[BrainSalat]

Sorry to tell you but NOS R2R is the least accurate way to convert digital to analog. If your goal is to recover the original analog music, NOS R2R is not it.

Could you please elaborate more?

 

[borkenarrou]

IMHO oversampling is not necessarily a bad thing in fact it is at core of all Delta Sigma DACs.

 

[devilboy]

IMHO oversampling is not necessarily a bad thing in fact it is at core of all Delta Sigma DACs.

Yes, but notice how R2R DACS have a reputation for being more natural sounding than Delta sigma.

 

[krude]

Yes, but notice how R2R DACS have a reputation for being more natural sounding than Delta sigma.

I sometimes switch OS on May for a minute or two just to remind myself why I spent all that money on a great NOS DAC :>

 

[borkenarrou]

Yes, but notice how R2R DACS have a reputation for being more natural sounding than Delta sigma.

Agree if compared with D/S chip based DACs, but one thing I noticed is Chord DACs even though delta sigma design is more natural than chip based ones, I guess it gets even better with higher end DCS and MSB DACs though haven't heard yet, so implementation maybe key here.

 

[krude]

Agree if compared with D/S chip based DACs, but one thing I noticed is Chord DACs even though delta sigma design is more natural than chip based ones, I guess it gets even better with higher end DCS and MSB DACs though haven't heard yet, so implementation maybe key here.

Isnt MSB also R2R? Chord is not a typical delta sigma, it's their proprietary approach, so is DCS.

Also worth mentioning Mola Mola, tube based DACs like Audio Note, Weiss etc. The field is pretty crowded around 6-8k with lots of options for every taste.

 

[dudeX]

I have owned the May for over 1 year and a half. For 44/48 content, it is a bit soft, but it is generally pleasing and organic. Even sources that are low frequency ( think old games ) sound pretty natural and pleasing rather than harsh and digital.
It can sound good with 96 kHz audio files without HQP or other samplers.

But when you use upsampling you really get a WOW factor in the music. For me that’s HQP with DSD.

 

[KrauserX91]

You can certainly get a very lively sound out of Cayins with HA300 being the daddy (Im also waiting for mine ... ) : ) just be prepared to roll some tubes to find your sound.

That being said I use Bliss a lot, as a pre into HA6a and also into a power amp for speakers and I love the sound I'm getting in both cases. Very refined, holographic, full, dynamic and relaxed and easy at the same time.

Ok so trying to copy you a bit, using the Bliss as a preamp to Euforia AE and I found a new sound of my liking. The Caldera (unlike Atrium) with Euforia directly from Spring 3 is terrible for me, but somehow the Bliss fixed this. I think the problem definitely is how laid back is the Bliss itself for me, but seems it can change completely the chain using it as a preamp. A thing I didnt mention (because is not suposed to be the purpose of this stuff) is I enjoy the Spring3/Bliss/Caldera a lot when gaming, I think it could be because the games, or most of them do oversampling already or who knows why.

 

[chesebert]

Could you please elaborate more?

Sample and hold - hence the step function. This is a typical R2R dac output. Practically speaking you can think of NOS R2R as an EQ filter of negative treble shelf.

My point has nothing to do with personal preference in audio reproduction. I roll different dac architectures depending on my listening mood and I treat NOS R2R just like a different form of EQ.

 

[borkenarrou]

Isnt MSB also R2R? Chord is not a typical delta sigma, it's their proprietary approach, so is DCS.

Also worth mentioning Mola Mola, tube based DACs like Audio Note, Weiss etc. The field is pretty crowded around 6-8k with lots of options for every taste.

Ah thanks for correcting.

 

[BrainSalat]

Sample and hold - hence the step function. This is a typical R2R dac output. Practically speaking you can think of NOS R2R as an EQ filter of negative treble shelf.

My point has nothing to do with personal preference in audio reproduction. I roll different dac architectures depending on my listening mood and I treat NOS R2R just like a different form of EQ.

Right, this is the unfiltered output of a R2R network stage of a R2R DAC, not the final output of a R2R DAC. I saw you took this image from this tutorial
https://www.tek.com/en/blog/tutorial-digital-analog-conversion-r-2r-dac.

and in order to compare apples with apples this is how it looks the unfiltered output of a Delta Sigma DAC (1-Bit DAC graph).

Practically speaking you can think of NOS R2R as an EQ filter of negative treble shelf.

Sorry but this is absolutely wrong. From where did you get that info?

Delta Sigma and R2R are two different methods to reconstruct signals accurately (theorically) but both suffer of different practical difficulties/trade-offs. On one hand DS DACs have to do oversampling and apply complex reconstruction filters that introduce ringing, moiré artifacts which are somehow reduced by increasing the number of parameters of the filters (tap count) but there more taps the more processing power you need (more expensive). That's the difference between the different Chord DAC models for example, or the different filters of the HQPlayer. R2R DACs don't suffer from the previous problems but little tolerances of the resistors (respect its nominal Ohm values) of the R2R networks (theorically they all should have the same Ohm value) can lead to imbalances in the DAC conversion. However recently companies found a way to manufacture resistors with a extreme low tolerance and combined that with FPGA chips that compensate for differences (HoloAudio, Rockna Audio, etc).

I have personally tested side by side the Chord TT2 vs the HoloAudio Spring3 KTE and while they share a lot of qualities like resolution, layering, clarity, soundstage the Spring3 sounds to me more "analog", like I would be listening from vinyl or reel and the TT2 more "digital". But both are great DACs, I would have been happy with any of them, but I decided for the more analog sound of the Spring3.

Until recent R2R DACs were not measurement very well on lab test but that have changed:
https://www.audiosciencereview.com/...o-audio-spring-3-dac-wpreamp-prototype.22083/

 

[chesebert]

Right, this is the unfiltered output of a R2R network stage of a R2R DAC, not the final output of a R2R DAC. I saw you took this image from this tutorial
https://www.tek.com/en/blog/tutorial-digital-analog-conversion-r-2r-dac.

and in order to compare apples with apples this is how it looks the unfiltered output of a Delta Sigma DAC.

To my knowledge there is generally no reconstruction filter on the NOS R2R unlike Delta Sigma. I suppose NOS dacs can always use some analog LPF which is not really ideal

The only way to get a nice and gentle LPF to work as a reconstruction filter is to upsample the signal to a few dozen MHz.

 

[BrainSalat]

To my knowledge there is generally no reconstruction filter on the NOS R2R unlike Delta Sigma. I suppose NOS dacs can always use some analog LPF which is not really ideal

For NOS R2R true, you are right!, but they could use output capacitors (passive filtering) and the output of nowadays top R2R DACs doesn't look like your sample picture above that represent a 4-bit R2R DAC (or a 16 levels DAC resolution) where you clearly see steeping which musically transforms to a very compressed dynamic range.

Previous R2R DAC generations used 16-bits (65536 levels), or today's entry level R2R DACs, where you almost won't see the steps on a oscilloscope. With these there was still some slight perception of lack of resolution compared DS DACs, but nowadays top R2R DACs like HoloAudio May/Spring3, Rockna Wavelight are 24-bit (16,7 Million levels) R2R DACs.

The Spring3 KTE with a SINAD of 116dB has 19-bits of effective resolution (free of noise)

 

[chesebert]

For NOS R2R true, you are right!, but they could use output capacitors (passive filtering) and the output of nowadays top R2R DACs doesn't look like your sample picture above that represent a 4-bit R2R DAC (or a 16 levels DAC resolution) where you clearly see steeping which musically transforms to a very compressed dynamic range.



Previous R2R DAC generations used 16-bits (65536 levels), or today's entry level R2R DACs, where you almost won't see the steps on a oscilloscope. With these there was still some slight perception of lack of resolution compared DS DACs, but nowadays top R2R DACs like HoloAudio May/Spring3, Rockna Wavelight are 24-bit (16,7 Million levels) R2R DACs.

The Spring3 KTE with a SINAD of 116dB has 19-bits of effective resolution (free of noise)

I believe most use analog LPF, so yes, no actual stepped analog output. This is why most NOS output has a negative treble shelf, which I said before was not desirable.

That 19bit figure you quoted is the dac in OS mode, correct?

Holo spring KTE. Green is NOS mode.

 

[BrainSalat]

I believe most use analog LPF, so yes, no actual stepped analog output. This is why most NOS output has a negative treble shelf, which I said before was not desirable.

That 19bit figure you quoted is the dac in OS mode, correct?


Holo spring KTE. Green is NOS mode.

This is the frequency response of the Holo Spring (1) KTE (already discontinued, current version is Spring3, quite more advanced) and all 3 curves you are showing here are NOS mode (not only the green one). The difference is the frequency of the input signal, green is 44khz, magenta is 96Khz and red 192Khz. This graph shows that in the worst case (input signal 44Khz) the attenuation at 20Khz is ~1.7 dB which is quite OK. When you design a pass band filter you expect the filter to reach -3dB at the cutoff frequency. For 192Khz input signal the attenuation at 20Khz is ~0.9 dB, excellent!. Right, with Delta Sigma DACs you have the advantage you can build sharper filters (with not so long decay like in the previous graph) , but at the cost of ringing, which is also not desirable, just compare the impulse response of a R2R DAC, against a DS DAC.

So I would not dare to say one system is better than the other, each has pros/contras, you get excellent results with both, but they do sound different and I believe when you are looking for a more "analog" sound, R2R DACs are the way to go.

BTW Spring1 and Spring2 had also OS modes but Holo remove it with Spring3, because you can get that better oversampling externally (software: HQPlayer/Hardware: M Scaler) and then you can enjoy the pros/contras of both worlds.