What would be important criteria for your purchase of a new DAC?

[Jud]

24 minutes ago, Dr Tone said:

 

Since the OP asked about what to look for in a DAC, I thought maybe we would see more points (some similar to yours) and reasons why related to DAC design like:

 

Balanced/true dual mono design.

Output stage methodology.

FPGA and/or DAC(s) chip.

Specific filtering, oversampling, NOS etc

Clocking, Isolation and power etc

 

Instead the thread seems to have just deviated in the direction of most other threads on this site, the latest and greatest buzz words on how to get input into the DAC.  Everyone might as well just say "it needs to let me pump the highest DSD rate available into it via HQPlayer".

 

PS) I love HQPlayer as much as the next guy.

 

Your 1st and 5th items I feel are definitely relevant, though many folks seem to be playing around with their own isolation and power solutions.

 

Can you tell me what you're thinking about with regard to your 2nd item (what sorts of output stages and possibly issues concerning them)?

 

 If someone is doing software upsampling to high DSD rates, then your third and fourth items do indeed seem to come down to "it needs to let me pump the highest DSD rate available into it."

[pkane2001]

42 minutes ago, Dr Tone said:

Instead the thread seems to have just deviated in the direction of most other threads on this site, the latest and greatest buzz words on how to get input into the DAC.  

 

That's true. I can certainly try to explain my choices and thinking behind them.

• USB DSD256/512 native

I have my content stored on a NAS drive. I'd like to have a computer to play back that content. After evaluating various playback software, I found HQPlayer producing the best sound with multiple DACs I've tried. And with HQPlayer, I prefer the sound of all my content (mostly 44/16 PCM) upscaled to DSD512. The difference between PCM and DSD for me is not huge, but still obvious. And DSD512 has a slight edge over DSD256. Native DSD is preferred to DoP since it is simpler and requires less bandwidth for the same rate of playback.

• I2S input

As most audiophiles I like to tinker and mod my system, I'd like to be able to add various components to try to improve the sound in the digital domain, before it reaches the DAC. Things like DDC boxes, SRC, jitter reduction, etc. To do this effectively, I want to have the lowest jitter, most direct interface as the input into the DAC, and that is I2S.

• Digital volume control with remote

I spent a long time evaluating various preamps, but never found one that didn't add some coloration or distortion into my system. As a general principle, I believe in simpler is better.  I prefer one less component, one less power supply, and one less set of cables in the analog chain.

• Balanced analog output

Balanced delivers cleaner sound through common mode rejection. But, primarily, I find that it just sounds better with my amps (Pass Aleph). A while back I paid a few thousand dollars for a pair of custom unbalanced interconnects I thought they were really amazing. Recently, I found that a $60 balanced pro-audio interconnects that sounded better with my amps. I'm using them now, the custom cables are in a box.

• Excellent LPS

I keep finding that good, clean power source is critical everywhere, from computer to DDC to the DAC. Small improvements, such as replacing voltage regulators with less noisy ones yield obvious improvement in SQ. So, the DAC should have an excellent power supply that I don't need to mod or replace.

[Dr Tone]

1 hour ago, Jud said:

Can you tell me what you're thinking about with regard to your 2nd item (what sorts of output stages and possibly issues concerning them)?

I was wanting to learn more than anything.

 

I was thinking descrete vs opamp and maybe further details like class, negative feeback and stuff like that.  It might be to broad of a subject.

[Dr Tone]

Proper output voltage is something that is overlooked as well I think.  

[esldude]

2 hours ago, pkane2001 said:

 

• I2S input

As most audiophiles I like to tinker and mod my system, I'd like to be able to add various components to try to improve the sound in the digital domain, before it reaches the DAC. Things like DDC boxes, SRC, jitter reduction, etc. To do this effectively, I want to have the lowest jitter, most direct interface as the input into the DAC, and that is I2S.

•  

Well the systems that let the clock sitting at the DAC do the clocking have the lowest jitter.  I2S is not equal to things like asycnh USB, wifi or ethernet input to DACs for lowest jitter purposes. 

[esldude]

1 hour ago, Dr Tone said:

I was wanting to learn more than anything.

 

I was thinking descrete vs opamp and maybe further details like class, negative feeback and stuff like that.  It might be to broad of a subject.

Discrete vs opamp has rather become a non-topic.  You can find examples of either at both price extremes.  Excellent opamps are fine and not a bottleneck to sound quality.  Opamps inherently use tremendous amounts of feedback and by their fidelity illustrate feedback alone is also a non-topic.  Opamps are typically operated class A for quality devices. 

 

So that pretty much covers class of operation, feedback and opamps vs discrete.

[jcbenten]

I would like a DAC to internally upsample &/or convert to max PCM & DSD available with purpose built logic/DSP chips.  I get to pick output format and sampling rate via tablet/phone.  I do not want a PC in the chain.  Connection will be wireless or ethernet.  Music is on an NAS (I know...technically a PC but I do not want pay for an i7 NAS).  And I want it at a reasonable cost...<$2K.

 

Probably not going to happen anytime soon.  In the meantime I am looking at the OPPO 205 for component reduction purposes (eliminate SB Touch and external DAC).

[Dr Tone]

6 hours ago, esldude said:

Discrete vs opamp has rather become a non-topic.  You can find examples of either at both price extremes.  Excellent opamps are fine and not a bottleneck to sound quality.  Opamps inherently use tremendous amounts of feedback and by their fidelity illustrate feedback alone is also a non-topic.  Opamps are typically operated class A for quality devices. 

 

So that pretty much covers class of operation, feedback and opamps vs discrete.

 

Interesting.

 

And yet, discrete output is considered the darling and opamp based output the bastard by many.

[jabbr]

7 hours ago, esldude said:

Discrete vs opamp has rather become a non-topic.  You can find examples of either at both price extremes.  Excellent opamps are fine and not a bottleneck to sound quality.  Opamps inherently use tremendous amounts of feedback and by their fidelity illustrate feedback alone is also a non-topic.  Opamps are typically operated class A for quality devices. 

 

So that pretty much covers class of operation, feedback and opamps vs discrete.

You are overstating this. There clearly can be differences in different designs, using different transistors etc. That doesn't mean that integrated opamps are necessarily bad but, for example, if one is trying to emulate a single ended triode and/or minimize global feedback, the op amp probably not the best design choice.

 

in any case nowhere is it written that different electronic designs will or are intended to sound the same -- that would be a rather boring world for amplifier designers who continue to innovate.

[jabbr]

7 hours ago, esldude said:

Well the systems that let the clock sitting at the DAC do the clocking have the lowest jitter.  I2S is not equal to things like asycnh USB, wifi or ethernet input to DACs for lowest jitter purposes. 

Right. Let me expand on this with two points  

 

1) goal to minimize distance between clock and DAC latches in order to reduce skew -- a type of jitter.

2) an external clock line as with I2S input, has an undefined phase relationship to the DAC clock, which eliminates any benefit. PLL needed for phase synch can increase phase noise -- assuming that the DAC has one of those really nice low phase error clocks that are becoming ubiquitous. So you can make I2S work well but either way, USB, I2S or Ethernet, you are crossing a clock domain.

 

3) If you consider these details irrelevant, and aren't concerned about phase noise, then there is no business considering anything other than USB.

[pkane2001]

46 minutes ago, jabbr said:

Right. Let me expand on this with two points  

 

1) goal to minimize distance between clock and DAC latches in order to reduce skew -- a type of jitter.

2) an external clock line as with I2S input, has an undefined phase relationship to the DAC clock, which eliminates any benefit. PLL needed for phase synch can increase phase noise -- assuming that the DAC has one of those really nice low phase error clocks that are becoming ubiquitous. So you can make I2S work well but either way, USB, I2S or Ethernet, you are crossing a clock domain.

 

3) If you consider these details irrelevant, and aren't concerned about phase noise, then there is no business considering anything other than USB.

 

I2S input does not rely on internal DAC clock for timing.  All other points are correct. Please allow me to quote myself  as to the context for why I'd like I2S input:

 

Quote

 I'd like to be able to add various components to try to improve the sound in the digital domain, before it reaches the DAC.

 

Just because a DAC is async does not guarantee low jitter. All it does (if implemented properly) eliminate jitter from the transmission. The onboard clock and the logic at the DAC can introduce its own jitter, which is why I would always want to have a way to bypass the internal clock and internal timing logic. In this context, I2S is the best input. 

[Jud]

59 minutes ago, jabbr said:

Right. Let me expand on this with two points  

 

1) goal to minimize distance between clock and DAC latches in order to reduce skew -- a type of jitter.

2) an external clock line as with I2S input, has an undefined phase relationship to the DAC clock, which eliminates any benefit. PLL needed for phase synch can increase phase noise -- assuming that the DAC has one of those really nice low phase error clocks that are becoming ubiquitous. So you can make I2S work well but either way, USB, I2S or Ethernet, you are crossing a clock domain.

 

3) If you consider these details irrelevant, and aren't concerned about phase noise, then there is no business considering anything other than USB.

 

Is #3 correctly stated?  If you *are* concerned about crossing clock domains, jitter/phase noise, shouldn't you be strongly considering async USB?

[Jud]

6 minutes ago, pkane2001 said:

 

I2S input does not rely on internal DAC clock for timing.  All other points are correct. Please allow me to quote myself  as to the context for why I'd like I2S input:

 

 

Just because a DAC is async does not guarantee low jitter. All it does (if implemented properly) eliminate jitter from the transmission. The onboard clock and the logic at the DAC can introduce its own jitter, which is why I would always want to have a way to bypass the internal clock and internal timing logic. In this context, I2S is the best input. 

 

But if an internal and external clock are equivalent in spec and implementation, the internal one will inevitably have less jitter.  So why is your goal to bypass the internal clock rather than to buy or DIY a really good one?

[pkane2001]

23 minutes ago, Jud said:

But if an internal and external clock are equivalent in spec and implementation, the internal one will inevitably have less jitter.  So why is your goal to bypass the internal clock rather than to buy or DIY a really good one?

 

Imagine that a new clock technology becomes available. I'd like to be able to leverage it without replacing the DAC. (This one is not imaginary, as I just did this with my DAC by placing SU-1 in front of it using I2S into my DAC.)

 

Or, if a new format becomes the standard, such as MQA or XYZ that are not supported by my DAC: I'd like to replace the PC interface with a new one without having to throw away the DAC. Or, perhaps thunderbolt or some other interface becomes dominant on PCs and my DAC only has USB input. Or, if in the future I want to add ethernet connection as an option. etc., etc.

 

[christopher3393]

Sound quality, decent headphone amp, has to work well with my present cans (Hifiman Edition X). Since the Chord Mojo works so well with my headphones, I'm saving for a Hugo 2.

[esldude]

4 hours ago, Dr Tone said:

 

Interesting.

 

And yet, discrete output is considered the darling and opamp based output the bastard by many.

And yet further make a list of DACs from $1k to $20k and put on that list whether they use opamps or discrete.

[esldude]

3 hours ago, jabbr said:

You are overstating this. There clearly can be differences in different designs, using different transistors etc. That doesn't mean that integrated opamps are necessarily bad but, for example, if one is trying to emulate a single ended triode and/or minimize global feedback, the op amp probably not the best design choice.

 

in any case nowhere is it written that different electronic designs will or are intended to sound the same -- that would be a rather boring world for amplifier designers who continue to innovate.

I must apologize for my mistake.  I was thinking in terms of fidelity to input.  In which case with sufficient fidelity there will be boringly the same sound.

 

If we are using the aural equivalent of a box of 64 Crayolas, then I will make my tiger with purple stripes and aquamarine eyes because it is pretty, and no one can say otherwise.

 

If I wanted the sound of a single ended triode or other device with no feedback, I would put it in circuit and follow it with a good opamp.  Of course that kind of design doesn't market well. 

[esldude]

1 hour ago, pkane2001 said:

 

I2S input does not rely on internal DAC clock for timing.  All other points are correct. Please allow me to quote myself  as to the context for why I'd like I2S input:

 

 

Just because a DAC is async does not guarantee low jitter. All it does (if implemented properly) eliminate jitter from the transmission. The onboard clock and the logic at the DAC can introduce its own jitter, which is why I would always want to have a way to bypass the internal clock and internal timing logic. In this context, I2S is the best input. 

Has anyone been able to use an external clock at any level of quality and gotten less jitter than a free running clock close to the DAC?

[esldude]

1 hour ago, pkane2001 said:

 

Imagine that a new clock technology becomes available. I'd like to be able to leverage it without replacing the DAC. (This one is not imaginary, as I just did this with my DAC by placing SU-1 in front of it using I2S into my DAC.)

 

Or, if a new format becomes the standard, such as MQA or XYZ that are not supported by my DAC: I'd like to replace the PC interface with a new one without having to throw away the DAC. Or, perhaps thunderbolt or some other interface becomes dominant on PCs and my DAC only has USB input. Or, if in the future I want to add ethernet connection as an option. etc., etc.

 

So what was jitter without the SU-1 vs with it?

[Dr Tone]

7 minutes ago, esldude said:

And yet further make a list of DACs from $1k to $20k and put on that list whether they use opamps or discrete.

 

Just to clatify I have no expertise in the matter or preference one way or the other.  But the list would definately be interesting.

[pkane2001]

Just now, esldude said:

So what was jitter without the SU-1 vs with it?

 

Better with it, worse without it

 

Ask most SU-1 owners and you'll find that they prefer the sound with SU-1 in the circuit.

 

SU-1 does not change bits, it reclocks the data and converts from one input format to another. So, any changes in SQ produced by it can be explained by:

 

1. Better clock, lower jitter than DAC clock

2. Lower noise due to better electrical isolation, better power supply, better design. This results in lower jitter and less noise transmitted into the DAC

 

Obviously, this all depends on how good the DAC USB/clock/async implementation is relative to the DDC. Mine is OK, but SU-1 is obviously better when connected over I2S.

 

I have no equipment to measure jitter levels. I've tried, but my crappy AD converter (Behringer) is too noisy to be of use for this purpose. While I see jitter bands in the FFT, I can't tell if they are coming from the SU-1/DAC or from the AD converter itself.

[pkane2001]

38 minutes ago, esldude said:

So what was jitter without the SU-1 vs with it?

If you are interested, here's SU-1 jitter performance measured by atomicbob on superbestaudiofriends:

 

 

[jabbr]

2 hours ago, Jud said:

 

Is #3 correctly stated?  If you *are* concerned about crossing clock domains, jitter/phase noise, shouldn't you be strongly considering async USB?

I don't think async USB is anything special in this regard as opposed to eg Ethernet, TCP/IP.

 

I2S has other issues one of which is that the R/L channels are interleaved on the same line. 

[jabbr]

2 hours ago, pkane2001 said:

 

Imagine that a new clock technology becomes available. I'd like to be able to leverage it without replacing the DAC. (This one is not imaginary, as I just did this with my DAC by placing SU-1 in front of it using I2S into my DAC.)

 

If a new clock "technology" becomes available and isn't part of the design of the DAC you are just putting lipstick on a pig. 

 

An analogy might be having a super high quality amp and then sending the final leg of the sound over a string & tin can.

[jabbr]

1 hour ago, esldude said:

I must apologize for my mistake.  I was thinking in terms of fidelity to input.  In which case with sufficient fidelity there will be boringly the same sound.

 

If we are using the aural equivalent of a box of 64 Crayolas, then I will make my tiger with purple stripes and aquamarine eyes because it is pretty, and no one can say otherwise.

 

If I wanted the sound of a single ended triode or other device with no feedback, I would put it in circuit and follow it with a good opamp.  Of course that kind of design doesn't market well. 

Same issue as sonic signatures of different amps. If you don't believe they exist there is no convincing you. I would say that since no real amp is perfect, that it is the pattern of harmonics (nonlinearities) that forms the sonic signature and the details of the circuit that decides which nonlinearities are corrected and which remain.

so regarding "sufficient fidelity" the question becomes what is sufficient. I'd define sufficient (for me) as inaudible (for me) and there we are