Connect with us

Bit-resolution and Clipping?

Discussion in 'General Electronics' started by Radium, Dec 21, 2003.

Scroll to continue with content
  1. Radium

    Radium Guest

    Can a wave file with stronger bit-resolution handle more decibels
    w/out clipping than a wave file with weaker bit-resolution?
  2. Dave Platt

    Dave Platt Guest

    Can a wave file with stronger bit-resolution handle more decibels
    Yes. The dynamic range is proportional to the number of bits,
    assuming that you're talking about linear PCM. The dynamic range is
    approximately 6 dB per bit.

    So, with more bits, you can handle larger signals at the "loud" end
    (without clipping) or smaller details at the "quiet end" (without
    losing them below the quantization-noise floor), or both. It all
    depends what signal levels you choose to treat as "full scale" and
    "least significant bit".
  3. Radium

    Radium Guest

    So a digital audio device with a bit-resolution of 100-bit can handle
    up to 600 dB w/out clipping. Right?
  4. Dave Platt

    Dave Platt Guest

    The dynamic range is
    In theory, yes, in the absence of noise and other dynamic range limits.

    In practice, it's useless to try. Even 32 bits is more than you can
    usefully expect to represent within reasonable physical limits. I
    think I recall reading an analysis a few years ago (JJ's, perhaps?)
    which pointed out that if you try to use a 32-bit linear PCM system,
    and set the least significant bit down to somewhat below the human
    ear's ability to hear any sound in a perfectly quiet room, then a
    full-scale 32-bit CD-quality signal would release so much energy that
    it would vaporize the listening equipment (and probably the listener,

    Doesn't make for a relaxing evening... a log in the fireplace, a nice
    glass of wine, and a pleasant CD on the
  5. Only if the analogue gear it's attached to can cope with this range!
  6. Radium

    Radium Guest


    I forgot to add that this theoretical digital audio device also
    contain a 100-bit DAC. Sorry.
  7. Which will of course have an output dynamic range of less than 22
    bits, as with all available '24 bit' DACs...................
  8. Radium

    Radium Guest

    You mean the DAC?
  9. Radium

    Radium Guest

    Speaking of digital vs. analog, analog clipping isn't as harsh as
    digital clipping. However, CDs have a dynamic range of approx. 90dB
    and tapes have it at about 60dB. This means a CD can handle 90 dB
    w/out clipping, while a tape can only handle 60dB. Right?
  10. Radium

    Radium Guest

  11. Dave Platt

    Dave Platt Guest

    Which will of course have an output dynamic range of less than 22
    Several reasons. Noise is one.

    All electronics generate noise. One type, known as "thermal" noise,
    occurs any time you have a resistance - the amount of noise depends on
    the resistance and the temperature. If you set the maximum output
    voltage of your DAC to a useful standard level (e.g. 2 volts
    peak-to-peak, as is fairly usual for CD players and other line-level
    outputs), you'll find that the thermal noise generated by the
    resistances in the DAC circuitry will be down in the 24-bit region.
    If you try to resolve signals smaller than that, they'll be buried in
    the noise.

    You can try cooling the DAC down with liquid nitrogen, and maybe
    that'll get you another bit or two in noise performance, but it's
    going to be expensive in the long run.

    Linearity is also an issue. In order to reproduce very-low-level
    signals with a DAC, the DAC's internal electronics must be very, very,
    VERY accurate... and at some point, the necessary accuracy moves
    past the "expensive" point, to "beyond the state of the manufacturing
    art" Twilight Zone.
  12. Dick Pierce

    Dick Pierce Guest

    Well, now, think about it. What do YOU think 600 dB means?

    Well, if you haven't gotten there, 600 dB means there is a factor
    of 1,000,000,000,000,000,000,000,000,000,000 between the amplitudes
    of the smallest and largest representable signals. That's 10^30

    Fine, so figure out what THAT means.

    Okay, didn't bother to go there yet? Fine.

    Let's look at hooking up a h-fi system to this rather silly device.
    Let's assume that you have adjusted the volume control so that the
    very smallest signal this device can put out generates a sound that
    is 1/100 as loud as the faintest signal you can hear, that is, at
    a sound pressure level of -20 dB re 10^12 watt/m^2 (that's a trillionth
    of a watt per square meter).

    Fine. How loud can it play? Well, according to you, 600 dB -20 dB or
    580 dB SPL. Great, what does THAT mean?

    Well, if it's 580 dB above 1 trillionth of a watt, that's 10^27 watts
    per square meter of acoustical power. Let's just, for the sake of
    simplicity, assume the entire acoustical radiation is confined to that
    1 square meter The amount of acoustic power you want to put into that
    small area is something like 1,000,000,000,000,000 times greater then
    the entire amount of electrical power consumed by the entire United
    States for all conceivable and inconceivable reasons. And let's assume that
    your speakers are pretty efficienct, say 10%. That means you are going
    to have to find yourself an amplifier capable of putting out somewhere
    in the range of 10,000,000,000,000,000 watts. Assume such an amplifier
    is running at about 40% efficiency. How much power would you need?

    Well, the entire output energy output of the sun is probably insufficient
    to the task.

    Got it?
  13. What are you trying to achieve, Mr Radium? Will you put out question
    after question until you feel educated?

    Methinks Radium is a troll. Nobody can be this stupid, based on the
    the questions asked so far.

    But it *is* fun when D. Pierce does his usual
    you-asked-for-the-reason-here-it-comes computing of needed power!

    I can see the ad in front of me: "The new Gizmo 600dB amp will
    outperform the sun!"

  14. And the microphones............
  15. It can be - that's too sweeping a generalisation.
  16. Because of the dynamic range of the analogue output stage.
  17. Because of the dynamic range of the analogue output stage. The
    standard full-scale output level is 2 volts rms, and the self-noise of
    most line stages is seldom less than 2 microvolts. This is a 120dB
    dynamic range, or 20 bits.

    Check out the specs of most '24-bit' DACs, and you'll very seldom see
    dynamic range quoted at more than 114 dB, which is only 19 bits. I
    believe that the current state of the art is represented by the dCS
    and Apogee professional converters, which run out around 125dB, just
    under 21 bits.

    Of course, bearing in mind that no music master tape in the history of
    the world has ever exceeded a dynamic range of 85dB, and that hardly
    any studio microphones have a DR of more than 100dB, you'll see that
    this in't a problem in the real world!
  18. Perhaps a simpler analogy would be that it represents rather more than
    the dynamic range that you would experience if you were standing in
    the New Mexico desert at midnight on a perfectly still night - and a
    cruise missile exploded ten feet away........... :)
  19. Peter Larsen

    Peter Larsen Guest

    The multimedia speakers for PC's will get there first .... it was indeed
    a well written and entertaining discourse Mr. Pierce delivered.

    Seasons Greetings / God Jul

    Peter Larsen
  20. Arny Krueger

    Arny Krueger Guest

    Not true. Both analog and digital circuits can put very nice flat tops on
    your music waveforms.
    Right. In practice this means that you can allow more headroom without
    getting fried by hiss at low levels.
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day