Don't see the graph. There is nothing to support this statement except flawed interpretations of answers to poorly posed questions. Yes, jump! One flaw might be the meaning of accurate, by ED's definition (<140dB) it can never work.
EDIT - found it, that was just an off the cuff comment with no data to support it and just plain wrong.
There are two issues here.
The first is are you trying to capture and reproduce what energy is actually in the music? If that is the case many instruments do produce acoustic energy well above 20 kHz. So a 200 kHz or even 500 kHz sample rate could be required to meet the Nyquist criteria.
Now the common assumption is that the bandwidth you perceive is limited to 20 kHz and therefore you only need to reproduce up to that. This ignores the data that shows younglings can often hear to 22,000 Hz and there is some data even normal young adults can detect up to 30 kHz at some level.
The second issue is dynamic range. At some frequencies normal folks can hear to - 6 dBc acoustic. The threshold of pain is somewhere around 132 dBc acoustic. So a 138 dB range would be required to reproduce that.
Now the assumption there is that no recording has that range, so you don't need to reproduce it.
My current systems are getting close to 120 dB dynamic range from the power amplifier through the loudspeakers. Of course the source material is not that good.
ES
Since no one that contends anti-aliasing at 20k is inadequate (Nyquist by definition for 44.1k) comes forth with anything but personal anecdotal tests, there is no point in going over this again and again. Toss off comments about 24/192k being obviously inadequate are only worth ignoring.
Your correct. The defenders of perfect theory are alive and well.
Perfect theory give perfect results. I said Practical results do not live up to the perfect theory in accuracy. With the real world imperfections, a higher sample rate may be needed to get rid of certain imperfections to the highest freq for testing of D.U.T. Personally, I think 1-5KHz test freq is sufficiently high. But, others want to go higher.
Thx-RNMarsh
Last edited:
For the past 9 months, I have acqiured several test instruments for measuring THD, THD+N and individual harmonics with various technologies and FFT/sound card devices and compared the results they give under various levels and freqs. [I am talking about the measurement accuracy of harmonic levels]. This is chronicled under another forum on ultra-low distortion signal generators (which then need to be measured). Inexpensive sound cards were the least accurate.
Thx-RNMarsh
Thx-RNMarsh
Last edited:
Well as soon as you can build a system with 200 kHz bandwidth and 140 dB dynamic range we can do a well controlled test.
The concepts that distortion and bandwidth fully describe a music reproduction system are just silly.
When I first mentioned the Sabin none here were familiar with that unit of measurement.
So let some think it is like a snake and others a tree or a wall. Pull the cord and get a different answer.
There ain't no stinking micro diodes, there is no such thing as DC and no single figure of merit for audio systems.
Digital vs. Analog
For those that think without limits digital is very confining. It has hard boundaries that cannot be breached- .001dB over full scale is hard clipping with disastrous consequences. The maximum frequency is a brick wall no matter what, the minimum level is that point below which you cannot go since there is noting there and the resolution from step to step is a hard fixed interval.
Contrast that with analog; no minimum or maximum level, frequency range unlimited and infinite resolution in levels. That is as much an illusion as the gold at the end of the rainbow.
The max level on an analog recording medium is clearly limited with significant impact long before you get close. The minimum level is set by the recording medium's density and in any case is limited by noise. The infinite resolution is also limited by that same noise. Even if you had a perfect medium those analog gremlins in the electronics will resurface and constrain your magic. For all intents and purposes a 192KHz/24bit recording is perfect. If the sounds originate in nature and are sounds a human can tolerate (less than 120 dB SPL) the dynamic range of available 24 bit ADCs can capture all there is before you reach any real noise limit in any real space occupied by humans, if the microphone can capture it. 100 KHz seems like it may also be a limit, but the wavelengths are so short that most recording microphones won't see them, or even much above 20 KHz if that. And those that do are noise limited.
I think effort in figuring out how to understand and work with those limits is more worthwhile than continuing to complain that they are not adequate.
If there is musical content that can affect human perception above 20 KHz (which their might be) doing some real research to see how much and how to reproduce it would be worthwhile I think. And then determine if it really does make a difference.
Its very telling that digital recordings can capture the nuances of different cartridges and turntables. I don't think the inverse works too well.
For those that think without limits digital is very confining. It has hard boundaries that cannot be breached- .001dB over full scale is hard clipping with disastrous consequences. The maximum frequency is a brick wall no matter what, the minimum level is that point below which you cannot go since there is noting there and the resolution from step to step is a hard fixed interval.
Contrast that with analog; no minimum or maximum level, frequency range unlimited and infinite resolution in levels. That is as much an illusion as the gold at the end of the rainbow.
The max level on an analog recording medium is clearly limited with significant impact long before you get close. The minimum level is set by the recording medium's density and in any case is limited by noise. The infinite resolution is also limited by that same noise. Even if you had a perfect medium those analog gremlins in the electronics will resurface and constrain your magic. For all intents and purposes a 192KHz/24bit recording is perfect. If the sounds originate in nature and are sounds a human can tolerate (less than 120 dB SPL) the dynamic range of available 24 bit ADCs can capture all there is before you reach any real noise limit in any real space occupied by humans, if the microphone can capture it. 100 KHz seems like it may also be a limit, but the wavelengths are so short that most recording microphones won't see them, or even much above 20 KHz if that. And those that do are noise limited.
I think effort in figuring out how to understand and work with those limits is more worthwhile than continuing to complain that they are not adequate.
If there is musical content that can affect human perception above 20 KHz (which their might be) doing some real research to see how much and how to reproduce it would be worthwhile I think. And then determine if it really does make a difference.
Its very telling that digital recordings can capture the nuances of different cartridges and turntables. I don't think the inverse works too well.
You haven't been looking at many CDs it seems - clipping regularly occurs on a remarkable number of them, I was quite taken aback when I came across examples on classical works, and on tracks from studios with reputations for high quality. And, the CD player doesn't explode when such a track is played, I had played tracks with significant clipping on them, which I wasn't aware of, many times and never noticed anything untoward ...
Does that mean you said the rest?
Now is there a 24 bit accurate 192k A/D?
Is there a microphone flat 20 to 200 kHz? (Let's not forget about uniform polar pattern also.)
Once you get those I'll take care of the loudspeaker.
Nyquist was right, so to see (say) the third harmonic you need three times the sampling rate. My concern was that some people might think that needing a higher sampling rate for instrumentation implies that ordinary audio needs faster sampling too. It doesn't.qusp said:
Using digital sampling to see the output of an amplifier requires anti-alias filtering again even if the amplifier input is band-limited, or you need a much much higher sampling rate. If you are then doing an FFT this should expose any aliases, as they will appear at anharmonic frequencies (probably relating to harmonics +- sampling frequency).
No, dither fixes this.1audio said:
Last edited:
Dick, if by "inexpensive sound card," you mean $20, I may agree with you. If you mean $200 sound card (which is damn inexpensive compared to stand-alone units), I strongly disagree. I have two 24 bit commercial units in that price range that give results down to -140dB that agree with one another and agree with the $30,000 R&S in our lab at work.
I don't think that I was amazingly lucky twice.
I don't think that I was amazingly lucky twice.
Does that mean you said the rest?
Now is there a 24 bit accurate 192k A/D?
Is there a microphone flat 20 to 200 kHz? (Let's not forget about uniform polar pattern also.)
Once you get those I'll take care of the loudspeaker.
Where's the evidence that any of this is needed?
sure, I understand your concern; which is why I made sure to restate my post. i'm wondering what the problem is though, good quality 192khz+ ADCs are fairly readily available.
- Status
- Not open for further replies.