only been out for a couple of months. Great thing with gizmos like this is you can use em with other DACs/servers/streamers too, or (as I do) in combination with additional ethernet filtration like switches or FMCs should one go down that route.

https://www.amazon.com/iFi-LAN-iSilencer-Electrical-Ethernet/dp/B0BV72SW8V#customerReviews
I think I already mentioned that in a previous post here (as well as the competing product from Sellarz). This is a kind of upgrade that I may try, if I keep the R26.

...

For those interested, I still consider whether I should keep or return the R26.

As I have said in various ways in previous posts, the extra level of detail and resolution going from the Ref 7.1 (w. DSP v7) has been underwhelming, considering the age difference. This made me think I should keep the Ref 7.1, and that made me ask myself why I should keep the R26 (and own both). Right now, I have come to the conclusion that I will only keep one of them, which makes it much more sensible to keep the R26 when selling the Ref 7.1. The logic in that would be that, while the extra level of resolution and detail out of the box is much smaller than I expected, the R26 is more future-proof, and easier to upgrade along the way having LAN, IIS etc. inputs and ability to be controlled via something like Audirvana with a software based parametric equalizer. I will most likely also sell my upgraded Squeezebox Touch when selling the Ref 7.1.
 
Jun 10, 2023 at 10:20 AM Post #6,413 of 8,833
It has something to do how HQPlayer operates internally. This software adds extensive noise shaping and the amount is handled internally by the settings in a dithering section. These two operations are technically different, but in HQPlayer there are linked together for some reason. For those who use HQPlayer, noise shaping and dithering depth become the same things, it is why you receive weird explanations mistakenly refering to noise shaping, instead of dithering.

I was always wondering why for the best SQ with HQPlayer you need to reduce word size of a dithering, "matching a ladder linearity". There is noise on all bits, including less significant bits that provide suficient randomisation of linearity errors. It is why I never accepted such explanation, you can see my responses a way back.

There is a question: Why there are benefits on some brands like Gustard or Holo Audio DACs and very little or none when using audiophile brands like Audio GD?

I had no real answer till finding opamps in the R26 audio path. Now it comes to a conclusion that answer is indeed not in dithering, but noise shaping. Noise shaped source is much easier for opamps to handle large signal fast transients. This is a big problem with opamps. If you look at any datasheet there are two figures for the opamp pulse response. One for a small signal ~50mV and another for a large signal ~2V. I will attach pictures when coming home. The former carries small overshots with oscilations. This is normal behaviour of a closed loop operation. On the large signal response plot there are no oscillations. It indicates that during a small period of time some internal amplification stages are saturated, output do not follow small changes of the input.
You are like the smartest MOST WRONG person I have ever encountered.
ENOB has nothing to do with HQplayer, although thankfully a program exists that lets us correct for this.
ENOB is real world resolution. It includes the effects from INL/DNL, distortion, jitter, etc.; which all contribute to the noise of the DAC.
ENOB does not include the effects of gain or offset error, clock error, intermodulation distortion, etc.
It’s the actual bit depth exiting your DAC headed towards preamp/amp. Internally HQplayer and the R26 are upsampling PCM to 64bit float.
When you upsample a base rate to a much higher rate, you can reduce the bit-depth without losing information. The higher rate offsets the lower bit depth and allows the corner frequency of the reconstruction filter to be moved much higher and away from the baseband where our ears are more sensitive to phase, which is affected by the reconstruction filter. Also, this can optimize the signal to the dac elements because a 24 bit dac usually isn't linear all the way down to the least significant bit. Moving the bits to a smaller range often make the dac perform better because it is being used in its more linear range.
 
Jun 10, 2023 at 10:20 AM Post #6,414 of 8,833
... finding opamps in the R26 audio path. ...
That is your guess. Seeing IC op-amps in the analog section is not the same as confirming it is in the audio path. Audio-gd points out, in their description of the Ref 7.1 DAC, that the IC's near the outputs are for DC servo. This is most likely the case with the R26, considering they call it "Discrete" and that Sound News opened the R26 and confirms that there are not IC's in the analog (EDIT) path.
 
Last edited:
Jun 10, 2023 at 2:47 PM Post #6,415 of 8,833
The R26 works as it does. It doesn't matter if these opamps are in the signal path or not. This knowledge will not change the sound of the device.

The R26 is a quality DAC, excellent considering the price.

Today I listened to a great recording of Arthur Fiedler and Boston Pops - Carmen Ballet for strings. It sounds like a tape recorder, incredibly good.
 
Jun 10, 2023 at 4:40 PM Post #6,416 of 8,833
I think I already mentioned that in a previous post here (as well as the competing product from Sellarz). This is a kind of upgrade that I may try, if I keep the R26.

...

For those interested, I still consider whether I should keep or return the R26.

As I have said in various ways in previous posts, the extra level of detail and resolution going from the Ref 7.1 (w. DSP v7) has been underwhelming, considering the age difference. This made me think I should keep the Ref 7.1, and that made me ask myself why I should keep the R26 (and own both). Right now, I have come to the conclusion that I will only keep one of them, which makes it much more sensible to keep the R26 when selling the Ref 7.1. The logic in that would be that, while the extra level of resolution and detail out of the box is much smaller than I expected, the R26 is more future-proof, and easier to upgrade along the way having LAN, IIS etc. inputs and ability to be controlled via something like Audirvana with a software based parametric equalizer. I will most likely also sell my upgraded Squeezebox Touch when selling the Ref 7.1.

Understood on all counts, incl your previous awareness of the iFi, which I am endorsing/recommending from personal experience as an easy, effective, affordable and hopefully quick (if locally in stock for you as it was for me) way to either aid your decision or feel better about it, whichever way you go.
 
Jun 11, 2023 at 8:44 AM Post #6,417 of 8,833
After considering my options and needs, I concluded that I will sell the Reference 7.1 DAC and keep the R26. This leads me to the obvious question:

How to maximize the performance of the R26?

I am looking for suggestions within a sensible price range like $100-200.

If we assume that the best input is the LAN port, given the data transmission is cleaned up, then I would like to get some specific advice (links or at least the full name on items) on how to isolate electrical noise from the Ethernet.

My speculation in this context is that gadget types like the Ifi LAN iSilencer may do a good job, but they apparently don't eliminate the noise issues completely. Is see that some recommend having two such LAN filters, which to me indicates that such units don't remove noise 100%. There is a reduction, but not a full elimination.

Instead, it would be better to get something that completely removes electrical noise from the line, like getting two fiber LAN converters (converting copper to fiber to copper) since the fiber only carries the data and won't carry noise, if I understand this correctly.

Another way to do it would be getting a separate Wi-Fi repeater or access point (?) with a LAN output which then is connected to the DAC with a regular copper LAN cable. However, I question whether the second option is as bullet-proof as the first one, since the repeater or access point may create some noise of its own...?

A third and easy way to eliminate computer noise getting into the DAC was mentioned by Currawong in his YouTube video: Simply use the optical input on the DAC. However, years ago I read that the optical digital connection is the poorest and generally should be avoided when playing hi res music. This may be wrong or technical advancement in DACs may have made this claim obsolete.

I should add that I have in fact tested the optical input using my Squeezebox Touch and it did sound nice. Whether it was better or worse than via LAN input is another story. I haven't does a side-by-side test of these two, and for obvious reasons, not with a notoriously clean data connection into the LAN port.

To me, the idea of going the fiber route seems like the safest bet and won't cost a fortune (which is important). I'd love to see some experienced/knowledgeable suggestions.
 
Jun 11, 2023 at 8:51 AM Post #6,418 of 8,833
The easiest thing to do will be to run the Fmc mode if you insist on Lan.
Other options would be the expensive variant of sonore at least on the clean side.
I think the dirty side could be neglected a bit.

The other variant and more constant solution is to go after what I personally find better a DDC via I2S connection.
Singxer Su2/Su6, Matrix Spidif3, maybe one or two other DDCs.

Congratulations on keeping the R26.
 
Jun 11, 2023 at 8:53 AM Post #6,419 of 8,833
After considering my options and needs, I concluded that I will sell the Reference 7.1 DAC and keep the R26. This leads me to the obvious question:

How to maximize the performance of the R26?

I am looking for suggestions within a sensible price range like $100-200.

If we assume that the best input is the LAN port, given the data transmission is cleaned up, then I would like to get some specific advice (links or at least the full name on items) on how to isolate electrical noise from the Ethernet.

My speculation in this context is that gadget types like the Ifi LAN iSilencer may do a good job, but they apparently don't eliminate the noise issues completely. Is see that some recommend having two such LAN filters, which to me indicates that such units don't remove noise 100%. There is a reduction, but not a full elimination.

Instead, it would be better to get something that completely removes electrical noise from the line, like getting two fiber LAN converters (converting copper to fiber to copper) since the fiber only carries the data and won't carry noise, if I understand this correctly.

Another way to do it would be getting a separate Wi-Fi repeater or access point (?) with a LAN output which then is connected to the DAC with a regular copper LAN cable. However, I question whether the second option is as bullet-proof as the first one, since the repeater or access point may create some noise of its own...?

A third and easy way to eliminate computer noise getting into the DAC was mentioned by Currawong in his YouTube video: Simply use the optical input on the DAC. However, years ago I read that the optical digital connection is the poorest and generally should be avoided when playing hi res music. This may be wrong or technical advancement in DACs may have made this claim obsolete.

I should add that I have in fact tested the optical input using my Squeezebox Touch and it did sound nice. Whether it was better or worse than via LAN input is another story. I haven't does a side-by-side test of these two, and for obvious reasons, not with a notoriously clean data connection into the LAN port.

To me, the idea of going the fiber route seems like the safest bet and won't cost a fortune (which is important). I'd love to see some experienced/knowledgeable suggestions.
Toslink optics are much worse in most cases. I solved the problem to some extent by means of a LAN switch and quality ethernet cables. For local streaming. To the extent that I have no need to do anything else. Tidal doesn't sound nearly as good so I think it would take extra steps for internet streaming. For example optical converters connected to each other by optical cables. It has nothing to do with Toslink optics.
After considering my options and needs, I concluded that I will sell the Reference 7.1 DAC and keep the R26. This leads me to the obvious question:

How to maximize the performance of the R26?

I am looking for suggestions within a sensible price range like $100-200.

If we assume that the best input is the LAN port, given the data transmission is cleaned up, then I would like to get some specific advice (links or at least the full name on items) on how to isolate electrical noise from the Ethernet.

My speculation in this context is that gadget types like the Ifi LAN iSilencer may do a good job, but they apparently don't eliminate the noise issues completely. Is see that some recommend having two such LAN filters, which to me indicates that such units don't remove noise 100%. There is a reduction, but not a full elimination.

Instead, it would be better to get something that completely removes electrical noise from the line, like getting two fiber LAN converters (converting copper to fiber to copper) since the fiber only carries the data and won't carry noise, if I understand this correctly.

Another way to do it would be getting a separate Wi-Fi repeater or access point (?) with a LAN output which then is connected to the DAC with a regular copper LAN cable. However, I question whether the second option is as bullet-proof as the first one, since the repeater or access point may create some noise of its own...?

A third and easy way to eliminate computer noise getting into the DAC was mentioned by Currawong in his YouTube video: Simply use the optical input on the DAC. However, years ago I read that the optical digital connection is the poorest and generally should be avoided when playing hi res music. This may be wrong or technical advancement in DACs may have made this claim obsolete.

I should add that I have in fact tested the optical input using my Squeezebox Touch and it did sound nice. Whether it was better or worse than via LAN input is another story. I haven't does a side-by-side test of these two, and for obvious reasons, not with a notoriously clean data connection into the LAN port.

To me, the idea of going the fiber route seems like the safest bet and won't cost a fortune (which is important). I'd love to see some experienced/knowledgeable suggestions.
Toslink optics are much worse in most cases. I solved the problem to some extent by means of a LAN switch and quality ethernet cables. For local streaming. To the extent that I have no need to do anything else. Tidal doesn't sound nearly as good so I think it would take extra steps for internet streaming. For example optical converters connected to each other by optical cables. It has nothing to do with Toslink optics.
 
Jun 11, 2023 at 9:09 AM Post #6,420 of 8,833
Hi everyone,

I'm interested in purchasing an R26 to replace my small A18. Should be a considerable upgrade of course.
What I'm confused about is the NOS mode.
Most reviewers fail to write in which mode they tested the unit.
My favorite and main format being still the good old CD ( played through an Audiolab CD transport via the SPDIF coax output, and passing through an iFi Audio SPDIF iPurifier 2) should I use OS mode or NOS mode? Does NOS mode work better with higher resolution formats? Wouldn't a 44.1KHz red book file be somewhat problematic in NOS mode?

thanks
@ Rolox - you & I seem to be the only ones here using CD. I also have an Audiolab, upgraded from 6000 to the new 9000 (worthwhile). I bought quite an expensive & good quality Coax cable but then read a comment somewhere that optical might be better so I substituted a good Toslink fiber optic cable, bringing a definite improvement all round. Not being one to sit still for long, I then bought a glass Toslink cable, and wow! a tremendous step forwards! I believe that one of the factors at play are eddy currents caused by the copper shielding of the coax, as there's no need for any shielding with optical.
 

Users who are viewing this thread

  • Back
    Top