Amplitude Resolution ...

[phobia]

Hi!
I have been following some interesting threads about the works of the digital domain
on this list.

I have just a little question that I have
been wondering about for some time.

- 0dB is considered digital max signal level.
- each bit would normally correspond to a ratio of 6 dB meaning that a theoretical headroom for a 24 Bit recording is 24*6 dB = 144.

My question is, how is the resolution of this headroom measured?

If we talk about sampling frequency (which is a different thing) we know that the higher the sample rate, the higher is the 'resolution' we use to capture a sample, from a time prespective.

But on the amplitud level, are ALL 24 bit digital ADs capturing accurate levels on the scale -144 dB -> 0 dB ? Can there be differences in accuracy or resolution here?

Regards,
// Joakim W

------------------
>>We need more time so that everything doesn't happen at once<<

[phobia]

Hi all!
I guess either is my question stupid or
nobody is interested.

I find it very usefull to gin knowledge of this though.

I'll try discussing in some other forums as well and cross/psot my findings.

Cheers,
// Joakim W

------------------
>>We need more time so that everything doesn't happen at once<<

[Zeus]

It's likely that the most drastic differences between converters is in the very low amplitude range. But that range is already full of useless noise from the recording environment, mics and preamps...If there were considerable non-linearities in the high amplitude range those would immediately result in audible distortion...?

Tech-heads, any opinions?

Z

[666]

Zeus wrote: "...the very low amplitude range. But that range is already full of useless noise from the recording environment, mics and preamps...."

I have the opposite opinion. The very low amplitude range determines the quality of your audio recording.

low level digital signal is just distortion unless you add some "noise" (dithering) that can mask the distorsion.

Try this: use the signal generator plugin and sweep the freq. You can hear a big distorsion / modulation as you increase the frequency.

Every sound you record that includes such frequencies will be distorted in this way. Adding a dithering the distorsion is "masked".

So the way a converter /DAW / Mixer /Plugin treats those low levels determines the quality of the audio output.

------------------
somewhere on the planet

[Zeus]

Excuse me 666? How is dithering (noise used to smooth low lovel digital signals) related to aliasing (distorsion caused by frequencies exceeding the Nyquist limit)? You are speaking in a quite confusing way. And what has the Digi signal generator (that is anyway with many bugs) at all to do with AD-converters???

666, getta grip on yourself, please!

Z

[Zeus]

Excuse me 666? How is dithering (noise used to smooth low lovel digital signals) related to aliasing (distorsion caused by frequencies exceeding the Nyquist limit)? You are speaking in a quite confusing way. And what has the Digi signal generator (that is anyway with many bugs) at all to do with AD-converters???

666, getta grip on yourself, please!

Z

[Zeus]

Excuse me 666? How is dithering (noise used to smooth low lovel digital signals) related to aliasing (distortion caused by frequencies exceeding the Nyquist limit)? You are speaking in a quite confusing way. And what has the Digi signal generator (that is anyway with many bugs) at all to do with AD-converters???

666, getta grip on yourself, please!

Z

[Zeus]

Sorry for the multiple posts! Something wrong with the browser or network here...

Z

[Sonsey]

To answer the original question...

Yes all 24 bit converters are capable of the theoretical 144db headroom. However headroom is the difference between the Loudest signal (in this case 0dbfs) and the noise floor. -144db Signal to Noise is equivalant to molecular vibration. The problem is that NO electric circuit is going to be that quiet. They are all going to add noise to the signal - the more noise the less headroom. Obviously the better the circuit the less noise it will induce into the signal. Hence ONE of the differences in converters. A noisy 24 bit converter may only give you 18-20bits of "actual" headroom before the signal drops below the noise floor. So the answer is NO - there is no 24bit recorder that can capture the full 144db of dynamic range- it's just a theoretical max. Of course there are many other factors such as jitter and the cutoff filter, that can affect the sound of converters as well if not more.

Howard
Atomic Productions

[F Umminger]

What you are hearing with the signal generator is aliasing. The saw, square, and triangle waveforms are implemented in a naive way that suffers from significant aliasing distortion. This is completely unrelated to bit depth issues; the signal generator would sound just as bad in 48 bit audio.

Fixing this has been a low priority because this plugin only exists for the purposes of very basic testing and troubleshooting.



------------------

Frederick Umminger
Digidesign Plugin Engineer

[phobia]

So I guess the answer is:
The significant/resolution of the mesurments done in amplitude level during AD-convertion doesn't matter, mostly due to that other
errors inflicted are larger?

------------------
>>We need more time so that everything doesn't happen at once<<

[Sugarite]

I've had my head snapped back many a time trying to argue that there is a direct relationship between digital dynamic range, and analog headroom. Either I've got a chronic mild concussion or I'm starting to see how they are too different to assume certain similarities.

One dubious thing I've done in the past that has worked is let recording levels clip occasionally on my portable DAT during a location recording. I keep the RMS signal well under 0dB, but if the occasional spikes go over it isn't audible on my particular recorder, nor are they audible once they've been transfered into ProTools.

Since every live recording is very different, I can't really say if it makes a significant improvement, but... shouldn't it? Since the clipped info is not really missed, it's as if I've expanded the dynamic range for the RMS signals at least.

The opposite extreme case is recording at very low levels where as we all know it's bound to lose resolution. Wouldn't every step in that direction result in a respective quality loss?