[liamstrain]

"Phonograph Playback: It's better than you think!" By Dr. Bruce Maier and Jon Risch  ( the Discwasher team  provided some of the best reasearch into phono ever made )

 
I'm certain they are totally unbiased, and have no commercial interest in making LPs more attractive to modern audiences. :/

 

[bigshot]

Remember CED video disks? They were disparagingly called "needle vision". They actually squeezed video out of a vinyl record- big, bulky, low resolution, they skipped. They sucked royally. Now we have blu-rays that contain hours of perfect high definition video and multichannel sound on a tiny silver disk. It's all about the amount of information packed on a disk and the accuracy of playback. Optical media wipes the floor with needles and grooves. Why would anyone believe that LP records sound better than CDs? Absurd.

 

[Greenears]

 
I agree, that's why I did warn about the work itself and its significance. but if they had truly kept people in the dark, then we would have missed the nice differences between sighted and blind listenings showing all the bias so many dignified audiophiles would never admit possible.
 
I've read about a few trials, usually with small number of participants, and each time whatever the result, you get the opposing team giving the same justifications.
if they succeeded in telling 16 from 24 or 16 from DSD, then we end up looking for problems in the setup and usually we find some.
if they couldn't tell the formats apart, the we end up with the fact that the test wasn't done on good enough gears, or was with songs the testies didn't pick, or even that the test is wrong and stressful or inviting null results.
so as always, the only thing we can know for sure, is if I as an individual can hear a difference. past that there will always be people saying that the results are wrong because else they themselves would be wrong and they can't fathom the idea.
 

Well ultimately we all have our limits.  The antidote for "gear not good enough" is simple, just pick out the absolute most you would ever spend on a chain of equipment.  Then audition that, if you can't get an ABX positive result, then ABX more modest equipment against the expensive one and if that is null you've saved yourself quite a bit.  Actually it may get overall better if you reassign your dollars to things like speakers. 
 
I have a practical antidote I've used to the AB loudness issue if you don't have the setups mentioned above handy.  Again you need a disinterested 3rd party to help you.  What you do is before each A-B switch, the assistant turns the volume control to very low.  Then after the switch you ask them with hand signals to bring it up to any level you like.  Remember the level is in your control and you can set it or change it as much and as long as you like.  But it takes away that volume cue.  It doesn't give you the nice instant switch, but a null result here is still very instructive, since we don't listen to instant switching I real life.  I found that if I can't tell A from B beyond a guess after going back and forth multiple times at multiple levels, I felt good that whatever A-B difference was really didn't matter to me. 

 

[analogsurviver]

   
I'm certain they are totally unbiased, and have no commercial interest in making LPs more attractive to modern audiences. :/

No, they were (at the time, 80s ) not - see my post above - yet what they say IS true. And IS audible with the right equipment that does not have to cost an arm and a leg, but is more expensive than what is required for CD.
 
It is indicative that Matsusiita decided to recently revive the Technics brand precisely because of hi-rez digital; 
despite being in hiatus for over a decade, Technics engineers did not lay looking in air, but kept on working and came up with some great new developments - and must have not been exactly thrilled by the CD being unable to give them a true workout.
 

 
The "normal people" range seems very interesting, and as an owner of Technics SB-RX50 speakers from mid 80s ( acquired three years ago and refurbished/modified with up to date parts ), I can tell you that this updated version will have an impact on the market; it is VERY hard to position it properly, it does require, despite small size, rather large(ish) room to really breathe, but those who will take the trouble will see their efforts compensated many times over - by the almost holographic imaging. IF recording does have it.  And LP/HiRez vs CD on SB-RX50 does sound like a bad joke - since this is head-fi, SB-RX50 was/is perhaps the closest one can get to Stax Lambda in loudspeaker world - or answer to the question "Is there a speaker one can listen after Stax?". The modern one has better non square box cabinet that should have far lesser diffraction effects and reduced standing waves within cabinet; if they managed to keep the HF performance of the coaxially mounted tweeter while allowing a bit better performance at its lower end of response, this one should be a clear winner.
 
Although the first question majority will have in mind (where is the turntable ?) is not that interesting to me, I do wonder if they plan to reintroduce updated phono cartridges - that to this day are second to none.

 

[bigshot]

  No, they were (at the time, 80s ) not

 
I remember the 80s. There were all kinds of spurious articles by experts claiming that digital audio was inferior to old formats. I remember reading one by Lincoln Mayorga and Doug Sax that had all sorts of info about stair steps and sampling flicker and complete hooey. Later on, they wised up and admitted they were wrong and just didn't understand how digital audio worked. But the seed was planted in golden eared audiophiles who report hearing stair steps and jitter to this day.
 
LP records are a fine format, capable of good sound, but redbook exceeds that on every single measure of quality. There were a couple of small bumps in the road early on, but digital audio has now achieved its goal of perfect sound.
 
"in terms of pricing, the description of 'reference' seems perfectly justified." <--- The stupidest thing I've heard so far today.

 

[Greenears]

 
As far as multi-bit ΔΣ goes, it's sad but unsurprising that the chipmakers don't reveal the details of their implementation in datasheets. We might be able to infer some features by looking at the filtering, though. For instance, an early DSD DAC like the PCM1792 boasts an extensive 4-stage analog FIR filter to get rid of the DSD noise. Move forward 5 or 6 years to the CS4362A and we see that the 50kHz DSD filter is now implemented in the digital domain before it hits the DAC, suggesting that they may indeed be extending the wordlength to full-scale PCM.

 
The updated version of PCM1792, the DSD1792A still has the analog FIR filter for DSD.  That's their top-end dual PCM/DSD DAC. With DSD and PCM performance well over 120 dB it is really hard to imagine that this is the limiting factor in any end-to-end system, given the pesky little detail that you need transducers. 
 
http://www.ti.com/lit/ds/symlink/dsd1794a.pdf
 
I'm not really with the PCM conspiracy camp on this.  It would be cool to know more detail on all of these, but the detail may have been withheld because it is boring or they don't want their competitors to copy them (IE because it is good not bad).  Hard to know for sure.  Note that all of the main 3 DACs convert all incoming formats to multi-bit

 

[Greenears]

  There is at least one outside the companies that did the homework - beyond the companies/manufacturer. iFi Audio - or its parent AMR. They went to test every imaginable chipset and found one relatively "vintage" one that not only can do PCM and DSD - both natively - but also can perform way above manufacturer's specifications; they wrote their own protocol in order to squeeze the last bit of performance available from this chipset. Modern day chips simply do not allow for this kind of performance - proving again the point that when any technology is being made available for the first time, they try to make it best they can - then come cost cutters in the subsequent generations...
 

I checked out the iFi Audio website and on page 2 of their technical brief there is a "square wave" with sloped edges that they show as fuzzy out of the DAC and "clean" after their magic circuit.
 
Let me put this in measured technical terms: It is complete hooey.  I'm sorry you've been had.  There is no such wave that you or any independent 3rd party has measured on the DAC output before or after their "cleanup".  This pulse makes no sense as a test waveform anyway, and I can assure you with modern DACs of >100 dB performance you cannot visually see anything like that on an O'scope. Period.
 
If you can A-B detect the iFi box, which I doubt, then compare it to a source with an equalization option.  All you are hearing is a simple low pass filter.  A laptop equalizer will match your result and you can A-B that.
 
I joined this "Sound Science" forum because I really liked the title.  There is valid science to discuss about sound, but we also need to retain the sound principals of science.  Brilliant title.

 

[bigshot]

The nice thing about the internet is that when someone types out a whole bunch of stuff that doesn't relate to the subject of the forum, you really aren't required to read beyond the first line or two. At least that is my theory!

 

[analogsurviver]

  I checked out the iFi Audio website and on page 2 of their technical brief there is a "square wave" with sloped edges that they show as fuzzy out of the DAC and "clean" after their magic circuit.
 
Let me put this in measured technical terms: It is complete hooey.  I'm sorry you've been had.  There is no such wave that you or any independent 3rd party has measured on the DAC output before or after their "cleanup".  This pulse makes no sense as a test waveform anyway, and I can assure you with modern DACs of >100 dB performance you cannot visually see anything like that on an O'scope. Period.
 
If you can A-B detect the iFi box, which I doubt, then compare it to a source with an equalization option.  All you are hearing is a simple low pass filter.  A laptop equalizer will match your result and you can A-B that.
 
I joined this "Sound Science" forum because I really liked the title.  There is valid science to discuss about sound, but we also need to retain the sound principals of science.  Brilliant title.

Please visit http://www.head-fi.org/t/711217/idsd-micro-crowd-designed-phase-3-show-a-little-leg-what-is-it-page-132 - it IS long, somewhere is the exact description of the DAC used - or was it here ? http://www.head-fi.org/t/683406/ifi-audio-nano-idsd-discussion-impression that the DAC was discussed in great detail - including someone trying to prove it can not do what it does according to DAC mfr specs - only to find that it is indeed capable of claimed performance.
 
I have not seen any square waves off nano or micro that were arrived at with real hardware, not PC generated - and did not look at anything iFi posted because of that. I will record some using analog square wave generator and record 1 kHz ( approx - potentiometer for frequency setting should get replaced, good ones are costly, for sine wave up to 20 kHz I prefer precision of SineGen anyway ...) to Korg MR-1000 DSD recorder at the same level in ALL formats/resolutions it is capable of:
 
DSD64, DSD128, PCM 24bit 192/176.4/96/88.2/48   16 bit 48/44.1
 
I will also record low frequencies to see any high pass filtering in nano - Korg MR-1000 has -3dB point very low, approx 5 Hz from memory but will re-check while at it. And for fun I will record say 6 kHz square wave - this should really show the difference among various resolution files. All will be played on nano, with which I will also do the ABX of PCM files. I can also resample the DSD128 to all the "lower" versions and compare them to native recordings in the same resolution. That will be quite some work - I am NOT looking forward to listening to square waves, but for the sake of science...
 
I will post pics off the analog oscilloscope screen for the above. This gives me creeps - I am the last person anyone should hire for photography...
 
Please do send me a link to the square wave and pulse from ifi in question.

 

[charleski]

  There is at least one outside the companies that did the homework - beyond the companies/manufacturer. iFi Audio - or its parent AMR. They went to test every imaginable chipset and found one relatively "vintage" one that not only can do PCM and DSD - both natively - but also can perform way above manufacturer's specifications; they wrote their own protocol in order to squeeze the last bit of performance available from this chipset. Modern day chips simply do not allow for this kind of performance - proving again the point that when any technology is being made available for the first time, they try to make it best they can - then come cost cutters in the subsequent generations...
 
Without their effort, DSD512 or corresponding DXD would not have been a reality. Now everything else has to catch up, about three months ago I was checking there was no DSD512 recorder available - yet.

iFi uses Burr-Brown DACs with a multi-stage FIR filter for DSD, see the PCM1792 datasheet I linked above. See the PCM1795 for a more modern version of this design. The disadvantage is that the analog audio output is going through a bank of switched capacitors followed by 4th-order Bessel filters, a direct-charge-transfer stage to buffer the transients, and finally a 2nd or 3rd-order RC filter (see section 7.5.1 of Schreier and Temes). It's relatively expensive to implement correctly, but capable of producing perfectly acceptable results. There's nothing especially notable about it, except that it takes some skill to design properly.
 
[Edit]I see one of your links talks about using the old DSD1700 architecture with its 8-tap filter. You need only look at the huge amounts of residual ultrasonic noise shown in Figure 8 to see why TI moved to a more modern design. This is probably workable in a completely-integrated design, where you can ensure you won't be driving instability in the power stages, but otherwise may produce undesireable results.
 
 
As a matter of interest, I came across this patent which explains the method used by CirrusLogic in more detail. It talks about 'volume control by processing directly on the 1-bit data' and this presumably relates to their digital implementation of the 50kHz filter as well. It turns out that they are considerably extending the word length in order to perform this, but then decimate down to feed to their 4-bit ΔΣ DAC: "Delta-Sigma modulator 201, whose noise transfer function is generally the high pass response shown in FIG. 2B, re-codes multiple-bit data generated from the volume scaling multiplication into multiple-bit data having m number of levels. For example, the scaled data may be 16 bit and the output of the modulator more completely quantized Delta-Sigma modulated 4-bit data." Since everything is going through their multi-bit DAC they don't need to include the aggressive filtering discussed above.

 

[analogsurviver]

  iFi uses Burr-Brown DACs with a multi-stage FIR filter for DSD, see the PCM1792 datasheet I linked above. See the PCM1795 for a more modern version of this design. The disadvantage is that the analog audio output is going through a bank of switched capacitors followed by 4th-order Bessel filters, a direct-charge-transfer stage to buffer the transients, and finally a 2nd or 3rd-order RC filter (see section 7.5.1 of Schreier and Temes). It's relatively expensive to implement correctly, but capable of producing perfectly acceptable results. There's nothing especially notable about it, except that it takes some skill to design properly.
 
As a matter of interest, I came across this patent which explains the method used by CirrusLogic in more detail. It talks about 'volume control by processing directly on the 1-bit data' and this presumably relates to their digital implementation of the 50kHz filter as well. It turns out that they are considerably extending the word length in order to perform this, but then decimate down to feed to their 4-bit ΔΣ DAC: "Delta-Sigma modulator 201, whose noise transfer function is generally the high pass response shown in FIG. 2B, re-codes multiple-bit data generated from the volume scaling multiplication into multiple-bit data having m number of levels. For example, the scaled data may be 16 bit and the output of the modulator more completely quantized Delta-Sigma modulated 4-bit data." Since everything is going through their multi-bit DAC they don't need to include the aggressive filtering discussed above.

Yes, it is PCM1792 - and yes, there is a reason why they did not use on paper better PCM1795 - "somewhere" in the nano and micro threads here on head-fi.
PCM1792 can be made to perform above spec - and iFi discovered that and wrote their own protocol for it to get maximum out of it. nano uses single PCM1792 , micro uses two and hence can double on sample rates.
 
Thank you for the CirrusLogic link - it is DAC used  in Korg MR-1000 and MR-1, will check exact models of DAC ASAP. 

 

[SilentFrequency]

Remember CED video disks? They were disparagingly called "needle vision". They actually squeezed video out of a vinyl record- big, bulky, low resolution, they skipped. They sucked royally. Now we have blu-rays that contain hours of perfect high definition video and multichannel sound on a tiny silver disk. It's all about the amount of information packed on a disk and the accuracy of playback. Optical media wipes the floor with needles and grooves. Why would anyone believe that LP records sound better than CDs? Absurd.

I guess that explains why my dad replaced his entire vinyl collection with CD's?

 

[The Walrus]

Hey guys,
Can someone refer me to a site that clearly explains the audio engineering terminology in their relevant context from the beginner level up? (like dither, dynamic range, PCM, bit depthe, bit rate etc. etc. ) 
 
cheers. 

 

[charleski]

  Hey guys,
Can someone refer me to a site that clearly explains the audio engineering terminology in their relevant context from the beginner level up? (like dither, dynamic range, PCM, bit depthe, bit rate etc. etc. ) 
 
cheers. 

Here's one:
http://www.jiscdigitalmedia.ac.uk/guide/an-introduction-to-digital-audio
Sound on Sound did a series of articles, they're a bit dated now, but cover the basics well:
http://www.soundonsound.com/sos/may98/articles/digital.html

 

[The Walrus]

  Here's one:
http://www.jiscdigitalmedia.ac.uk/guide/an-introduction-to-digital-audio
Sound on Sound did a series of articles, they're a bit dated now, but cover the basics well:
http://www.soundonsound.com/sos/may98/articles/digital.html

Many thanks charleski
 
Cheers.