Digtal vs Analog; Why Isn't Digital Better? (long)

[sunnysal]

that is wierd I did not post that last post....I guess I have to change my password...tony

[DizRotus]

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

On 5/9/2005 11:53:59 PM scriven wrote:

When the computer reads a CD it will get an exact copy of what is on the CD......."

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

Thanks Mark. That's what I thought, but I wasn't certain.

Tony-

That's disturbing about a unauthorized post in your name.

[colterphoto1]

I've been purposely avoiding this thread for fear of running into a flame war, but all looks peaceful in Klipsch-land. Good info here, esp. like DrWho's posts. I volunteer to buy your old college texts when you get tired of them. You're on a very interesting run of courses there, young friend. Keep it up!

Michael

[Smokem]

Wow! I thought switching from Audio Cassette/VCR to CD/DVD was joy to my ears. And upgrading to KLIPSCH speakers of course! Digital vs Analog? My ears can't tell the difference. I guess I'm not a BIG audiophile.

[formica]

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

On 5/9/2005 9:03:07 PM sfogg wrote:

Tony, "I say use what you like...just listen!"

I agree.

I was just pointing out some incorrect info posted about how digital works.

Shawn

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

And it's appreciated.

It's unfortunate that some people will still refuse to understand (accept?) the true nature of the pros and cons of each format. There is so much false reasoning floating around, promoted by some manufactures and the magazines they sponsor *cough* stereophile *cough*.

That is one thing I like about the Klipsch forums, that it has a greater than average number of technically informed members.

Rob

[pauln]

3DZ is right.

Consider how many points (samples) there are available to define the waveform when you sample at 44K/s.

Freq Number of data points to define the wave

20 2252.8

39 1126.4

78 563.2

156 281.6

313 140.8

625 70.4

1250 35.2

2500 17.6

5000 8.8

10000 4.4

20000 2.2

Looks good for the low end, but happens with high frequency sounds? Now you see the problem with digital sampling - for example, the two points used to define the 20KHz sine wave are used to define the top and bottom points of what becomes a sawtooth wave. The lack of points available to characterize the waves manifests itself by sounding like the harmonics and overtones are missing. These are the signals that distinguish the different instruments and give musical sounds their timbre.

Digital to Analog conversion is all about making up for this loss by using various electronic guessing. At the current sampling rates these guesses are audibly off. Higher sampling rates will improve this, but it has a very long way to go - remember that rock on plastic (stylus in groove) is a 40 bit signal at the molecular level. This is a continuous sample rate that approaches infinity as far as reproduction is concerned.

paul

[sfogg]

"Consider how many points (samples) there are available to define the waveform when you sample at 44K/s."

And *again* all you need is more the two samples per cycle of the wave to reproduce it.

"but happens with high frequency sounds?"

Exactly the same thing that happens for the low frequency sounds. After all as you showed there are more then two samples per cycle....

" Now you see the problem with digital sampling - for example, the two points used to define the 20KHz sine wave are used to define the top and bottom points of what becomes a sawtooth wave. "

The samples are not trying to draw the analog wave... it is NOT connect the dots.

Until people realize that they are simply not going to understand how digital works.

The distortion and noise characteristics of Redbook CD is no different at 20kHz then it is at 20hz. Additional samples per cycle on a wave buy you *nothing*. You run a 20kHz signal through an A/D and a D/A and you will not in any way shape or form end up with a sawtooth wave. You will have a sine wave... just like you do at 1000hz... just like you do at 20hz.

"is all about making up for this loss by using various electronic guessing. At the current sampling rates these guesses are audibly off."

Completely wrong, there is no guessing at all.

" is a 40 bit signal at the molecular level. "

40 bit would mean it has a dynamic range of 240dB. The loudest sound the atmosphere can support is around 198dB. So quite literally you are out of this world if you think vinyl has 40 bit dynamic range. Again, what is the signal to noise ratio of your phono stage? That alone will give you an idea on the max dynamic range you have, and in reality it will be below that. The Blueberry has 70dB of dynamic range... that is a equivalent to a little over 11 bits of resolution.

"This is a continuous sample rate that approaches infinity as far as reproduction is concerned."

So then you are claiming vinyl has unlimited bandwidth? Funny then how most measurements of bandwidth (meaning good signal delivery) put the upper end limit closer to 16kHz or so but it of course varies by cartridge and recording.

Shawn

[3dzapper]

"Each advance in digital technology is a step in the right direction. Maybe at some point the sample rate will get to a point where all the music is there and the device does not have to fill in the gaps with it's sometimes faulty logic."

I think the main point of my original post (above) has been lost in the quibbling over the obsolete 1970 technology that is Red Book CD.

The next generation of 21st Century technology, using blue lasers, will provide a true lossless, uncompressed, representation with a data stream of 1.5M Hz or better and a sampling rate for 7.1 audio of 24bits @ 96Khz along with HD quality video.

Pure audio standards have not yet been established but there is no space limitation on the disc to prevent a 36 or 48 bit word totally lossless with no compression. Do I expect LP performance? No, I expect better.

I have a graph of a sine wave recorded at 24 bits at the same level as the first which shows a very nice waveform with minimum filtering required. And, a screen shot of a sine wave from an LP but the Forum's picture loading function is down.

Rick

PS: Please don't try to parse my statements or put words to my statements that are not there. I know what I mean even if I lack the eloquence of Longfellow or Whitman. Even though I live at the confluence of Whitman and Longfellow it did not rub off.(

[formica]

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

On 5/10/2005 10:17:36 AM pauln wrote:

...the two points used to define the 20KHz sine wave are used to define the top and bottom points of what becomes a sawtooth wave. ...

....remember that rock on plastic (stylus in groove) is a 40 bit signal at the molecular level. This is a continuous sample rate that approaches infinity as far as reproduction is concerned.

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

I guess you didn't read all the posts about Nyquist and the way the data is interpreted?? The data points are not used in a "join the dots" topology... but rather one to define a series of sine ways. Your 20kHz wave will be reproduced in it's entirety... and without the distortions introduced by vinyl's limitations.

PSG did a better job than I can at explaining the math in a simple "visual" way...

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

On 5/9/2005 10:02:32 AM psg wrote:

Most people look at a 10 KHz sine wave and wonder how can 4 samples reproduce that? The math works. In my field, it gets used for sea-level and currents analysis (tides) and I use it to quantify turbulence and mixing rates from 100 Hz measurements of water temperature. I think the prrof of the misunderstanding is well summerized by Rick saying:

No matter how mant bits you add, the waveform will still be a semblance of the original until the sample rate reaches into the mega BPS range.

To him, the only way to reproduce a sine wave digitally is to sample it infinitely. All he sees are the increasingly smaller square staircases as we increase the sampling frequency or the bit rate. Most people don't understand the reconstitution of the original as a series of sine waves with different frequencies.

I'll give it a shot, for those who don't understand Nyquist. (This is not how Nyquist is usually explained in school). Say you have an analog square wave and you want to represensent it as a series of sine waves of various frequencies. You'll start with a sine wave with the same wave length as the square wave. Then you'll add a harmonic (2x) to start capturing the squareness of it, then a higher harmonic (3x) to get a better fit, etc. The square wave is decomposed into an infinite series of pure sine waves. Now, if you go in the opposite direction and start from a crude digital sample. It represents a basic sine wave and a series of higher harmonics. If you know that there's no content about some cutoff frequency, then you can filter them out of the digital squareness and leave only the frequencies less than half the sampling rate. They will be correctly reproduced.

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

later...

Rob

[meuge]

But IT'S NOT a sawtooth wave. It's a mathematical derivation, based on the samples provided.

You still don't understand the way data is being stored. For example, let's say you need to store the following coordinates: (0,0; 1,1; 2,2; 3,3; 4,4; 5,5; etc... up to 999,999). In an analog format, you have to keep specifying each coordinate. Digitally, you can simply write "x=y(if x=0-999)". Notice that in this case the analog storage is precise only to an integer value, while the digital waveform is infinitely accurate.

Same with sound waveforms - you don't need to record every point on the waveform in order to reproduce the waveform, you just need to derive its equation. Just try it... get a graphing calculator and get a couple of samples and have the calculator derive you the equation. As long as it knows that it's supposed to be a sine wave, you'll see how well it can do it.

44.1kHz isn't enough... but 192kHz is. Just like one cannot hear a 0.05dB frequency response changes caused by speaker wire, there is no way you could possibly hear a difference between 192kHz digital vs. the original (assuming that the transducer/speaker was 100% perfect). Coupled with an insane noise floor of the 24bit recording, SACD/DVD-A is unquestionably the medium of choice for audio storage.

[formica]

This is fun... isn't it? Kinda like a first person shooter...

Shawn's got the better reflexes...

[pauln]

Let me ask a leading question, while I beat out the last of the embers of my scorched keyboard...

Let the actual original waveform to be digitized and returned to analog be a sawtoothwave at 22KHz (2 samples). Will the CD player emit a sawtooth waveform or a sinewave?????

The Nyquist approach is fine for constant frequency signals that make predictable composites of sine waves. In the time domain, real music is not like that - the overtones, harmonics, tones of Tortini, and envelope components are not all sine waves and not all regular and calculable. The digital to analog conversion is as you guys say, "an interpretation".

So how does it interpret a sawtooth wave at 22KHz? As a sawtooth or a sine wave?????

Flameproof Paul

[sfogg]

Rick,

Why do you think anyone needs 36 or 48 bit delivery formats. We can't even make normal electronics that support 24bit resolution let alone things that have a dynamic range beyond what would be possible to reproduce in the atmosphere.

24 bit gives in theory 144db of dynamic range. The resolution/dynamic range of the system is defined by its weakest point.

A 50ohm resistor at room temperature with 2v through it (output from the player) has a thermal self noise of just about -144dB.

IOW, if you want to build a pre-amp that has a SNR of 144dB the entire signal path of the pre-amp can't have more then 50ohms of resistance in it. Otherwise the resistance alone (even totally passive) will be killing the resolution of the signal you pass through it.

Since most pre-amps have input impedances of thousands of ohms and output impedances of at least 100ohms the above isn't so likely. And in reality the entire signal path would need to be below 50 ohms (as noise is cumulative), meaning the output circuitry of the player, the pre-amp, the amp and the speakers. Not going to happen. And that is to support 24bit, you go higher and it gets even worse.

Set the max playback level in a room to 110dB in a room with a 50dB noise floor and the entire system is limited to 10 bits of resolution.

Pump the 24bit signal through a pre-amp that has hum and noise at -100dB and you have just lost over 7 bits of resolution due to the noise just in the pre-amp.

" I have a graph of a sine wave recorded at 24 bits at the same level as the first which shows a very nice waveform with minimum filtering required. "

If it is still recorded at -90dB on a 24 bit delivery format then there is no doubt that would be the case. And that is because the sine wave is further away from the noise floor of the format.

And it is still just as pointless because recordings don't use anywhere near that kind of dynamic range. There isn't a system on the planet (or a persons ears) that could support a system that fully used 24bits of dynamic range. In a dead silent room with 'perfectly silent' equipment (SPL in room = 0dB) it would take 10,000w into a K'Horn to accurately reproduce 24bits of dynamic range. The maximum playback level would be 144dB which will cause permanent hearing damage/loss.

As far as next generation delivery format we already have a delivery format that gives 6 channels of 24bit@96kHz audio. It has been out for years... DVD-Audio. It uses data compression (MLP) but it is lossless compression that is bit perfect. It can also support 2 channels of 24 bit @ 192kHz.

" Do I expect LP performance? No, I expect better."

And there is the rub.... how do you want to quantify that?

By any technical standard (noise/distortion/fr/phase response...etc..) that has been done.

Yet many still prefer vinyl.

If you like e-mail me the pictures you have and I will try to get them online for you.

Shawn

[sfogg]

Paul,

"Let the actual original waveform to be digitized and returned to analog be a sawtoothwave at 22KHz (2 samples). Will the CD player emit a sawtooth waveform or a sinewave?????"

You mean if you input a sawtooth at 22kHz? In that case the output would be a distortion of the signal.

A sawtooth is defined as a fundamental frequency with odd and even harmonics at higher frequencies. Those higher frequencies wouldn't be captured or reproduced by a 44.1kHz sampling frequency so the sawtooth wouldn't be reproduced properly at that frequency and at that sampling rate. It should look more like a sine wave if you pass it through an A/D-D/A.

" In the time domain, real music is not like that - the overtones, harmonics, tones of Tortini, and envelope components are not all sine waves and not all regular and calculable."

Actually, you can break any signal down to its sine wave components. An FFT will show you that. Real music is just a bunch of sine waves.

Things like square waves,triangle and sawtooth waves don't occur naturally in music (or anywhere else really). Even if they did microphones and speakers will distort them badly as would phono cartridges because most don't have the bandwidth needed for the sine wave components

Shawn

[DizRotus]

This is exactly the kind of intelligent and spirited, but civil, dialogue that I was looking for to shed light on the subject.<?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" />

Its obvious that the collective expertise on the topic is over my head, but, nevertheless, I feel that I understand more now, after reading the posts; and that scares me. As you know, A little knowledge is dangerous.

Im reminded of a record I once had where Dr. Edward Teller was explaining the size and nature of the universe. In doing so, he commented regarding Einsteins theory of relativity (paraphrasing now), If you think that you dont understand it you might understand it. If you think that you do understand it, then you probably dont understand it.

Understand?

[3dzapper]

You can stick a fork in SACD and DVD-A. They're done. Sony is betting it's future on Blue Ray, DVD-A has never gotten a foothold.

By the end of the year, the first readily available Blue Ray players will make their appearance and they are coming at the right time. SACD, DVD-A and for that matter Quadraphonic LPs (don't forget the Edsel:) all made their appearances during economic downturns. Blue Ray is coming amid an expansion period and will make headroom fast. The first disks will be HDTV videos/movies as that is where the largest market lies. They will also have a DVD or CD layer to be read by a red laser for backwards compatability. The audio portion will be in DTS++ 24/96 uncompressed at greater than 1meg.

Computers will ship next year with Blue Ray read/write drives.

There is nothing that anyone can do to stop this train. At least until some company develops an ultraviolet disc.

The choices are going to be: embrace this new technology or be religated to MP-3.

Rick

[sunnysal]

sean, I have been told there are some intruments that generate squarewave signals, perhaps not all the time but...anyway, what about a trombone? I have heard tat given as an example. regards, tony

btw I changed my password let's see what happens...

[grillecloth]

I been following this debeat. Whilest some have made good cases; one stands as a questionable statement by someone that says they record many things.

"people don't wn't to hear guitar distortion." Jimi Hendrix, Eric Clapton, Drew Abbott was with Seger, REOs, SRV, Ted Newgent, Joe Walsh blow that out of the water.

Maybe you just meant microphone over laod Doc. For the rest I play what I wanna hear

[sfogg]

Tony,

They might generate 'squarewaveish' signals but they wouldn't look the same as a square wave out of a function generator.

Keep in mind it actually doesn't really matter if you think of it simply as frequency response.

A square wave or a sawtooth wave is nothing more then a sine wave with a specific distribution of harmonics of the main sine wave frequency. Those harmonics are also just sine waves.

As such if you tried to record an instrument at a 44.1kHz sampling rate and that instrument did in fact produce a square or sawtooth wave you would capture the main sine wave and any harmonics that were below 22kHz in frequency. If it had harmonics above that point they would be lost in the recording.

It is really no different then if your amp or speaker or whatever didn't have enough frequency response/bandwidth.... those harmonics would be lost. That is why an amp could say be flat to 25kHz and reproduce perfect sine waves at 20kHz. But if you feed it a square wave (20kHz sine wave and sine wave harmonics at 40kHz,60kHz,80kHz...etc..etc.. ) it would reproduce the first sine wave but doesn't have the bandwidth for the harmonics so if you looked at it on a scope you would not see a square wave.

Make sense?

Shawn

[Klewless]

yea,

It is hard to accept that those same sine waves that accompanied the train when it hit the living room are the same ones that come from a sweet sounding pipe organ! Like the cook said, "it's all in the receipe".