Digital Technology Collection

Negotiableterms

Administrator
Staff member
Admin
Subscriber
The purpose of this thread is to collect links, articles, whatever, on how digital sound works. When the collection is big enough, I'll rearrange things into a coherent thread that can be used as a reference by anyone wanting to understand things like quantization, dither, jitter, etc.

Here's an example of the kind of thing we're looking for, which is a fairly good coparison article on SACD and DVD-A:

http://www.digit-life.com/articles2/sacd-dvd-a/index.html

Everyone, please join in. If you have or find something good, POST!
 
"Notes on the Troubleshooting and Repair of Compact Disc Players and CDROM Drives"

http://www.repairfaq.org/sam/cdfaq.htm

CD players, how they interact with the CD (very long, technical, not too dry), what breaks, what you can fix yourself. LOTS more information than just troubleshooting.
 

The video has some problems. First, a scope won't reveal if the sine wave is "exactly" and/or a "perfect" sine wave. For instance, if the fundamental frequency is altered slightly, the scope will never reveal it. Even an amp with 5% or more harmonic distortion will be quite difficult to see on a scope. Now consider an entire orchestra, singing. Evidently, we must believe what he says.

If other frequencies are introduced, the scope will never reveal it. What do I mean by that? The distortion analyzer shows nothing but the fundamental frequency and harmonics. If the fundamental frequency is slightly altered, it will not be seen.

When dealing with bit depth, "stair steps" (16 bit, 65,536 values, 24 bit, we have 16,777,216 values), very rarely is the analog signal going to be exactly on a "value"/"step" during the sample period. The signal will be in between, so which value is chosen, an upper or lower value? Whichever value is chosen, the slope/rise time is altered between samples by definition, thus the fundamental frequency is also slightly altered between samples, said instrument won't reveal to us.

We won't have to worry about the slope exceeding 20khz unless the harmonic is quite high in frequency. Eight bit alters the slope/rise time even more, plus 8 bit lacks dynamics, inner detail even more than 16 bit. The inference that 8 bit, even 16 bit is enough for high quality music is the opposite of what Philips engineers believed, but RCA marketed 16 bit players anyway, and the rest is history.

Notice the Gibb's effect. It is within 20khz which is to be expected, Notice its amplitude value is high compared to the rectangular wave. Any Intermodulation distortion in the system, whether it be from speakers, electrical components, will cause mixing with this ringing, and cause non musical tones in the audio band.

It takes two samples to recreate a sine wave. At 10khz, there are only 4 samples per cycle, 5khz only 8 samples. However, music is not a simple sine wave nor a rectangular wave with equal repetitive waveforms. Music is complex with all sorts of phase relationships between instruments and their waveforms. Think it can reproduce the music perfectly when parts tolerances enter the picture and values are altered?

Ever notice the "digititus" I have heard it called, especially at higher frequencies. A grain, roughness, grittiness etc, cymbals don't sound right, sounds like glass breaking. Attempting to claim analog is 13 bit is not entirely accurate, as the needle follows the grooves in a continuous motion, not just periodically sampling. 44.1k means a cutoff of ~20khz, yet study after study demonstrates that each "ear" can perceive to at least 5us (5 microseconds) rise time differences, in layman's terms comes across as attack times.

16 bit may be good enough for one's average system, but some want better, with both higher sampling rate and higher bit, wider bandwidth. So why be against higher quality, memory is easily available and cheap.

Now I like digital and use it often, convenient, but I use 20 bit or higher if possible.

Continue on with your discussion.

keep on truckin
joe
 
Last edited:
Um, no. You need to educate yourself. Did you even pay attention to the video?

Much of the technology used for lossless compression and streaming was invented by Monty. I assure you he knows his stuff FAR better than you do. Ever hear of FLAC? Monty.
 
Um, no. You need to educate yourself. Did you even pay attention to the video?

Much of the technology used for lossless compression and streaming was invented by Monty. I assure you he knows his stuff FAR better than you do. Ever hear of FLAC? Monty.

Your reply demonstrates your lack of understanding of the science stated in my post. As such, we see no civilized argument in your defense.

Continue on sir.

keep on truckin
joe
 
Last edited:
Um, no. You need to educate yourself. Did you even pay attention to the video?

Much of the technology used for lossless compression and streaming was invented by Monty. I assure you he knows his stuff FAR better than you do. Ever hear of FLAC? Monty.

His discussion of the stepless output is based on pure sampling theory, where each sample is a zero-width Dirac delta function (his 'lollipop diagram').

That's correct in theory, but in the real world, a DAC does implement the zero-order hold he mentions, and the output of the DAC chip DOES have steps; it does not produce zero-width delta function outputs. A DAC chip is therefore followed by a Nyquist reconstruction filter, to remove the steps.

A real-world ADC will also generate a sampling (or quantisation) error, as, unless you have infinite bits, there will be an error between the real sample, and the nearest ADC quantising value. His lollipops are assumed to be perfect samples, with zero quantisation error.

Sampling theory assumes perfect, 'brick-wall' Nyquist filters. In the real world, these do not exist. Real filters have problems like roll-off rates, and ripple in passband amplitude & phase.

I was once asked to look at the design of a digital radio transceiver. They were having trouble with harmonics in the transmitted RF, violating the spectrum mask. It turned out to be the the use of a 4-bit, 12 sample per cycle sine wave generator used in the digital IF mixer (implemented in an FPGA). The quantisation error caused by the 4 bit samples was causing the harmonic distortion. I increased the size of the samples, which allowed us to reduce the harmonics to an acceptable level. The analysis of the problem needed nothing more than a look at the VHDL code to understand the sample scheme, and creating an excel spreadsheet to create a set of repeated cycles, on which I got excel to compute an FFT, which showed the harmonics.

It would have been illuminating for him to have changed the precision of his samples, from the 16 bits he used, to 8 bits, or even 4 bits. Using 16 bits, the quantisation error will be below the noise floor of the analyser he was using. It's not a good idea to try to claim an effect doesn't exist because you can't measure it.

Much of the effort expended in the design of DACs has been to try to eliminate that zero-order hold step problem; to smooth the transitions in some way. A first order interpolator would draw a straight line between the samples, and higher order interpolators would try to draw a smooth curve. The most common approach is a digital oversampling filter, which generates additional samples to fill in the gaps, with a higher precision DAC; e.g. a 4x oversampling DAC, generating 176.4kSa/s, 18-bit samples. Then there are the 'noise-shaping' DACs, such as Bitstream and MASH, that use an entirely different approach to reconstruction, with a stream of very high rate pulses; they push the noise way above the Nyquist frequency, allowing simpler, more linear filters to be used.

I started my career working on the development of the GSM standard, and the first network and handsets. In particular, the frequency synthesis and modulation. We used a technique called Digiphase, a type of fractional-N synthesiser. It did direct digital modulation by constantly changing the synthesiser frequency. It used a third-order interpolator, combined with digital predistortion to meet the modulation and spectral mask requirements. Essentially, a noise-shaping DAC.
 
Last edited:
Not sure what audiophile magazine BS is being debunked in that article, but the author does do a fine job of contradicting himself at the end in the summary.

The “stair-steps” you see in your DAW when you zoom up on a digital waveform only exist inside the computer. (The spoon only exists inside the Matrix !) When digital audio is played back in the Real World, the reconstruction filter doesn’t reproduce those stair-steps – and the audio becomes truly analogue again.

Whoa, wait. If the reconstruction filter doesn't reproduce those stair-steps, then the stair-steps must have existed in the first place in order for the reconstruction filter not to reproduce them. Poof! The whole premise of this article is gone. Also, if the audio truly becomes analog again, then what is output from DACs that claim their superiority by NOT having reconstruction filters? Like Audio Note DACs. They would have stair-steps on their output since they lack the reconstruction filter to not reproduce the stair-steps, so their output is not analog? Perhaps it is analog, but not truly analog. Must be one of them fake analog outputs.
 
The good thing about audio is that, unless one is deaf, all arguments are moot, once the soundwaves hit the eardrum. Just like there is no way to know exactly what someone else is thinking, there is no way to know if someone is lying when they indicate preference to a specific audio stimulation.

Most 'arguments' in audio seems to assume the listener is deaf.
 
Also, if the audio truly becomes analog again, then what is output from DACs that claim their superiority by NOT having reconstruction filters? Like Audio Note DACs. They would have stair-steps on their output since they lack the reconstruction filter to not reproduce the stair-steps, so their output is not analog? Perhaps it is analog, but not truly analog. Must be one of them fake analog outputs.

Sorry to contradict, but Audio Note have never said that their DACs don't have a reconstruction filter, just that that they don't use digital filters and don't oversample! Any lowpass filter that cuts at or under 1/2 the sampling frequency is a reconstruction filter, and it can perfectly be analogue - hence the "digital filter-free" claim in the article they cite on their website (http://www.audionote.co.uk/articles/reviews/t3systemreview.shtml).

Actually all the DACs have always had, and still have an analogue lowpass filter. That's very basic electronic engineering, and Audio Note never claimed they had gotten rid of it, on the contrary they explain themselves how they make at least part of their analogue filter (http://www.ankaudiokits.com/agrovedac.html), using a loaded transformer to limit the bandwidth - they don't exactly say it in these terms, but that's what "This in itself reduces overshoot and ringing and because the system is slightly overdamped the rise time is reduced to an acceptable rate as well" boils down to when you translate the sales pitch and the (very rough) schematic into technical terms.
 
Sorry to contradict, but Audio Note have never said that their DACs don't have a reconstruction filter, just that that they don't use digital filters and don't oversample! Any lowpass filter that cuts at or under 1/2 the sampling frequency is a reconstruction filter, and it can perfectly be analogue - hence the "digital filter-free" claim in the article they cite on their website (http://www.audionote.co.uk/articles/reviews/t3systemreview.shtml).

Ah, thanks for setting that straight. I hate to pass on misinformation on the Internet. It would seem you are correct, as I could not find any claim of a lack of analog filter in their DAC products. Instead, there is this comment about their CD-1.1x:

This player, like its bigger brothers, features no oversampling, no digital or analogue filtering.

So, in my text that was quoted, I should have said, "Like Audio Note CD players," instead of, "Like Audio Note DACs." Thanks for the clarification!
 
Well, that's interesting! So if I get it right, they claim that they don't do any kind of anti-aliasing filtering at the analogue output? Meaning that they voluntarily create a very audible distortion? Either it's a typo, or those guys have missed the signal theory 101 course :idea:

If you read in reviews that those Audio Note CD players "sound different" or "bring something to the sound", then that's the explanation. Some people might find this pleasing on some music, but that's definitely the sign that they are distorting the sound in an odd way instead of properly reproducing it, and shows that the anti-aliasing filter is necessary for a proper reproduction of the signal.

Ah, this seems to confirm what I was saying: https://www.tnt-audio.com/sorgenti/audionote_cd11x_e.html. Among other things, the reviewer does say that "the ear will also need some time to adjust to the character of this CD player, as it clearly does things in a slightly different way as most other players" (translate: it noticeably colours the sound). Incidentally it's kind of funny to see how people can be fooled into finding "purist" a blatant marketing pitch telling them that less processing respects the sound better - answer: no, not when you remove an essential part of the process and its absence is bound to cause artefacts.

Which might well explain why they don't mention "no anti-aliasing of any sort" for their external DACs: the result was probably not to the taste of everyone and they had to find something else to sell. It's just a hypothesis but it's plausible.
 
A stepped signal at the output of a DAC is still analogue. It has lots of levels, not just the two of a binary digital stream.

That's what a Digital to Analogue Converter does...

The purpose of the reconstruction filter is to reduce the sinc imaging alias signals that are inherently produced by a zero order hold function (the DAC output sampling). That's also a very good reason to use a digital oversampling filter, as it shifts the sampling frequency up, thus shifting the sinc images up, and relaxing the requirement for the reconstruction filter, or increasing the image attenuation if the filter is left unchanged.

You don't even need to use a filter in your oversampler; simply doing a zero-stuffing upsample will move the sinc images up to the upsampling frequency..
 
Last edited:
Back
Top Bottom