Digital that sounds like analog
Posted 10th December 2012 at 04:18 AM by abraxalito
Updated 15th February 2013 at 06:11 AM by abraxalito
Updated 15th February 2013 at 06:11 AM by abraxalito
For those who missed Frank (fas42)'s link on a thread I started then here's where I'm continuing my minimal oversampling DAC developments for the time being : Digital that sounds like analog
Total Comments 108
Comments
-
Another thread over there, talking about jitter, demonstrating how ritualistic thinking builds up over a period of years, finally secures prime position for a time, and takes seemingly a similar number of years to fully fade away. Used to be with slew rate distortion, and in the digital age it's jitter.
Every time now digital sound is influenced by something, it [I]has to be[/I] a jitter effect. So, people dutifully measure real jitter parameters, find nothing, and conclude that everyone was imagining things ... :spin: :spin: :spin:
FrankPosted 5th March 2013 at 10:36 PM by fas42
Updated 5th March 2013 at 10:42 PM by fas42 -
Your comment just now about recent Decca is my experience: contemporary classical recordings are a curate's egg, easily the "poorest" standard of production of any era, irrespective of label ...
FrankPosted 9th March 2013 at 01:20 AM by fas42 -
Interesting reading that you and Pano are agreeing about the speaker 'disappearing act' and nobody else is curious. That's the effect I get when the electronics has very low IMD (or noise modulation). I have the same experience as Pano - when I put my ear close to the speaker its not like the sound comes from the cone's surface, rather behind the cone, inside the speaker. It happens with many speakers so long as the electronics is up to snuff.
Posted 12th March 2013 at 01:46 AM by abraxalito -
Exactly. But don't mention it "over there", :D, the abuse I copped for daring to suggest such things was quite amazing really. I guess for those who've spent big money on achieving something not quite so capable, it was a bit confronting ...
It's all a journey ... :cool:
FrankPosted 12th March 2013 at 02:22 AM by fas42 -
([I]Was not in a brilliant mood, felt like having a whinge ... got rid of the content ...[/I])
FrankPosted 14th March 2013 at 11:32 PM by fas42
Updated 15th March 2013 at 05:20 AM by fas42 -
Just tried reading Lynn's piece in Positive Feedback, Mountains and Fog: the Sound of Digital Converters, Part 1, and I can't take him seriously. Far too much flaky, and downright silly, thinking intrudes and it gets worse as you go along. I only made it less than halfway through, and gave up ...
It's difficult enough tackling the gung-ho objectivists, but then even the most reasonable of subjectivists do themselves no favours when they go into twaddle mode ...
FrankPosted 19th March 2013 at 03:47 AM by fas42
Updated 19th March 2013 at 03:49 AM by fas42 -
Posted 20th March 2013 at 03:01 AM by abraxalito -
Okay, ... ([I]takes deep breath ...[/I])
* First of all, he [I]does[/I] do an excellent job with the intro, the rundown on the subjective "what's wrong with the sound of a lot of digital today"
* Nice point about the dynamic characteristcs of paralleled output devices not matching, which doesn't help non-linear behaviour
But, then:
* "All the reviewers that wrote the glowing reviews were using high-powered (200 watts or more) Class AB transistor amps with plenty of feedback (very low distortion, high damping factor, etc.)" - the subjectivists prick up their ears, "bad" amps being used ....
* " My electronics are all-triode throughout, have zero local or global feedback" - every electronic circuit has feedback effectively as part of its operation, intentional or otherwise, they would sound horrendous otherwise. Point being made, is that he's listening with "good" electronics ...
* "The physics of the amplifying device dominates the sound, like it or not" - what??!! Yes, a SET device will "flavour" the sound, everyone acknowledges this, but -100dB distortion of a high performance SS device is "dominating" the heard sound?
* "we come to the DAC. My feeling is the converter itself, and its principle of operation, is what dominates the sound of digital sources." - you may disagree, but this is just wrong, wrong, wrong ...
* "lossy-compression digital sounds (much) worse; what do you expect when 90% of the sound is discarded and thrown away at the record end, and has to be guessed (interpolated) on final reconstruction?" - it doesn't sound much worse, if it is intelligently reconstructed, and there's no guessing involved.
* "Sony and Philips set the bar too low with a sample rate of 44.1kHz and a bit depth of 16 bits as the hard-and-fast standard for Compact Discs" - this old chestnut, yet again ...
* "A more serious problem is low bit depth; although (barely) acceptable for a consumer medium, 16-bit resolution proved unacceptable in the studio, since each track of a multitrack master is recorded, and later mixed, at different levels, and 16-bit systems have no extra resolution to throw away when track 1 is -3dB down, track 2 is -10dB down, and so on" - barely acceptable?? And,yes, if one is playing around with a DAW you need high bit resolution, this was nothing but a programming "issue", trivial to solve.
* "The difference between 16, 20, and 24-bit doesn't sound like much, but it is huge." - Rubbish, I've done enough experiments on real music tracks to know this is nonsense
Sorry, by this stage I felt I was back in the shop of the local hifi guru, in the 80's, who hated CDs with a deep passion. There may be something of value further in the article, but this anti-digital tirade was really irritating me strongly by this stage.
I gave up shortly after this ...
FrankPosted 20th March 2013 at 09:18 AM by fas42 -
Posted 20th March 2013 at 09:51 AM by abraxalito -
Yes, well, probably didn't help that it was late in the evening and I wasn't in a brilliant mood at the time; but every now and again it really rubs me the wrong way when all the typical half-truths and twisted viewpoints about digital appear, one after the other. Normally, I would probably skip over them and get to the "meat" of the piece; this time I wasn't so forgiving ... ;)
FrankPosted 20th March 2013 at 10:41 AM by fas42 -
1) 'Bad amps being used' - well he explicitly says he's not saying his amps are 'better' than those 200W transistor ones, so looks like your conclusion of 'bad amps' is not really supported by what Lynn's saying.
However here there is a generalization - that those kinds of amps tend to be ones picked out by people at least partly buying on technical specs, rather than purely by listening. The kind of sound will include more noise modulation than Lynn's valve amp (I conjecture - feel free to disagree). Apart from your amps being less powerful, they do fit the mould that Lynn's describing, so did you feel irritated that he's critiquing your own preferences? Amps with lots of feedback tend to have been designed 'by the numbers' and trade open loop gain for linearity.
This 'gain for linearity' brings me to a point raised in Bob Cordell's book about power amps, where he's got a chapter on zero-fb amps [gets book out]. The part I'm thinking of is a few pages before that chapter where he's introducing the controversy. Looking now at p502 in the section titled 'Spectral Growth Distortion' - he's showing a JFET, common -source config, and how its THD varies in harmonic structure as feedback is applied (fig24.3).
The figure looks rather damning for feedback because all the higher order harmonics grow in moving from zero to 30dB of feedback and only 2nd and 3rd improve. But then he later says this isn't really representative because open loop the performance is so bad (like almost 10% 2nd harm distortion). So I'm left wondering what happens with a starting point which is much lower (say 1% open loop distortion)? Guess I'll have to do my own sims on this.Posted 20th March 2013 at 11:12 AM by abraxalito -
The implication, carried through to the S-D DACs evaluation, is that feedback causes problems. Which of course it can - like everything, it's all about the implementation. The 'evil twin' of feedback is PSR, if the latter is not sufficiently optimised then FB most certainly will exacerbate unpleasant audible distortion, because the FB will drive the voltage rails into delivering sharp current pulses to offset the non-linearities.
I have no preferences for topologies: what works, works. All that matters is the final sound, the means to getting there is irrelevant so long as it doesn't introduce annoying artifacts, or a 'sameness' to the sound. Valve sound that I've heard has ranged from excellent, to highly syrupy and clouded; like SS, it runs the gamut.
Getting back to FB, chip amps are all about FB, whether internal or external, that's the only way they can work. And they obviously can do a good enough job. Where FB [I]may [/I]be causing problems in the S-Ds, even though it's digital and therefore theoretically shouldn't be a problem, is that it means that there is very high speed processing within the chip itself which is completely unrelated to the musical signal, and if there is internal crosstalk between this highly CPU-like digital dancing and the analogue output related electrical areas then this is where one could get what you call "noise modulation", in an audible form.
I would be careful with Bob's results with distortion ... in his work he has an "excellent" amplifier design which he has sim'd to give superb distortion results. However, attach anything vaguely like real power supplies, rather than the ideal voltage sources he used, and see the distortion figures go up in smoke ...
FrankPosted 20th March 2013 at 10:25 PM by fas42
Updated 20th March 2013 at 10:35 PM by fas42 -
No, the issue with S-D DACs is not that 'feedback causes problems' - rather that truncation causes problems and feedback is being used to make those problems go away. Which it only does to a certain extent - on average, according to the measurements (which always contain averaging). Nobody really understands putting feedback around a discontinuity, but rather like GM foods its put out into the market without full understanding.
So then you're not at all interested in 'Spectral Growth Distortion' ?Posted 21st March 2013 at 12:49 AM by abraxalito -
I think you might need to have another read of that section in Cordell's book: he very clearly states, in 2 places in fact, that it is a phenomenon that manifests with using low levels of feedback, which as Putzey points out is the "worst" way of incorporating FB - either don't use it all, or if you do, use lots of it.
Specifically, "(if) FB exceeds about 20dB the spectral growth stops and all orders of distortion decrease as FB is increased"
Also, "the audio community read too much into it, wrongly generalizing the results ..."
FrankPosted 21st March 2013 at 09:58 AM by fas42
Updated 21st March 2013 at 10:00 AM by fas42 -
Posted 21st March 2013 at 01:59 PM by abraxalito -
There is no problem at lower frequencies, typically: taking the usual gain of 20 for power amps, meaning 26dB closed loop gain. Open loop gains are usually of the order of 100dB, so 75dB or so of feedback.
Where things become interesting is at the 20kHz mark, a simple compensation scheme or internal compensation usually results in open loop gain around 60dB for this frequency, far less wriggle room for FB. Your passion for high speed opamps should be rewarded here ...
Of course, this compensation should also help tame the open loop distortion at these higher frequencies. This is where it all starts getting very complicated, and where the subtleties of "opamp" sound probably originate. Extremely accurate simulations probably would help unravel the finer points of what's happening, where distortion could be causing audible problems.
Personally, I would like to see the distortion plot remain relatively flat up to 20kHz, no starting to rise earlier than that, which to me indicates that FB is starting to lose the battle ...
FrankPosted 22nd March 2013 at 10:03 AM by fas42 -
Yes, I recall reading in Self that a gain crossover frequency of 2MHz is probably about the best that can be hoped for with a poweramp, thus only 40dB of feedback is available at 20kHz. But when I extrapolate the Spice sims in Cordell's book from 30dB to 40dB I don't see the higher harmonics going below their original levels with zero feedback. But there's a plot of output stage distortion and here all the distortion harmonics fall with increased NFB. My hunch is that output stage distortion, while dominant for large signal THD, isn't the dominant factor in SQ.
Posted 22nd March 2013 at 12:39 PM by abraxalito -
On that DSD vs. PCM thread, it's so depressing seeing the same old rubbish and clueless thinking repeated over and over again -- 50% loss going from hi-res to RB ... oh, dear!
FrankPosted 26th March 2013 at 03:21 AM by fas42 -
And now on a positive note, [url=https://www.whatsbestforum.com/showthread.php?10027-DCS-upsampler&p=183000&viewfull=1#post183000]DCS upsampler - Page 2[/url]. I'm impressed that dCS has discerned on their very high quality DAC the same phenomenon I determined for my cheap and nasty PC motherboard chip: that upping the resolution of the digital signal before feeding it to the circuitry improves things.
mep is bewildered that this happens; he's not able to separate the medium from the mechanism processing the medium ...
FrankPosted 27th March 2013 at 01:32 AM by fas42 -
Posted 27th March 2013 at 04:00 AM by abraxalito