Is the dynamic range of CD sufficient?
- Thread starter Al M.
- Start date
Julf said:
Thought not.
You seem to think that a 1-bit system can't be PCM just because it is a 1-bit system. OK, so a 16-bit PCM system is a PCM system. A 8-bit PCM system is a PCM system too. Wouldn't you agree that a 2-bit PCM system is a PCM system? Why is it so hard to accept the idea of a 1-bit PCM system - a system that is just the same as, let's say, a 8-bit PCM system, but has a word length of only 1 bit? What makes it a PCM system is not the number of bits. What makes it a PCM system is the fact that you have samples at fixed time intervals (and with no time-varying "pulse width"), with values that represent the signal. And yes, a 16-bit PCM signal has to be converted if you want it as 1-bit - just as 16-bit PCM has to be converted if you want it as 8-bit PCM.
You still CANNOT GRASP A 1-BIT SYSTEM MUST USE PDM-PWM, this is NOT THE SAME AS PCM; the encoding is very different and their definitions are very different otherwise SDM when created would had also been called PCM, it is not and is a 1-bit system.
If you are having problems grasping fundamental difference of PCM and PDM encoding it is your issue not mine
Yes giving in now before I start to swear at you .
Again you ignore every paper and quote I give to you; maybe you should ask yourself why they call one a 1-bit system and the other PCM, or why SDM internally has to convert from DSD-PDM to PCM and vice versa....
And now I have provided Sharp who are heavily involved with DSD components using exactly same distinction as Philips Research Labs and the separate application note (from another SDM/DSD manufacturer) I quote.
Is it me and these who are wrong with the distinction and being funny, or you trying to win an argument?
And yes I think I am right to mention EE/DSP because you argued the stair-step in the same way I have seen those who educated in DSP rather than EE (which funny enough I pointed out you can easily find on the internet for stair transfer function or if you have access/linked with to degree EE lecture notes).
You really don't seem to grasp that not every 1-bit system is PDM or PWM. Yes, there are 1-bit PDM systems. Yes, there are 1-bit PWM systems. But there are 1-bit PCM systems too - just as there are multi-bit PCM systems. What makes the number "1" so special to you?
I do understand the difference between PCM and PDM. Do you? (Hint - it is not the number of bits per se).
*Multi-bit* PCM. Please understand that that is the important distinction. See my previous response that you don't seem to have read.
I am afraid you are focused on a very low-level basic electrical engineering college "intuitive" mental picture of what PCM and DSD are. Try to think, on a more abstract level, about what it is that distinguishes PCM from PWM. Don't just think about a picture of a wave with steps - think about the mathematical properties and the definitions.
This really isn't going anywhere. You clearly don't have a clear idea of a definition for PCM. Until you can at least define what you mean with PCM, this is really pointless, and this has, in any case, drifted pretty fart away from the original question in this thread, so I am happy to stop here, unless you provide your own clearly articulated definition of PCM so that we can have a meaningful discussion.
This thread is about audio-original, and argument started with PCM being said to be the same SACD-DSD; so your first response is academic (not every 1-bit system is PDM-any similar solution or PWM) to what has been argued so far.
Really so that means Philips Research Labs did not got it nor any of those application notes I quoted do not "grasp what differentiates a 1-bit system to PCM".
And I guess you do not realise what you have been basically describing is actually DPCM; but ah well thought you might actually get it with my hints on why one is called PCM and the other 1-bit system in all of those actual DSD papers, or why there must be a conversion from PCM to any 1-bit sytem encoding.
Was not going to mention DPCM/Delta Modulator because that is now truly going to screw with many reading this thread; but DPCM is NOT PCM.
To put it to rest one more time; DPCM represents differences between samples (your 1-bit system that evolves to modern use SDM or DSD-PDM-etc), PCM represents specific absolute sample values (based upon discrete sample interval with amplitude of signal at that point) - this then has implications for quantization/aliasing/noise/etc depending upon which encoding solution one uses.
BTW it is meaningless how you keep on about multibit PCM in the context of DSD, I cannot remember nor that bothered now but I think one of those DSD Philips papers explain why, if not them then it is explained within another paper by either Philips,AD, or Sharp.
Anyway your right I know nothing and bow down to your wisdom and even though they are not the same now accept your logic; PCM is in fact DPCM which is in fact DM which is in fact SDM, meaning PCM must be SACD and DSD.
Anyway thanks for pointing out I know nothing and that everything I posted is meaningless because it is low-level basic EE and not a more "abstract level".
Would help if you can link a real 1-bit PCM DAC used in audio, that does not use SDM (bear in mind delta modulation-DPCM fits in with Sigma Delta Modulation)/some sort of PDM-PWM.
As I already mentioned much earlier, the Philips Research paper mentions DSD does not necessarily rely upon SDM but still needs a comparable solution (but show me a real one that enables such 1-bit system without SDM).
Really so that means Philips Research Labs did not got it nor any of those application notes I quoted do not "grasp what differentiates a 1-bit system to PCM".
And I guess you do not realise what you have been basically describing is actually DPCM; but ah well thought you might actually get it with my hints on why one is called PCM and the other 1-bit system in all of those actual DSD papers, or why there must be a conversion from PCM to any 1-bit sytem encoding.
Was not going to mention DPCM/Delta Modulator because that is now truly going to screw with many reading this thread; but DPCM is NOT PCM.
To put it to rest one more time; DPCM represents differences between samples (your 1-bit system that evolves to modern use SDM or DSD-PDM-etc), PCM represents specific absolute sample values (based upon discrete sample interval with amplitude of signal at that point) - this then has implications for quantization/aliasing/noise/etc depending upon which encoding solution one uses.
BTW it is meaningless how you keep on about multibit PCM in the context of DSD, I cannot remember nor that bothered now but I think one of those DSD Philips papers explain why, if not them then it is explained within another paper by either Philips,AD, or Sharp.
Anyway your right I know nothing and bow down to your wisdom and even though they are not the same now accept your logic; PCM is in fact DPCM which is in fact DM which is in fact SDM, meaning PCM must be SACD and DSD.
Anyway thanks for pointing out I know nothing and that everything I posted is meaningless because it is low-level basic EE and not a more "abstract level".
Would help if you can link a real 1-bit PCM DAC used in audio, that does not use SDM (bear in mind delta modulation-DPCM fits in with Sigma Delta Modulation)/some sort of PDM-PWM.
As I already mentioned much earlier, the Philips Research paper mentions DSD does not necessarily rely upon SDM but still needs a comparable solution (but show me a real one that enables such 1-bit system without SDM).
Last edited:
Sigh,
remember how I had to correct you much earlier because you made a pretty poor screw up and then it seems really do not understand the differences or deliberately misapplying them.
Here try another research related engineer for yourself, because obviously I know jack **** - YOU DID READ HOW THE ENCODING OF DPCM AND PCM ARE FREAKING DIFFERENT IN MY PREVIOUS POST.
As I said: "To put it to rest one more time; DPCM represents differences between samples (your 1-bit system that evolves to modern use SDM or DSD-PDM-etc), PCM represents specific absolute sample values (based upon discrete sample interval with amplitude of signal at that point) - this then has implications for quantization/aliasing/noise/etc depending upon which encoding solution one uses."
THEY are not the same from a technical perspective.
Going by your argument you are saying PCM is Sigma Delta Modulation because DPCM utilises Delta Modulation, this then evolved into SDM....
Yet as I mention DPCM is like PDM and not PCM; hence PCM has to be converted, anyway please show me where we use DPCM in the audio world that is to do with this thread...
Bear in mind as I said (and so do other engineers) DPCM/DM has much more in common with DSM and PDM, it is closer to them than PCM.
Sure your going to now argue how you always meant DPCM and ergo SDM/DSD is the same as PCM even though the principles and encoding is radically different.....
I swear your going to make me now post engineer papers/links where they explicitely mention how DPCM is not the same as PCM.
If they were the same you would not be calling it DPCM, it would be 1-bit PCM encoding.
But hey lets ignore they use very different techniques and that DPCM is actually closer to modern day PDM.
Just because it is called DPCM does not mean it has the same nature-behaviour-definition as PCM, but hey strange how it was me who mentions DPCM and not you with regards to 1-bit operations.
Anyway looking forward to you still showing a real world audio product using:
a) DPCM.
b) 1-bit solution that is not DSM (PDM-PWM).
Last edited:
How can I be thinking of LPCM when I have been providing research papers who worked on DSD that show the difference between PCM and SACD/DSD.
So not only myself have it wrong in context-definition but the research engineers involved with DSD do as well, and so do the application notes by various manufacturers.
Even when scrolling through this thread so that reading is impossible, this thread manages to be a tedious, head-banger. I may have to go argue with micro about something just to clear my head. 
Tim
Tim
Look Julf,
IF DPCM/Delta Modulation was a subset in the way you present and therefore the 1-bit PCM you were always saying is same as DSD; Philips Research Labs engineers involved with DSD would not go on to say PCM and DSD are different, NOR would others such as Merging Technology/Audio Precision/other application notes for DSD components; this is clear in the papers and quotes I provided.
All of them differentiate between PCM and 1-bit DSD, or PCM and PDM (which is closer to DPCM/DM than PCM is)
Anyway.
The only audio stream ADC products that exist (hence why I asked you to find specifically a 1-bit PCM product in audio) is either PCM (not 1-bit) or DSD (1-bit).
In terms of DACs we have.. PCM (again not 1 bit), or SDM (1-bit) that must convert for PCM or can keep native DSD and bypassing the conversion processes.
Technically you can have DPCM/DM but again that in operation is more like PDM than PCM, however again find me a DPCM ADC/DAC in the audio world, or one that says 1-bit PCM.
And yes I have worked with ADPCM extensively in the distant past when it was more relevant than today.
Anyway there is a reason why all those engineers differentiate between 1-bit system and PCM (suggesting they do not see DPCM/DM in same way as you do).
Orb
Last edited:
Even when scrolling through this thread so that reading is impossible, this thread manages to be a tedious, head-banger. I may have to go argue with micro about something just to clear my head.
Tim
This all kicked off because JJ said the following where I then went on to point out SACd/DSD actually use PDM (or something very similar) rather than PCM and I was informed how wrong I was by both JJ and Julf, then eventually I ended posting various engineering application notes or actual Philips Lab Research engineer paper on SACD/DSD where they also clearly state PCM and DSD are not the same thing.
I just was being picky because it is not actually PCM-encoding, since then it snowballed on arguing 1-bit PCM is same as DSD although doing so would mean changing the encoding from PCM to DPCM/Delta Modulation (but Julf never mentioned that and for some idiotic reason I decided to recently point that out to Julf after he stated my grasp of 1-bit PCM is lacking) and even then DPCM/DM is closer to PDM than it is PCM.
That said I did not expect this to snowball to 15 further pages on what is or is not PCM and DSD, although pointing out the stair transfer function-etc within EE probably managed to take up a good couple of those as well
Cheers
Orb
Last edited:
This all kicked off because JJ said the following where I then went on to point out SACd/DSD actually use PDM (or something very similar) rather than PCM and I was informed how wrong I was by both JJ and Julf, then eventually I ended posting various engineering application notes or actual Philips Lab Research engineer paper on SACD/DSD where they also clearly state PCM and DSD are not the same thing.
I just was being picky because it is not actually PCM-encoding, since then it snowballed on arguing 1-bit PCM is same as DSD although doing so would mean changing the encoding from PCM to DPCM/Delta Modulation (but Julf never mentioned that and for some idiotic reason I decided to recently point that out to Julf after he stated my grasp of 1-bit PCM is lacking) and even then DPCM/DM is closer to PDM than it is PCM.
That said I did not expect this to snowball to 15 further pages on what is or is not PCM and DSD, although pointing out the stair transfer function-etc within EE probably managed to take up a good couple of those as well
Cheers
Orb
It's all good. Didn't understand it anyway.
Tim
First, let's sort out all of the mistakes accumulated there.
1) Delta mod is not SACD, or vice versa, Delta mod is 1 bit DPCM, no more, no less. It requires a decoder with a predictor, just like DPCM. So arguments about that are not germane to this thread.
2) SACD is demonstrated absolutely and incontrovertible to be a form of PCM. Simply put an antialiasing filter after the DAC (1 bit or otherwise) and presto, with no further ado, you have a linear system with added noise.
3) The fact that people made up other names for 1 bit noise shaped PCM proves nothing more than that they made up other names. People call 16 bit PCM at 44.1kHz "redbook", that doesn't mean it's not PCM. SACD is exactly similar in that regard.
4) Go ahead and write all of the excuses for advertising nomenclature you want, DSD and SACD are still forms of delta sigma, and therefore PCM. The fact that one can go all the way to 16 bits with the same technique ought to be a hint. If you look at some of the noise feedback methods used in Redbook by a few folks that shape the noise floor, it's the same technique. Nobody calls those PDM, but they are just the same thing as SACD, except for the bit resolution and noise shaping filter. It's all a continuum of PCM techniques.
5) Engineers do not distinguish automatically between the two. You've now claimed to know more about the subject that two people who have attempted to educate you in this thread, and you make it clear from your faux-apology that you are simply trying to play to the uneducated in some kind of "king of the hill" nonsense.
PCM/DSD/SACD are all a continuum of the same technique.
This PDM thing does not carry the appropriate information to require that the behavior of the system is linear with added noise. It does not describe the behavior of SACD. It is a poor descriptor for SACD at best. It is a necessary, but incomplete description, while "PCM with noise shaping" is a complete description. A*PCM as in Jayant-Noll can be (but does not have to be, it is a superset) demonstrated to fully describe both DSD and SACD performace, without the step-size adaptation, even. That, by itself, should be a complete proof that your assertions are based in advertising semantics.
As there has been a lot of confusion and misunderstanding, I just want to summarize the points from this discussion, so that we can move on to more interesting things. Some of it is relevant to the original topic, some of it isn't.
Just in case someone still wants to debate them, I am presenting them as numbered items. If you disagree with any of them, please indicate which specific points you disagree with, and why.
JJ and others, please feel free to correct things I have gotten wrong or expressed badly.
Can we move on now? Somehow appropriately, my morning paper had this Calvin and Hobbes strip:
Just in case someone still wants to debate them, I am presenting them as numbered items. If you disagree with any of them, please indicate which specific points you disagree with, and why.
JJ and others, please feel free to correct things I have gotten wrong or expressed badly.
- There is no staircase involved, except at the intermediate stage of a binary-weighted DAC (such as a non-oversampling R-2R ladder DAC) before the required reconstruction filter.
- The talk about transfer functions is irrelevant. A transfer function describes the relation of the output to the input in a system, so it might apply to an amplifier or a DAC, but it does not apply to a signal.
- The function describing a digital signal is a time series of values at fixed intervals. Thus the function is only defined at the specific sample points, and undefined elsewhere. It is *not* a staircase function (despite often being displayed as such on processing software displays.
Thus like this:
and *not* like this:
- There are different encoding methods. One of them is PCM - Pulse Code Modulation. There are many variations of PCM, such as:
- LPCM - Linear Pulse Code Modulation
- DPCM - Differential Pulse Code Modulation
- Delta-Sigma modulation
- DSD (SACD) uses delta-sigma modulation. SACD uses a 1-bit form. There are also multi-bit forms of DSD ("DSD-wide"), not to be confused with DXD, a 24-bit LPCM format.
- There are different encoding methods. One of them is PCM - Pulse Code Modulation. There are many variations of PCM, such as:
- This all only matters to the original topic in that 16-bit, 44.1 kHz-sample-rate PCM is totally adequate to represent the
- The entire audible frequency range
- The full dynamic range of the original signal with a noise and distortion margin well beyond limits set by audibility and source signal
- Timing to a much higher precision than a single sample interval
Can we move on now? Somehow appropriately, my morning paper had this Calvin and Hobbes strip:
JJ why you say I am making mistakes when you misconstrue the whole context of my posts-narrative (which Julf I see is doing more so now).
a) I clearly said to Julf Delta Modulation is part of DPCM BUT this is NOT PCM (nor SACD but DM has links to SDM),Delta Modulation is closer to PWM-PDM than PCM;
Is PCM predictive coding and suffers from slope overload (both associated with DPCM/DM not classical PCM) or encode based upon code words that represent differences between samples (again DPCM not PCM)? answer is no they are not PCM to both questions.
Furthermore Delta Modulation IS part of the discussion considering both you and Julf go on about both DSD and SDM (that is an evolution of DPCM-DM and overcoming the many issues associated with DM) being PCM.
It seems to me your oversimplifying this back to PCM, in same way Ethan oversimplifies distortion to a basic point where you both want to ignore the definition-nature-behaviour of specifics.
By your logic SDM/PWM-PDM and PCM are the same, which they must be for you to state:
Anyway by your logic Audio Precision,Merging Technologies/Philips Research Lab (who defined a practical from concept Delta Modulation in 1950s,and much later coinvented DSD)/Analogue Devices/Sharp/Motorola/etc are all wrong, especially when they differentiate between PDM (and similar encoding) and PCM, or 1-bit systems (that MUST use some kind of PWM-PDM encoding)and PCM.JJ said:
Bear in mind this discussion thread scope/context IS about the existing audio world.
Anyway it was myself that pointed out DPCM/DM/SDM to Julf and how he was really arguing about that rather than PCM; but then he does not see any difference between PCM and DPCM and importantly the even further simplified Delta Modulation/SDM, using that logic all IIR and FIR filters are also same in definition-nature-behaviour and so are minimum phase/linear phase filters....
As I said to Julf, yes I appreciate what DPCM is because I worked extensively with ADPCM (including at a low level framing structure) when it was more relevant to the world of digital transmission and requirement for efficient use of bandwidth, I assume you did as well due to being at Bell Labs back then.
Anyway if you must say "PCM with noise shaping" that is relevant to this thread and audio then actually say DPCM-DM/SDM (oversampling-noise shaping), rather than trying to prove a narrative that PCM (definition-nature-behaviour and encoding) is same as SACD (which utilises PDM not PCM) as they are not.
Or you could argue I am applying (and all those others I mentioned who technically differentiate between PCM and 1-bit systems-solutions-encoding) the traditional/classical definition-framework to PCM (includes application-nature of encoding), which still existed even when DPCM/DM was invented btw and evolved into SDM.
But then this started and snowballed when you and Julf said I was wrong that SACD/DSD use PDM encoding (or something like that) and you both said it is PCM....
Julf [personal commentary omitted] and deliberately being obtuse or not bothered about technical/EE differences so giving up.
Last edited by a moderator:
Just in case you somehow again overlooked or misread what I wrote, here is a reminder:
"You are wrong because the guys who were promoting the stuff described it differently back in the day" and calling people names is not very helpful.
I am afraid that I have to take your inability to actually address any of the factual points as a sign that you don't actually understand them.
Can we now move on to more interesting things, or at least actually discuss specific items (with proper definitions for what we talk about) instead of endless "I am right because I think that is whet the Philips guys said"?
"You are wrong because the guys who were promoting the stuff described it differently back in the day" and calling people names is not very helpful.
I am afraid that I have to take your inability to actually address any of the factual points as a sign that you don't actually understand them.
Can we now move on to more interesting things, or at least actually discuss specific items (with proper definitions for what we talk about) instead of endless "I am right because I think that is whet the Philips guys said"?
Thanks for clarifying those distinctions. Makes sense now, looked at in this manner.
This whole discussion, as a biproduct, just invalidated the entire "analog can capture nuances that digital can't" argument. Why would anyone listen to vinyl records with 3% THD and -65dB surface noise, versus a clean digital signal?
Thanks for clarifying those distinctions. Makes sense now, looked at in this manner.
This whole discussion, as a biproduct, just invalidated the entire "analog can capture nuances that digital can't" argument. Why would anyone listen to vinyl records with 3% THD and -65dB surface noise, versus a clean digital signal?
Because everyone who loves LPs are basically very stupid people who just don't understand how much better CDs sound than LPs. Maybe one of these days all of the idiots that love LPs will come to their senses and realize how wrong we are for preferring the sound of LPs over their digital third cousin who is twice removed.
It's worth noting that there are also a fair number of music lovers / audio enthusiasts (I don't know if they are really "audiophiles") who don't think it's worth using any equipment made or designed in the last 25-30 years. We all have our preferences and it should be obvious that verbal (i.e., written) discussion is unlikely to change that, especially if the argument is over whether one is "better" than the other.
Similar threads
