Article: Understanding Digital Audio Measurements

[Blumlein 88]

Unfortunately, that is the limitation. But the R2R ladder for me is "better". Too much math for the Delta Sigma for me ... hahaha ... makes my head ache. Further, designing good loop response ... that is the challenge ...

Clearly a challenge that has been met. In essence the world of digital audio makers found meeting various challenges to approach the potential of DACs was done better and less expensively with sigma delta than with R2R.

 

[Blumlein 88]

Feedback loop for correction ...

That is a potential problem. One that has been solved. There is no law that feedback is impossible to work. Most of the electronics in the world use lots and lots of feedback. All things considered it is a boon to performance. There are DACs and ADCs at much higher frequencies than audio. Audio is slow in such a world.

 

[rascal101]

Clearly a challenge that has been met. In essence the world of digital audio makers found meeting various challenges to approach the potential of DACs was done better and less expensively with sigma delta than with R2R.

No feedback loop for R2R. I like this approach. No need for checking Gain Phase Margin. Less engineering but more research into materials to make very precise resistors -- maybe some time in the near future we can get "better" resistors and have better R2R ladder DAC.

 

[Blumlein 88]

Clearly a challenge that has been met. In essence the world of digital audio makers found meeting various challenges to approach the potential of DACs was done better and less expensively with sigma delta than with R2R.

No feedback loop for R2R. I like this approach. No need for checking Gain Phase Margin. Less engineering but more research into materials to make very precise resistors -- maybe some time in the near future we can get "better" resistors and have better R2R ladder DAC.

What benefits do you expect? Effective linearity is down possibly to 24 bits or below 20 bits. So low it becomes hard to measure. Frequency response is very flat down to DC. Distortion is vanishingly low. Noise is very low. What areas still stand to benefit that would be heard?

 

[BDWoody]

Less engineering but more research into materials to make very precise resistors -- maybe some time in the near future we can get "better" resistors and have better R2R ladder DAC.

To what end? I can see it as a curiosity/challenge, but otherwise I don't see the broad appeal to those who just want the box to do it's job.

I am just writing because the R2R ladder is not being given the response it deserves.

I guess I'm wondering what you think it deserves. It is an outdated technology that offers no benefit I can see, unless you are claiming that it 'sounds better,' whatever that may mean. If so, some evidence that a difference could even be determined between good examples of R2R vs D/S DACs might be a good start.

I could build a steam engine for my truck, and maybe get it to work great, but why?

 

[rascal101]

To what end? I can see it as a curiosity/challenge, but otherwise I don't see the broad appeal to those who just want the box to do it's job.



I guess I'm wondering what you think it deserves. It is an outdated technology that offers no benefit I can see, unless you are claiming that it 'sounds better,' whatever that may mean. If so, some evidence that a difference could even be determined between good examples of R2R vs D/S DACs might be a good start.

I could build a steam engine for my truck, and maybe get it to work great, but why?

Looking into DAC architectures you can argue that R2R is an "old" technology but it is still the simplest out there. The key for R2R is precision resistors. Delta Sigma is a little more complex in that several factors need to come into play -- as stated from Analog Devices AN-283 -- oversampling, noise shaping, digital filtering and decimation. Over the years this has been perfected so it is now that this technology is showing its benefits. R2R has been stuck because there is no more or very little research into new materials that can provide very precise resistance in an economical package. The key here is research. But there is enormous benefits for the simple R2R in that if such materials can be developped Delta Sigma will have similar fate as it is too complex.

 

[BeeKay]

But there is enormous benefits for the simple R2R in that if such materials can be developped Delta Sigma will have similar fate as it is too complex.

Quite a statement. Any scientifically relevant facts to support it? Or more of an opinion?

 

[rascal101]

Quite a statement. Any scientifically relevant facts to support it? Or more of an opinion?

Not a statement. It is facts. Just read DAC architectures application notes from Burr Brown or Analog Devices or TI etc.

 

[rascal101]

Look into how your Delta Sigma vs your R2R ladder is being manufactured from the statements from the designer/manufacturer. Or just look into the block diagram. As stated earlier you can check AN-283 from Analog Devices or read the D/A conversion handbook from Analog Devices. It is all there.

 

[rascal101]

Check below link to read that the very first DAC architecture shown in this 1986 D/A conversion handbook is the R2R ladder on page 199

https://www.analog.com/media/en/training-seminars/design-handbooks/Analog-Digital-Conversion-1986/Part2.pdf

 

[danadam]

Looking into DAC architectures you can argue that R2R is an "old" technology but it is still the simplest out there. The key for R2R is precision resistors. Delta Sigma is a little more complex in that several factors need to come into play -- as stated from Analog Devices AN-283 -- oversampling, noise shaping, digital filtering and decimation.

Wouldn't you still need oversampling (which implies digital filtering) even with R2R?

 

[rascal101]

Wouldn't you still need oversampling (which implies digital filtering) even with R2R?

Sampling is a basic requirement for any DAC. Nyquist theorem requires 2X of max frequency. However, the R2R architecture itself does not have this. The input to the R2R architecture requires this.

 

[NTK]

Different DAC architectures have their strengths and weaknesses and are suitable for different applications.

 

[Sokel]

Different DAC architectures have their strengths and weaknesses and are suitable for different applications.
View attachment 440104

The same TI did a 20-bit one recently though:

https://www.ti.com/lit/ds/symlink/dac11001a.pdf?ts=1743196624217

It does quite well at Schiit Yggdrasil, ‘More is Less’ version:

(link)

 

[pkane]

Sampling is a basic requirement for any DAC. Nyquist theorem requires 2X of max frequency. However, the R2R architecture itself does not have this. The input to the R2R architecture requires this.

Proper sampled waveform conversion to analog requires a reconstruction filter at the output. That's what makes the difference between a stair-step analog output with added imaging distortion vs. a smooth and clean analog waveform.

 

[rascal101]

Proper sampled waveform conversion to analog requires a reconstruction filter at the output. That's what makes the difference between a stair-step analog output with added imaging distortion vs. a smooth and clean analog waveform.

Sampling is done at the INPUT due to Nyquist criteria --> What I said and ... reconstruction is done at the OUTPUT --> What you said. We are both correct. No need to be emphasizing in italics "at the output". This is Basic D/A conversion.

 

[pkane]

Sampling is done at the INPUT due to Nyquist criteria --> What I said and ... reconstruction is done at the OUTPUT --> What you said. We are both correct. No need to be emphasizing in italics "at the output". This is Basic D/A conversion.

We are talking about DACs and not ADCs, correct? And no, sampling is done at the ADC and not at the DAC input.

 

[rascal101]

We are talking about DACs and not ADCs, correct? And no, sampling is done at the ADC and not at the DAC input.

Going to the input of the DAC. There is no sampling inside the DAC.

 

[rascal101]

Sorry but I am not a native english speaker. In our language we say "papasok" which means an action prior to coming in.

 

[danadam]

Sampling is a basic requirement for any DAC. Nyquist theorem requires 2X of max frequency. However, the R2R architecture itself does not have this. The input to the R2R architecture requires this.

Sorry, not sure what you're trying to say here or why do you bring up sampling.

I was answering to your mention of over-sampling. You said that R2R is simple and that DS is complex because of "oversampling, noise shaping, digital filtering and decimation". But one reason for oversampling is so that the analog filters, anti-aliasing in ADCs and reconstruction in DACs, don't have to be steep. And that's the same both for R2R and DS.