Jump to content
IGNORED

DSD Offshoot Discussion From MQA Topic


Recommended Posts

12 hours ago, daverich4 said:

 

Except, the question hadn’t been asked ten times. I’m sure the topic of how DSD works has been brought up here in the past but I don’t remember seeing it in the 2 1/2 years I’ve been here. Lee made a statement about DSD that got the response from Mansr that Lee had no idea how DSD works. I have no idea if Lee does or not but I didn’t so I asked and got the response that I was too stupid to understand. Paul R weighed in with a link to an article by Paul McGowen explaining how DSD works and the response was that Paul McGowen was too stupid to understand. Is that really the way you respond to questions from your daughter? Telling her she’s too stupid to understand? Mansr apparently knows the answer to how DSD works but he’s not sharing.

 

Well said.  I am not an expert on DSD but I know enough to do recordings in double rate DSD.  It sure sounds good and, in general, all else being equal, I prefer it to hirez PCM as I said before.

Link to comment
1 hour ago, Paul R said:

Sure - try AES Convention Paper 5396, it is an overview at a basic level , and old enough to have been suitably refuted if there was much of anything incorrect in it.

I didn't read it, so I very well may be wrong but it seems that you don't have to look far: AES Convention Paper 5395 - Why 1-Bit Sigma-Delta Conversion is Unsuitable for High-Quality Applications (googlefirst link pdf)

Link to comment
30 minutes ago, crenca said:

 

Just to clarify (or perhaps muddy the waters further)  that last sentence means that the density (the D in PDM) does not actually change but rather remains constant right?

But there is a density of +1 or -1 values, so I can see why some call it a PDM scheme, even though I don't doubt that experts in signal processing would say it is different from PDM.

Link to comment
27 minutes ago, mansr said:

That DSD isn't PDM.

 

That's sloppy writing on their part. The output from a sigma-delta ADC can be multi-bit PCM. What they mean is that a 1-bit sigma-delta ADC differs from other kinds, such as flash or successive approximation, in the way described. These other types always produce a multi-bit output, unless you consider a simple comparator a kind of ADC. Or put differently, the only (practical) 1-bit ADC is of the sigma-delta variety.

 

Once in the digital domain, DSD is just a degenerate case of PCM with one bit, and thus two possible values, per sample.

 

You are perfectly clear, and also perfectly wrong.

 

Perhaps a source of confusion here is that the binary representation of two-level samples, i.e. the bit, is usually referred to as 0 or 1. In this case, that becomes misleading since the values actually represented are +1 and -1. It may help to instead think of the bit values as high and low.

 

Consider also unsigned binary representations, commonly used for 8-bit audio. Here, 0 corresponds to the most negative value of the signal, 255 to the most positive. The numeric values here must be interpreted relative to the reference point of 128. For instance, one cannot reduce the amplitude by half simply through dividing each sample value by two as that would create a DC offset.

 

In PCM, the sample value usually can be though of as representing the current location of the microphone, with zero being zero deflection. At least, that is what I was taught rather long ago, and I do not think it has changed. (forgetting about quantization error and all that jazz, just for the sake of argument.) 

 

What do you think that a -1 or a +1 represents? In PCM, it would represent a full deflection, 

 

In DSD it means one quanta towards full deflection of the indicated sign.  

 

Not the same thing.

 

While it is just conceivable you could label DSD as a degenerate form of PCM, it is much clearer to say that DSD represents audio encoded in a PDM format.  You could certainly do the same thing in PCM by labeling each sample as +1 or -1, but that would be silly. And the common ADC and filters would not work. 

 

We quite obviously each believe the other wrong and too stubborn to see the crystal clarity of our own opinions, so let's just drop it here, shall we? We can agree that DSD is a special case of - something. 
 

It's time to go listen to some music. :)

-Paul 

 

 

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment
38 minutes ago, danadam said:

I didn't read it, so I very well may be wrong but it seems that you don't have to look far: AES Convention Paper 5395 - Why 1-Bit Sigma-Delta Conversion is Unsuitable for High-Quality Applications (googlefirst link pdf)

 

(grin) Talk about a case of wishful reading, or lucky typo...   5396 - The paper's actual title is: 

 

Why Direct Stream Digital is the best choice as a digital audio format

 

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment
7 minutes ago, Paul R said:

 

(grin) Talk about a case of wishful reading, or lucky typo...   5396 - The paper's actual title is: 

 

Why Direct Stream Digital is the best choice as a digital audio format

 

 

Those are two different papers:

 

http://www.mariobon.com/Articoli_storici_AES/SACD_1.pdf

http://www.mariobon.com/Articoli_storici_AES/SACD_2.pdf

Sometimes it's like someone took a knife, baby
Edgy and dull and cut a six inch valley
Through the middle of my skull

Link to comment

I'm a visual sort of the guy, and I found this page to be good as a way of getting a handle on things - http://digitalsoundandmusic.com/5-3-7-the-mathematics-of-dithering-and-noise-shaping/.

 

Yes, it's talking about translating to 4 bits representation, but the principle for 1 bit representation is the same - just think of the digital coding changing state more rapidly with noise shaping, so that the average density can match the actual waveform more precisely at the frequencies that count, that of the audio signal - at the expense of the digital encoding being 'noisier', at frequencies "that don't matter".

Link to comment
16 hours ago, Paul R said:

In PCM, the sample value usually can be though of as representing the current location of the microphone, with zero being zero deflection. At least, that is what I was taught rather long ago, and I do not think it has changed. (forgetting about quantization error and all that jazz, just for the sake of argument.) 

 

What do you think that a -1 or a +1 represents? In PCM, it would represent a full deflection, 

 

In DSD it means one quanta towards full deflection of the indicated sign.  

In addition to my earlier reply, the concept described here is differential coding, which DSD also isn't. Suppose it were this simple. Everybody agrees that DSD64 has at least as good frequency extension as 44.1 kHz PCM. A PCM signal is permitted to do almost a full swing from one sample to the next. For a 64x oversampled 1-bit differential coding to achieve this rate of change, there could be only 64 steps from one extreme to the other. That would give it a resolution roughly equivalent to 6-bit PCM. Everybody agrees it does a lot better than this.

Link to comment
10 minutes ago, mansr said:

64 steps from one extreme to the other. That would give it a resolution of roughly equivalent to 6-bit PCM.

 

Math.  Nice.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

So if I can confuse things further with another sloppy analogy :) :

 

I think of DSD as being something like driving a car and making steering wheel adjustments.  Think of the signal as the "road" the "car" is traveling along, and the -1 or +1 value as moving the steering wheel to the right or left.  This also for me intuitively gets to the necessity of a relatively high sample rate to accurately follow the signal (relatively frequent small steering wheel adjustments to follow the road).

 

Multibit PCM, on the other hand, I think of more like drawing a map of the road (again the "road" is the signal), with the multibit values specifying the location of the road at any given point in time.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
5 minutes ago, NOMBEDES said:

Does anybody really know what time it is?

 

https://www.timeanddate.com/time/time-zones-interesting.html

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
6 minutes ago, Jud said:

I think of DSD as being something like driving a car and making steering wheel adjustments.  Think of the signal as the "road" the "car" is traveling along, and the -1 or +1 value as moving the steering wheel to the right or left.  This also for me intuitively gets to the necessity of a relatively high sample rate to accurately follow the signal (relatively frequent small steering wheel adjustments to follow the road).

That's the control theory connection Miska pointed to.

Link to comment

Back to the PDM topic:  Since the value of a DSD bitstream can only be evaluated as -1 or +1 in terms of the signal (that is, the "steering wheel" must as each bit arrives be "turned" "left" or "right," it can never simply be held still - though by turning rapidly back and forth, it can very accurately follow a straight line), and since the bits arrive once each clock cycle (though that cycle is very rapid), then the "pulses" (bit values) can never vary in density.  And thus a DSD bitstream can never truly be PDM.

 

At least, that's the impression I (unfortunately not having had any math courses for the past ~47 years) get from following my intuition based on my sloppy analogy.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
1 minute ago, mansr said:

A slight correction/clarification is in order here. When the bitstream says +1, it means you're supposed to turn the steering wheel all the way to the right at once, and left for -1. It's the mechanics (inertia, etc) of the car, corresponding to the low-pass filter of a DAC, that prevent it from jumping like made across the road.

 

Does the high sample rate help at all?  That is, the "car" can't move very far in a split second.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
1 hour ago, mansr said:

In addition to my earlier reply, the concept described here is differential coding, which DSD also isn't. Suppose it were this simple. Everybody agrees that DSD64 has at least as good frequency extension as 44.1 kHz PCM. A PCM signal is permitted to do almost a full swing from one sample to the next. For a 64x oversampled 1-bit differential coding to achieve this rate of change, there could be only 64 steps from one extreme to the other. That would give it a resolution roughly equivalent to 6-bit PCM. Everybody agrees it does a lot better than this.

 

(groan) I just spent twenty minutes answering this, only to loose the text somehow. That has never happened before. 

 

The gist is simple; this is the condensed version: 

 

Delta Modulation is based upon quantizing the change from sample to sample rather than absolute value of each sample. This is pretty basic, and is described well in the only reference I really have immediately to hand; David Goodman's article in the Feb 1969 Bell Systems Technical Journal, Volume 48. "The Application of Delta Modulation to Analouge-to-PCM Encoding."  

 

Sigma Delta modulation is an extension of Delta Modulation. It essentially combines the two modulators needed for Delta Modulation into one, and moves it before the modulator. This is possible because it is a linear operation, where the error is quantized from signal to signal and actually used as a predictor. This also has the wonderful effect of making the quantization noise frequency dependent. And, while Delta modulation is very sensitive to the rate of change of the signal, Sigma Delta modulation is not. (I believe it encodes the integral.) 

 

The key being that in both modulation schemes, the data used to reproduce the original signal is the change between samples not the absolute value of the samples. . It is plauged by quantization noise of course, but that is (one of) the reasons for the very high sampling rate.

 

So again, I come back to what I originally said. I still think it is a clear and concise way of describing DSD vs PCM in a non-technical, but correct, manner, and targeted for consumption by a non-technical audience. What exactly do you see is wrong with it, and how would you change it to be better?

 

-Paul 

 

 

DSD is the result of modulating a free running clock signal with an analog signal so that the output is a bit stream whose density is relatively proportional to the input analog signal. One of the defining differences between PCM and PDM is that in PCM, the numbers actually mean something - signal values are actually *encoded* in the values. (Pulse Code Modulation).  In PDM (DSD) the values themselves are meaningless - except in relative reference to the values surrounding them. 

Anyone who considers protocol unimportant has never dealt with a cat DAC.

Robert A. Heinlein

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...