[gyrodec]

They are great, assuming you never want to go past 16bits. It become impossible to laser trim the DAC resistors accurately enough, and you end up with very servere destortions creaping in in the low levels under 16bit. Even at 16 bit the non-linearity is getting rather bad, which is why the D/S DACs were created (not that they don't have their own issues).

If you wanted to create a discrete R2R DAC, and hand select the resistors, then you could get there I guess.

 

[Roam]

Quote:

Originally Posted by gyrodec /img/forum/go_quote.gif They are great, assuming you never want to go past 16bits. It become impossible to laser trim the DAC resistors accurately enough, and you end up with very servere destortions creaping in in the low levels under 16bit. Even at 16 bit the non-linearity is getting rather bad, which is why the D/S DACs were created (not that they don't have their own issues).

And you'd be wrong. See the Stereophile measurements of the Simaudio Andromeda which uses the PCM1704, which is an R2R DAC. Effectively unmeasurable linearity errors with 16 bit data.

 

[gyrodec]

Actually, its linear to nearly 19bit, so I'm definately a little out of date with my info. I got the info from Bob Stuart, the digital guru at Meridian - I worked for KEF and they used to own Meridian. But this was when the 208 replaced the 207, so over 20 years ago, but I had imagined the production processes could not have improved enough to change that much.

Intrestingly the 1704 isn't a simple R2R. Internally its a dual-differential 23bit R2R converter with an additional current source to add 24th bit current dither. BB datasheets are very well written and very informative.

 

[sheldaze]

Do both R2R and S-D DACs filter bits through the process known as "decimation"?

 

[evillamer]

  Do both R2R and S-D DACs filter bits through the process known as "decimation"?

 
If you are talk about at the actual digital to analog conversion level, then only Sigma Delta does decimation.
 
Sigma Delta is a derived form of Delta Modulation
https://en.wikipedia.org/wiki/Delta_modulation
 
Input Waveform data is analysed and assigned either a positive(+) or negative(-) value at a defined sampling rate. 
 
 https://en.wikipedia.org/wiki/Continuously_variable_slope_delta_modulation

Like other delta-modulation techniques, the output of the decoder does not exactly match the original input to the encoder.

Note the link by wikipedia at the "Does not exactly match"
 

 In information technology, lossy compression is the class of data encoding methods that uses inexact approximations (or partial data discarding) to represent the content. 

 

[sheldaze]

   
If you are talk about at the actual digital to analog conversion level, then only Sigma Delta does decimation.
 
Sigma Delta is a derived form of Delta Modulation
https://en.wikipedia.org/wiki/Delta_modulation

Outstanding! Thanks for the fast response!
 
I am unclear on the science behind R2R and S-D - I am currently trying to read as much as I can.
 
I have just purchased an R2R DAC, and reading through the threads (on S-D DACs), they stated that one of the issues is the bits must go through a process known as decimation. It is, according to the post, at this processing point that the S-D DACs can change their sound quite dramatically. This is due to the math behind the process using a Fourier transformation, and that math cannot optimize for both frequency and time domain, simultaneously. Thus the DAC designer must make certain decisions, based on his/her intent in how the DAC should sound, at this point of the chain.
 
I have not read through the links you posted, but it sounds as-if this design decision is not required in an R2R DAC. I will read further, but again, thanks!

 

[aqsw]

What a stupid thread. Implementation is everything. D S is usually smaller and cheaper. Properly implemented they are as good , if not better. If you think technology regresses, you may disagree with me.

 

[kothganesh]

^. OK, agree to disagree.

 

[mikoss]

Lol also disagree... Have you actually heard an R2R DAC in comparison with DS counterparts of the same price? As mentioned above, a well implemented R2R has a level of accuracy DS chips simply cannot reproduce, both on paper and in sound. I do agree that there is far more to a DAC than the chip doing the conversion, but DS is simply inferior if the designer is given free reign.

 

[Wildcatsare1]

https://parttimeaudiophile.com/2017/10/08/rmaf-2017-more-must-see-rooms/ Some excellent press on a great R2R DAC,

 

[rkt31]

Lol also disagree... Have you actually heard an R2R DAC in comparison with DS counterparts of the same price? As mentioned above, a well implemented R2R has a level of accuracy DS chips simply cannot reproduce, both on paper and in sound. I do agree that there is far more to a DAC than the chip doing the conversion, but DS is simply inferior if the designer is given free reign.

Any link of measurements of a r2r dac measuring better than the totl measuring ds dacs like smsl su 10 or topping d90se?