1.     Don't be a dotard!
    Read these "rules" AND introduce
    yourself     before your first post
  2. A healthy approach to understanding measurements
    Being true to what the artists intended
  3. After being asked for the 308th time, Merv lists top 5 headphones
    (opinion / entertainment piece)
    Dismiss Notice
  4. $1000 to the first to design a headamp using discrete components with better measurements than Magni 3+
    Dismiss Notice
IMPORTANT: New Members Read This Don't Be a Measurement Fascist Merv's Top 5 Headphones - 2023 Edition $1000 Challenge

Delta Sigma Woes

Discussion in 'Digital: DACs, USB converters, decrapifiers' started by Madaboutaudio, Oct 19, 2015.

Page 2 of 2
  1. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Makes sense. I think in general SAR ADC have higher conversion rates than DS ADCs. So settling times are probably smaller for SAR. But usually DS ADC provide a bit more resolution.

    I thought he used one 1-bit modulator going through a 32-tap discrete digital "flat-top" filter. But I'm not very clear about what was actually implemented.

    Though perhaps not as much, one will likely still have deterministic ultrasonic noise/distortion components due to sample and hold with SAR/R2R, which will necessitate an analog reconstruction filter. Though perhaps of lower order and not as involved.

    I believe one of the things preventing DS from giving more than 24-bits of resolution in some cases is the analog front end, not the DS approach itself.
     
    Last edited: Oct 21, 2015
    ultrabike, Oct 21, 2015
    #21
  2. firev1

    firev1 Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    44
    Trophy Points:
    18
    Location:
    Singapore
    Probably not the best case study but comparing a 4Msps part like the AD1675, the specified settling time for the low latency filter is around 2.65us while the ADS8881 a 1Msps part is specified(it has a Anti-Aliasing filter on the front end) is specified to 290ns settling time. Yeah for most applications resolution of SAR parts are more limited but they seem to be getting really good.

    On the Mola Mola, essentially it works like a regular R2R or SAR in a sense, shift registers and all that jazz, I could be really generalising things. Also a fair bit of nitty gritty that I don't fully comprehend myself.

     
    firev1, Oct 21, 2015
    #22
  3. Madaboutaudio

    Madaboutaudio Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    545
    Trophy Points:
    93
    Location:
    Singapore
    Madaboutaudio, Oct 21, 2015
    #23
  4. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    This is indeed a very interesting paper. Will read. Being a Philips paper it focuses on "classic" 1-bit delta sigma (no MASH or other stuff).

    They have special discusion on dithering of 1-bit delta sigmas, which other folks have found to be problematic as it can result in overload and all kinds of bad juju (for 1-bit delta sigmas, not N-bit ones).

    Thanks! :)
     
    ultrabike, Oct 21, 2015
    #24
  5. firev1

    firev1 Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    44
    Trophy Points:
    18
    Location:
    Singapore
    I have just begun to really crunch the theory behind Laplace and Fourier series and I would like to ask something. Is the Z domain functions presented here the same as like Laplace transform in the complex frequency domain denoted by S? Or is it another transform that I have yet to touch on? Would really love to crunch the derivation to fully understand it, especially the loop functions also presented by the ESS dude for knowledge's sake.
     
    firev1, Oct 22, 2015
    #25
  6. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Z-Transform is indeed sort of the equivalent of the S-Transform (Laplace) in discrete time.

    It is however another transform one will touch on later. It is related to the DTFT (which is not an DFT/FFT, FS or FT). Also, while both have poles and zeros, the Z-plane is a little different from the S-plane.

    When designing analog filters one tends to use Laplace and other random stuff including amps and crap (till it gets too complicated and one hits SPICE or something). When doing digital filters one tends to use Z-transform (till it gets too complicated and one hits some other tools).

    It's not that bad though.
     
    Last edited: Oct 22, 2015
    ultrabike, Oct 22, 2015
    #26
  7. firev1

    firev1 Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    44
    Trophy Points:
    18
    Location:
    Singapore
    Yeah I see that S transform is extensively used especially with regards to application of high order feedback loops, thanks! I will start on it after I get my basics down by next week ^^; Still working on the more basic dynamics problems, barely stepped into control stuff.
     
    firev1, Oct 22, 2015
    #27
  8. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Control is a bit more complex IMO. You start getting into State Space equations for multiple inputs and outputs (and internal states).

    You usually see the Z-transform before you get into control.

    Control you will need before you get into Kalman filters and stuff like that.

    Z-transform is not that bad. Don't think you need full on State Space equations for basic Delta Sigma concepts.
     
    ultrabike, Oct 22, 2015
    #28
Page 2 of 2

Share This Page