The well segmented dac-24 bit 384khz thermometer/r2r sign-magnitude discrete dac

Forum Digital line The well segmented dac-24 bit 384khz thermometer/r2r sign-magnitude discrete dac

Viewing 31 reply threads
  • Author
    Posts
    • #1596
      The Well Audio
      Participant

      Since the original thread on diyaudio.com has been deleted the topic is renewed here.

      The DAC Lite is a multibit discrete DAC. The resistor ladder network is segmented, the first 3 MSBs are thermometer decoded while the other 20 LSBs are R2R decoded.
      The DAC implements a sign magnitude architecture to get bipolar output, so there are 2 x 23 bit ladder for each channel, one for the positive rail and one for the negative rail.
      The dirty signals like bit clock and data are optically isolated to avoid any interference with the latch signal wich drives the DAC switches.
      The bit clock is stopped as soon as data have been loaded into the shift registers, so every time the DAC switches to a new word the bit clock cannot affect the latch.
      It uses a custom protocol provided by the TWSAFB-LT FIFO Lite.
      The FIFO Lite also provides a digital calibration for the DAC Lite in order to compensate the error due to the resistors tollerance used in the ladder.
      It’s a voltage output DAC which provides 1.1V to 1.8V rms depending on the Vref power supply, from 3V3 to 5V. The output impedance is 625 ohm, so the DAC can drive directly most amplifiers.
      There is an optional buffer if the load wich has to be driven is lower than 20K.

      Attachments:
      You must be logged in to view attached files.
    • #1605
      Crom
      Participant

      Congratulations Andrea – looking forward to some interesting discussion.

    • #1608
      jmmbarco
      Participant

      Hi Andrea!. Hope we can finally get useful info for our projects without trolling!

    • #1611
      Ilgavro
      Participant

      Hi Andrea!. Hope we can finally get useful info for our projects without trolling!

      Exact! Maybe here we learn something!

    • #1614
      Ivan
      Participant

      Hi guys. Excellent news! Congratulations.

    • #1690
      pinnocchio
      Participant

      Yes, on DIYaudio it was a total nonsense about trolling. We can finally get details here without issues!

    • #1693
      pearse
      Participant

      DAC Lite continuing to delight here, I went through a period where the sound was not as good but I had been experimenting with cables and power supplies.

      After any change it needs several days to settle down it seems in my system.

      I added a 3 channel Paul Hynes SR7 power supply to the front end SD playback units and added a Musetec USB100 Interface on channel 2 of the Fifo recently.

      I also built the TWTTB Vacuum Tube Line Buffer and added it to the system in the past week, early days but I think I will be keeping it in the system.

       

      I plan on finally transferring the build into a case in the coming weeks now that I am happy with all the tweaking.

    • #1697
      minionas
      Participant
      pearse Said

      I also built the TWTTB Vacuum Tube Line Buffer and added it to the system in the past week, early days but I think I will be keeping it in the system.

       

      Hi,

      What it brought to your system? Im wondering about its transparency (if details or resolution are a bit lost) and its “tubiness” (if it add any tube goodness?

    • #1699
      pearse
      Participant
      pearse Said

      I also built the TWTTB Vacuum Tube Line Buffer and added it to the system in the past week, early days but I think I will be keeping it in the system.

       

      Hi,

      What it brought to your system? Im wondering about its transparency (if details or resolution are a bit lost) and its “tubiness” (if it add any tube goodness?

      Hiya,

      No loss in detail or resolution and it has taken over a week to settle in.

      What I am hearing now is a little more weight and authority on bottom end and highs are a little sweeter without any loss of detail, you could say a little tubey on the highs.

      Listening to Steely Dan, Two Against Nature at the moment and it sounds amazing.

    • #1728
      Sebastian
      Participant

      Hi Andrea, I have a question regarding Vref and the DAC’s output (1.1-1.8Vrms).  Is there an ideal voltage for Vref (best sound, lowest distortion, etc)?  I ask, because I am able to accommodate for variance with my preamp, and now that I am planning, it would be helpful to know that. I won’t be using the buffer, as the input impedance of the preamp is high.

       

       

    • #1733
      The Well Audio
      Participant

      Hi Andrea, I have a question regarding Vref and the DAC’s output (1.1-1.8Vrms).  Is there an ideal voltage for Vref (best sound, lowest distortion, etc)?  I ask, because I am able to accommodate for variance with my preamp, and now that I am planning, it would be helpful to know that. I won’t be using the buffer, as the input impedance of the preamp is high.

       

       

      We have used 4V and 4V7, and in our opinion there wasn’t any SQ difference.
      We have powered the DAC Lite with batteries, but it looks like someone preferes shunt regulators.

      We will test shunt regulators in the next listening session.

    • #1741
      pearse
      Participant

      I added 2 x 500F paralleled supercapacitors (1.6-7v in each Cap) on each of the 3.3v Vref supplies and I found a very nice improvement in low to mids body and presence in my system. I also added pairs of supercapacitors on the VDD lines but did not hear any change in sound with these in place.

      Spent some time sorting Caps to find pairs that charged and discharged evenly and so far so good with balance maintained in pairs. I never shut down power so no worries about startup but I have disconnects in place in case I ever need to shutdown and recharge the Caps on startup.

      Will monitor Caps voltages over the coming weeks to see if I need to add balancing resistors.

    • #1742
      pearse
      Participant
      pearse Said

      2 x 500F paralleled supercapacitors (1.6-7v in each Cap) on each of the 3.3v Vref supplies

      The above did not make sense 🙁

      (2 x 500F series set) in parallel on each of the 3.3v Vref supplies.

    • #1790
      Sebastian
      Participant

      Hello Andrea,

      is there a cap DC blocking cap at the DAC output?

       

      S.

    • #1792
      The Well Audio
      Participant

      There isn’t a DC blocking cap at the DAC output because the DAC Lite is a bipolar DAC (sign-magnitude architecture), so there is no DC issue.
      In other words, when there isn’t any input signal all the DAC switches are tied to ground.

    • #1793
      Sebastian
      Participant

      Thank you Andrea!

    • #1797
      Hifoli
      Participant

      Hi Andrea,

      could you please elaborate a bit on the calibration of the DAC? How much impact does it have, e.g. a difference in the output signal would be super interesting to see.

      BR
      Happy WE
      Oli

    • #1800
      The Well Audio
      Participant

      Due to the tolerance of the resistors used in the ladder, the accuracy of the DAC without any calibration is around 15 to 17 bit, depending on the resistors mixing.

      In the attached pictures you can see the voltage measured at DAC output for each bit and for all the bits.
      A dedicated Windows app automatically sends the signals to the DAC and measures the output voltages for each bit.
      The Windows app also manages a 7 1/2 digits DMM, so the measurement is fully automatized.

      Then the application computes the measured data and provides the calibration file in order to correct the accuracy error.

      Finally the calibration file is loaded into the FIFO lite, which provides the needed correction for each bit.

      This way the accuracy of the DAC is increased to 22-23 bit.

      Attachments:
      You must be logged in to view attached files.
    • #1813
      lasercut
      Participant

      Felt I should give small initial update because things are moving slow and might be a while before everything is optimised enough to give a more conclusive and thorough report.

      DAC lite is set up in semi-rough state, with basic enough power supplies (1 rail and 2 ultrabib between XO supply and FIFO Aux+DAC supply).
      Using PPG 22MHz for now,  DRIXO 5.6MHz eventually.
      CS8416 toslink to I2S for FIFO input

      Still, I’m hearing an almost unbelievable level of detail from the DAC already…
      “detail” doesn’t even seem like the right word for it, familiar tracks almost sound unfamiliar there is so much new information being exposed… but in a natural, not overly forward way, which has got to be one of the most common problems in DACs I’ve experienced.
      dam1121 for example, genuinely very resolving compared to most with a realistic presence (the reason I chose to buy dac lite as they share similar architecture) but had a particularly bright and forward character.
      It was forgivable, but very glad not to have to deal with it.

      The dac lite has a sort of dense tonality that I didn’t love it at first but is now starting to grow on me, though with burn-in time still in single digits and an endless amount of things left to experiment with and optimise it makes little sense forming any judgements yet.

      It is very clear already this system has some huge potential, had plenty of faith it would but you never know until you listen for yourself.

      How much the DAC, XO and FIFO are each contributing will be interesting to see also.

    • #1824
      lasercut
      Participant

      I’m noticing shortly after turning everything on the DAC puts out 2.5V DC for about a second, audible as 2 pops.

      Is this normal?

      Only thing I could think of is not following any turn on sequence for XO, FIFO and DAC supplies, everything turns on simultaneously from mains switch.

      • This reply was modified 1 month, 3 weeks ago by lasercut.
      • This reply was modified 1 month, 3 weeks ago by lasercut.
      • This reply was modified 1 month, 3 weeks ago by lasercut.
    • #1828
      Kazuma
      Participant

      Generally it is recommended as a rule of thumb, that you turn on equipement from source to end device (amp being the last to be turned on).

      And turn off everything in reverse order (first amp, then other devices up the chain).

      I think it was put up in practice to offset such possible unwanted effects, as you described, between different equipement.

    • #1829
      The Well Audio
      Participant

      I’m noticing shortly after turning everything on the DAC puts out 2.5V DC for about a second, audible as 2 pops.

      Is this normal?

      Only thing I could think of is not following any turn on sequence for XO, FIFO and DAC supplies, everything turns on simultaneously from mains switch.

      When the DAC starts, their switches are in unknown state until the FIFO begins to send data.
      Some DC at the output of the DAC is possible, because the FIFO takes longer than the DAC to start up.

      So Kazuma’s suggestion is a good practice, the ampli should be the last to be turned on in order to avoid DC to reach the speakers.

    • #1832
      lasercut
      Participant

      Ok, it will be left on most of the time anyway so not a big deal.

      Also noticed it wont play at 176kHz, but it’s more likely a problem with cheap CS8416 board, will confirm 176khz operation eventually when using SD player as source.

       

       

    • #1833
      lasercut
      Participant

      I was wondering for volume control, could you place a shunt resistor on the output to form voltage divider with output impedance  of ladder?

      Assuming this is not detrimental to DAC, it seems preferable to adding series resistance.

       

    • #1834
      The Well Audio
      Participant

      The DAC Lite plays up to 176/192 kHz with 5-6 MHz oscillators, and up to 352/384 kHz with 11/12 MHz oscillators.
      Please, check if the CS8416 runs at 176 kHz.

      We can place a shunt resistor at the output of the DAC in the next revision. Maybe it will takes a few months.

    • #1924
      Sebastian
      Participant

      Hello Andrea, I have two questions:  first, I believe you wanted to do some experiments with different power supply options on Vref.  Have you been trough this?

      Then, I have a few Mumetal sheets and I was wondering if you have any suggestions to shield FiFo/dac and theUSBtoI2S (Jlsounds in my case).  I think I will have all three in the same chassis and a long USB cable to keep the RPi/Odroid C2 source away.

       

    • #1925
      lasercut
      Participant

      To update the 176kHz playback problem was the CS8416 interface. it was one of those very unusual design choices you often see on generic chinese boards like this, a cap to ground slapped on the output of the toslink receiver, plays fine after removing it.

    • #1929
      The Well Audio
      Participant

      We have experimented a little with batteries and a pair of shunt regulators.
      It looks like the DAC sounds better using shunt regulators on Vref.
      We have used our own tracking regulators (not yet available) and a shunt regulator from the member ilgavro.
      Not yet tested Salas shunt, but I have read someone is using it as Vref and the DAC sounds fine.

    • #1979
      Sebastian
      Participant

      Hi Andrea,

       

      I am preparing the power supplies.  For the digital part of the DAC (+/- 3.3 to +/- 5v), the clean side of the FIFO (3.3 to 5V) of the DRIXO (12-18V) is there any advantage for any particular voltage in those ranges?

      I will be using the UltraBib 1.3 and while it can be brought down to 3.3v with special parts, it is easier to get 5v.  So If there is no penalty, I wold go with 5v for the first two.  For the last I can aim at whatever voltage in the rage, buy I have to take into account the transformers sec. VAC.

       

      For Vref I will also use the Ultrabib, but in this case I might have to do down to +/-3.3V because of my systems gain structure (not yet sure on this).

       

      Best regards,

      S.

    • #1980
      The Well Audio
      Participant

      Hi Sebastian,

      to power the DRIXO oscillators I suggest 16-17 VDC, as clean as possible.

      The voltage range of the power supply of the DAC Lite affects the output voltage only. The greater the power supply voltage the higher the voltage output of the DAC.
      We usually power the DAC Lite with 4VDC, both digital and Vref.
      There is no problem powering the dital section with 5V and Vref with 3V3. But if you use 5V for the digital section of the DAC you have to use the same 5V for the clean section of the FIFO buffer (J18).

      Hope yhis helps.

      Andrea

    • #1981
      Sebastian
      Participant

      Andrea, yes it does!

      I will then be using 5V for digital in DAC and FIFO, 16-17V on clock, and Vref yet to determine.

    • #1987
      G600
      Participant

      Andrea told me that you don’t need pos and neg rails, full positive is fine!

Viewing 31 reply threads
  • You must be logged in to reply to this topic.