1. Register NOW and become part of this fantastic knowledge base forum! This message will go away once you have registered.

1-Bit recorders

Discussion in 'Pro Audio Equipment' started by Cucco, Sep 2, 2009.

  1. Cucco

    Cucco Distinguished Member

    FWIW, I own both the MR-1 and the MR-1000 and use them both all the time. I love them for so many things and have only minor gripes about both of them.

    As far as sound quality - both in PCM and DSD (SR of DSD that they offer).

    The 1-bit recording methodology is a little over simplified by Korg, but it is *essentially* a mapping of increases and decreases in the sinusoidal wave or Fourier Transform. Given its extreme quantity of samples per second, it's able to replicate a near-instant pulse with amazing accuracy. Of course, one issue is that many 1 bit converters achieve this high sampling rate by oversampling, not by native sampling at the prescribed rates. Another issue with 1 bit recording is the inherent digital noise (although, it is "shaped" and sculpted way out into nearly the "Medium Frequency" range. Some claim that this has the potential to damage electronics; however, there is little to no factual or empirical evidence to suggest that this claim has any merit.

    From a purely "sound" perspective on these (few-year-old) units, they are great and their specs (96dB SNR) belie their quality.

    Here's a recording sample of a recording that went straight to the MR-1000 - The chain was as follows:
    Sennheiser MKH 8040 (mains) / Mojave MA-100 (omni flanks)
    True Systems Precision 8
    Dangerous D-Box (summing)
    Korg MR-1000 at 5.6MHz
    Downsampled to 32Bit Float/192kHz for editing (fade in and out only - no effects whatsoever)
    Conversion to MP3 performed by Sequoia's default algorithm.

    (Dead Link Removed)

    Please don't mention the title or composer. This track was not and is not released for the public, but can serve as a decent example.

    Cheers!
    J
     
  2. audiokid

    audiokid Staff

    Hey, cool topic J, sounds pretty nice to me. I wish we lived near each other.
    I have the MR2000S DSD coming this fall. I plan on using it as a replacement for my old... very old DAT and recording a lot of live stuff. I'm more excited after reading this topic and hearing your track.

    Why did you buy yours and what do you love the most about these?


    Thanks for sharing this. (y)
     
  3. Link555

    Link555 Well-Known Member

    This is a very interesting topic indeed. So much to learn!

    Codemonkey was kind enough to give a basic Idea, but I really need to dig deeper into this concept.

    It would be a very interesting PCB layout that had both analog signals and 5.6Mhz clocks bouncing though it.

    At least the physical size of the filters would be smaller.

    Does anyone have any technical links on the process? I did some searching but found only marketing crap.
     
  4. Boswell

    Boswell Moderator Distinguished Member

    This indeed is an interesting topic, and one for which there is much mis-information floating around.

    A quick overview of the principle of operation of single-bit audio is as follows: imagine a square wave switching between 0V and 3V. Considering average values of the wave, because it is symmetrical, the average value is constant at 1.5V and contains no information. Now if you modify the wave to leave the 0-to-3V transitions in the same place but shift the 3-to-0V transitions backwards and forwards in time about their original position, the average value varies. Make the time-shifting vary with your input signal, then the average value of the output will be a representation of the input signal. In most versions of single-bit systems, both transition directions (high-to-low and low-to-high) are shifted.

    The "1-bit" or "single-bit" name comes from there being only two levels (0V and 3V in our example), which in amplitude terms can be represented by the state of a single bit. All the signal information is contained in the times at which that bit changes state and not in its amplitude..

    From this can be seen that the limit to the resolution of a single-bit audio system is the coarseness of the time instants at which decisions are made whether or not to change the state of the single bit. In principle, the higher the sampling rate, the higher the resolution. In practice, there are numerous other effects to consider to make a usable system, but the basic principle of operation is as described.

    Incidentally, the front end of a single-bit system is identical to an over-sampled PCM system. The single-bit (DSD) stream is extracted at an earlier stage in the digital filtering process than for PCM. Also, the mix of precision analog circuitry and high-speed clocks is not new for DSD, and has been a major challenge in high-precision converters for many years. My day job includes designing these types of systems and making them work in audio and industrial fields.
     
  5. Codemonkey

    Codemonkey Well-Known Member

    It appears I was spectacularly wrong. Sorry link.
     
  6. Link555

    Link555 Well-Known Member

    No worries Codemonkey. I still think your idea is a cool one.

    Thanks Boswell so it sounds a bit like pulse width modulation triggered by the input audio signal? This is probably not true but it does sound bit like it.
    So the front end filtering could be reduced for a purely DSD system?

    What types of other issues have your designs encountered?

    Thanks again boswell, just trying to understand.
     
  7. Boswell

    Boswell Moderator Distinguished Member

    Not usefully. The filtering I talked about is digital filtering, primarily a cascaded integrator comb filter, often found inside the ADC chip but can be done by external DSP. (Reference)

    Any design has to work to a specification and within a list of constraints that includes a manufacturing budget. Some of the most difficult design constraints are not directly audio related, such as electromagnetic compatibility (EMC), power supples, mechanical housing, and testability, not to mention ease of manufacture and having to second-guess the availability of all the chosen components for at least 5 years after a design goes into production.

    At the audio level, the big challenge is trying to achieve the noise and distortion figures demanded by the design specification, and this is where mixed-signal problems (low-level audio co-existing with high-speed digital circuitry) can occur. Even simple things like the exact point at which analog and digital grounds are inter-connected can have a big impact on the resulting performance.

    Open up a cheap piece of audio gear and also a well-respected high-quality piece of a similar function, and at first sight they may appear to be much the same: a printed circuit board with I/O jacks and some user interface controls. Examine more closely, and you may see that the high-quality equipment uses 6- or 8-layer circuit boards, the boards are several revisions into their history, the power supply is well separated, screened and has separate looms for digital and analog power. It will smack of attention to detail. This is the only way to get quality results.
     
  8. Link555

    Link555 Well-Known Member

    Thanks for the response boswell do you know of any good references for the way 1-bit ADC's work?
     
  9. Boswell

    Boswell Moderator Distinguished Member

    This is an article by Korg outlining their 1-bit technology.
    Here is a more interesting paper from a group at QMC in London proposing various methods of applying EQ and effects to single-bitstreams. [Page 2 does not show up on my PC; you may have better luck]
     
  10. Link555

    Link555 Well-Known Member

    Thanks again Boswell. The filtering aside Is there a site you know of, that talk about the actual ADC conversion aside from the filtering section?
     
  11. Boswell

    Boswell Moderator Distinguished Member

    Try this from Maxim. Think of DSD data being taken out of the ADC at the point where the article shows "1-bit data stream". The details of a true DSD converter are slightly different, but the principles are similar enough for this to be a useful description.

    The DSD Pro Audio site has an explanation page

    For what it's worth, there is also a Wikipedia page with a jumble of information about DSD.
     
  12. Link555

    Link555 Well-Known Member

    Thanks! The maxim link is exactly what I was looking for!

    In your work have you designed Sigma-Delta ADCs at chip level?
     
  13. Boswell

    Boswell Moderator Distinguished Member

    No, unfortunately. Mixed-signal chip design is a very specialized skill that I regret I have never had the opportunity to acquire. However, I have used many of the different ADC architectures in my equipment designs including the type where you can get in and re-program the internal digital processing to adapt the basic design for use in new and sometimes weird applications.
     

Share This Page