1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Thoughts on a bunch of DACs (and why delta-sigma kinda sucks, just to get you to think about stuff)

Discussion in 'Dedicated Source Components' started by purrin, Dec 5, 2013.
First
 
Back
178 179 180 181 182 183 184 185 186 187
189 190 191 192 193 194 195 196 197 198
Next
 
Last
  1. purrin
     
    I wouldn't mind owing one. I'd stick it a bit above the SFD-1mk2/SE+ upgrades but below the Gen V. Really good DAC. Better than any of the modern ladder DACs from the big names.
     
  2. Clemmaster
     
    That would be interesting to reproduce this with the Yggy (with proper dithering / noise shaping), to see what's actually left out in the process.
     
    Can someone explain the rational behind the 0.01% resistors limit the resolution to 14.5 bits? Do they even know how these resistors are used in this converter to make such a claim? Are all the resistors value the same? Can't the ladder be broken down into different segments (MSB / LSB and, possibly, intermediary segments), like the new Metrum Pavane?
     
  3. prot
    http://youtu.be/nLEhfieoMq8. A pretty nice demo & explanation for the effects of lowering bit rates. There's a good chance that 16 to 14 bit with proper dithering will not be audible. With a very good recording, a very resolving system and high sensitivity speakers it may be though.

    As to your second Q, not sure about that relation between resistor values and bit depth. That math sounds kinda fishy. Maybe someone can clarify
     
  4. Sorrodje
    @purrin : Ok I see.  Soekris who made R2R ladder dac on DIYaudio says that the Totaldac does not use a sign magnitude architecture if I understand well ( a friend gave me this informations) but my friend said me that some totaldac measurements ( THD ? ) don't match wit a 14 bits resolution... Work in progress to understand this.
     
  5. Clemmaster
    Thanks for the link, will check it out.
     
    As for the resistor accuracy and bit depth, Wikipedia gives some explanation : http://en.wikipedia.org/wiki/Resistor_ladder
    But they also mention that R-2R is not the only way to build a ladder DAC. I guess we would need to dig some more to get an understanding of how the TotalDAC is implemented.
    Measurements a la Stereophile would be great, too.
     
  6. drez
     
    Yep, I think Stereophile style measurements would be what's needed.  Seems local hifi store will be taken over my company that distributes Schiit, so hopefully they will have Rag/Ygg for demo.  Less for me to lose than using the evaluation period of Schiit.
     
  7. Clemmaster
    This video is very cool: http://xiph.org/video/vid2.shtml
     
  8. purrin
     
    Six Loons seems to imply the Total DAC is an R / 2R architecture. Just plug in randomly generated R and 2R resistor values according to a specific tolerance  into a speadsheet and with the formulas / calculations for that architecture.
     
     
    You never know, I may have already done it. Maybe even modeling specific D-A chips of yesteryear according to generated INL error plots. You never know.
     
  9. Clemmaster
     
    I would be interested in seeing your generated INL plots.
    I did generate a transfer function for the DAC8581 based on the LE plot and a lousy CV algorithm, but the plot's resolution is not good enough to get a decent TF [​IMG]
     
  10. prot
    Thx for that link, seems clear now. And if my math isnt too rusty 0.01% accuracy should be enough for ~13.3 bits (100 / 2 power noBits = neededAccuracy). Which is indeed quite lame.

    However wiki also says "integrated circuits may push the number of bits to 14 or even more". Anyone who can clarify that part?
     
  11. Clemmaster
    I think they mean we can achieve greater accuracy with the technology used to manufacture integrated circuits (take a PCM1702/4, for instance), which is more precise than for a discrete resistor, thanks to techniques like laser trimming, etc.
     
  12. arnaud Contributor
    Well, if 13.5 bit effective resolution is what makes the TotalDAC sound so good, so be it! I have just never ever heard that much detail in my CDs and it's not like I have been using ladder dacs all along. Too bad also the TotalDAC designer can't step in and explain how he got 140dB effective dynamic range if all we got are 14bits. Maybe I can try to get an explanation from him. I guess I get confused between dynamic range and precision.
     
    I still think there's a lot of babbling around to justify the Iggy's design (be it the effective resolution of the ladder or the zillion taps long OS filter) and, maybe, be even more assertive on how it is going to be the best thing since sliced bread, killing each and every dac ever made :wink:.
     
    I will be very very curious to give it a run, in my rig, once the BHSE is in the house for example :)
     
    arnaud
     
  13. Sorrodje
    The totaldac is definitely a r2r ladder DAC. Soekris with his diy board or msb seems to achieve a 24 bits dynamics or resolution with sign magnitude architecture. V.brient said me the totaldac fully use 24 bits dynamics too but didn't tell me if he used the same method.
     
  14. Golfnutz
    Arnaud, he did try to explain it on a different forum, as a similar question was asked.
     
    Q - Member. "If 0.01% is good for 14-bits of discrete voltage steps, what are the other 10 bits used for? Does the 24-bit ladder become a graceful 14-bit down converter or is there something else in play that utilizes the other 10 bits ???"
     
    A - Totaldac. "In audio 24bit resolution is used to get the ability to resolve the large signals and also the very small signals, that's the dynamic capability. Such a R2R DAC has the ability to produce high level with 0.002% low distortion but also very low levels with a very good resolution. The FFT spectrum given in my web site shows the ability to resolve the very small signal with a very low noise which a 14bit DAC couldn't resolve, not even a 16bit nor a 20bit DAC, the noise floor is achievable only by the best 24bit DACs. For large signals (0dBFs) the LSB resolution is not necessary, and simply impossible to get by any DAC, calculate the distortion that would be admitted by a mathematically perfect 24bit DAC, it would be a distortion impossible to get and also completely useless. In audio people often forget that dynamic is the characteristic number one in audio.
    For example take the ear or a good microphone or a good loudspeaker, their distortion is higher than most DACs and than most transistor amplifiers but their dynamic is extremely high. I tried to make a DAC that fits these accoustics characteristics."
     
    arnaud likes this.
  15. Sorrodje
    @Golfnutz: Vincent gave me the same answer but the question is how the totaldac can achieve this 24 bits depth.
     
    I borrowed yesterday a HD800 and a Metrum Octave and give the A1 more hours of listening in next days.
     
First
 
Back
178 179 180 181 182 183 184 185 186 187
189 190 191 192 193 194 195 196 197 198
Next
 
Last

Share This Page