Analog Defined: Stairways to Bad Audio?
I belong to an email list that caters to the reel-to-reel crowd (believe it or not I’m a fan of analog tape and have a long history with this equipment!). It’s been a little quiet over the past month or so but came to life when one of the members asked readers to offer up their definition of analog. Now there’s a question that’s guaranteed to energize a conversation. The questioner wasn’t interested in the standard platitudes about the “warmth and emotion” of analog vs. the “cold and harshness” of digital formats. He wanted people to express their preference in other terms. The responses have been interesting.
There were links to a website that made outrageous claim after outrageous claim about the superiority of analog tape over high resolution digital. I won’t bother to refute them one by one now, but it’s obvious that the proprietor of the business (which specializes in providing analog tape copies to fans of reel-to-reel tape) has a vested interest in spewing bullet points of misinformation. Here’s a wonderful example of the pseudo science included in the piece:
“A tape captures transients and resolution extremely well, with LPs being inferior (mechanics inertia and alike). As to high resolution digital – it has some ‘sharp edge’ – too-sudden artificial transitions/jumps, associated with the binary ‘bits’ and the sampling-oversampling process, which sounds fast, but unnatural, hence often fatiguing, to the brain. Just like in the sea-sickness process, where what the brain gets from different receptors is contradicting, leading to undesired experiences.”
How do you present a rational argument against statements like this? What is a “sharp edge” in relation to the capture and reproduction of an analog waveform?
If you would like to read the article for yourself, here’s the link: http://highfidelity.pl/@main-306&lang=en
One of the most often heard arguments against PCM digital was stated this way, “Hard to put into words, but the human ear responds to waves or cycles of sound. Digital is a stepping of a wave in amplitude, and I believe that some humans may not know what makes it sound different, but they perceive two unique sounds.” The emphasis here is on the phrase “Digital is a stepping of a wave in amplitude.” Hogwash!
PCM digital encoding does allocate discrete values to the amplitude axis of our analog to digital conversion process. At each of the samples, the output of the encoder can’t be selected from an infinite number of values…the so-called smooth purity of analog (whether a groove or a magnetic domains). Given the number of bits in the digital words used there are more or less values to choose from. This establishes the dynamic range that is possible within that ADC. And yes, if you want to think of them as discrete levels, then fine. I prefer to conceptualize them as “points” of data associated with the samples. That’s the totality of the information we have available to us in a PCM system. We know the time of a sample and we know its value. End of story. And Nyquist proved that you get all of the original signal back when converting back to analog.
The output of the AD converter is stored on a hard drive or optical disc as individual letters would be stored in a sequence of neatly spaced neighborhood mailboxes. When the stored data is output for conversion, each amplitude level is equated with an analog voltage that in turn gets amplified and sent to a speaker. The actual movement of the speaker cone isn’t “stepping” in and out. It moves in exactly the same smooth “analog” fashion that comes from any other format…including vinyl LPs and analog tape. There is a low pass filter on the output of the DAC that removes any frequencies caused by the “discrete values” that the DAC receives form the hard drive or optical disc pickup.
There is no such thing as “stepping of a wave” in digital audio…but still the myth persists. This is not open to opinion…it is a well-establish fact that the output of a PCM digital system is a smooth replication of the input waveform. Yes, we have imperfections like jitter and quantization noise but they are nothing compared to tape hiss or scratchy vinyl LPs.
Each format has a flavor and plenty of people obviously like the “classic” analog sound…but then again plenty of people have graduated to the clarity and accuracy of HD-Audio and even surround. Just listen to the samples available through this site and see for yourself.
Let’s comment on the following statement: “And Nyquist proved that you get all of the original signal back when converting back to analog.”
Although this is true in theory, it’s only for pure intellectual satisfaction but has no practical value. Here is why in this excerpt from professor Milind N. Kunchur, University of South Carolina (see http://www.physics.sc.edu/~kunchur/papers/FAQs.pdf) :
“This Whittaker–Shannon interpolation is exact as long as the following conditions are met:
(1) The discrete levels represented by V(n) are correct in the first place. (Some error will result because of the
finite vertical resolution. Other, more serious, errors can result because of the way the V(n)’s are synthesized).
(2) V(t) is band limited and does not contain frequencies above the Nyquist value 1/[2T]. (This can be a
problem when one is trying to reproduce or generate square waveforms, unless special tricks and techniques are used. The problem can be exacerbated when the square waveform and sampling rates have incommensurate periods; sine and other smoother waveforms are more forgiving in this regard as explained further below.)
(3) The sum stretches from minus to plus infinity. (In practice the sum is always finite).
(4) The interpolation function corresponds exactly to the sinc function, where sinc(x)=sin(πx)/πx. In the
frequency domain, this corresponds to an exact “brick-wall” low-pass filter. (In practice this is impossible to
realize exactly.)
(5) There is no error in the timing (jitter) at any point in the process: the voltage samples are taken exactly on
time at the sampling frequency during the analog-to-digital conversion process and similarly the digital samples are converted to analog voltages exactly on time during the DAC process. (In practice no clock is perfect; but in modern high-end audio equipment jitter seems to have been reduced to negligible levels.)
In reality, all of these conditions are violated to some extent.”
See also a very interesting selection of M.N. Kunchur publications in the audio field at http://www.physics.sc.edu/~kunchur/Acoustics-papers.htm.