Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › 44.1kHz/16bit & 192kHz/24bit versus 2.822MHz/1bit!
New Posts  All Forums:Forum Nav:

44.1kHz/16bit & 192kHz/24bit versus 2.822MHz/1bit!

post #1 of 37
Thread Starter 
I've been playing around with my ModWright LLC Pioneer Elite DV-59AVi settings for the past several days. It uses three Burr Brown 1738E DACs which feed into two AD8620 operational amplifiers. I have the option of using the proprietary Pioneer HiBit and Legato PRO features which convert 16-20bit word lengths to 24bit and 44.1-48kHz sampling frequencies up to 352.8kHz (after which the Shannon-Nyquist thereom takes into effect).

With regard to Red Book CDs only, I have conclusively found that native 44.1kHz/16bit sounds more natural with greater precision in terms extracting low-level information, clarity, and immediacy. The music sounds so life-like and natural. In my humble opinion, it exceeds the CARY Audio Design 303/200 and is on par with the Meridian G08 that I have auditioned it against at my local Hi-Fi shops. Even the owners of the shop had to agree with me and they asked me more about my particular source component. What strikes me is that nothing in particular stands out: the whole musical communication is so transparant, detailed, nuanced, and alive that I am still dumbstruck by its ability to seduce me into the musical experience itself without effort.

When I do turn on the HiBit and Legato PRO features, the sound changes radically. To my ears, I can not tell the difference between 192kHz/24bit Red Book CD and my AIX Records 192kHz/24bit DVD-Audio discs that were recorded and mastered at that native resolution. Music becomes much more holographic and three dimensional with a greater sense of spaciousness and fewer aliasing jaggies throughout the audible frequency spectrum allowed by my Ue-10 PRO. Trebles are far more extended and airy with a greater capture of ambience cues such as the recording venue's dimensions. Midrange extends futher in terms of front to back imaging depth and there are fewer compression folds so to speak. Bass extension goes down further with greater low level retrieval and there is a far lower noise floor inherent in the music.

It is my opinion that either 44.1kHz/16bit or 192kHz/24bit sounds far more precise and musical than 2.822MHz/1bit DSD. Transients do not suffer from gloss, dynamic shading and micro-dynamics come through more freely, and there is a nearly inaudible noise floor. To my ears, DSD can not seem to overcome these limitations in sound due to its architecture. For the record, my ModWright LLC Pioneer Elite DV-59AVi does not perform DSD->PCM conversion nor does it employ steep filters whatsoever. It uses purely discrete signal lanes and the triple TI Burr Brown 1738E DAC can discriminate and process the SONY DSD generation II DAC signal independently of PCM.

In conclusion, I can see where preferences become very important in determining whether 44.1kHz/16bit or 192kHz/24bit is "superior or inferior," but there leaves little doubt in my mind that 2.822MHz/1bit DSD sounds "incorrect" for the above stated reasons.
post #2 of 37
Yep. You hit upon one of my conclusions on DSD purely from the math of it. It is absolutely unable to track quick transients. Also implying it's inability to faithfully reproduce larger amplitude high-frequency content.
post #3 of 37
well Welly, things aint that simple as you might think.. there is nothing like recorded in native 24/192 nor 16/44.1, they are recorded much like DSD in fact, it's just for example 5bit/5.6448MHz instead of 1bit/2.8224MHz.. the 192 or 44.1 is just output from decimator after digital low pass filtration.. and similarily those DACs you have are not working at 16/44.1 nor 24/192 but also multibit delta sigma structures running ~5MHz using ~5bit quantizing.. input being 16/44 or 24/192 or DSD doesn't matter..

and why do you think DSD cannot reproduce transients? it is the thing DSD does and does it better than common PCM samplerates.. the fact is it looses precision as the frequency rises, however the transients are reproduced much closer to reality than common PCM rates.. and the beauty of DSD is it's almost infinite resolution at frequencies our ears are most sensitive to and where 90% of the musical signal lies..
post #4 of 37
Thread Starter 
Before I left for work, I wanted to add this comment or opinion with regard to 44.1kHz/16bit oversampled to 176.4kHz/24bit: it seems to add more of a voluptuous body to each note and warmth to the sound, but it comes at a price of a slightly higher noise floor. At least, that is how the Pioneer HiBit and Legato PRO is handling it. Vocals take on an added lush warmth and velvety smoothness, but tends to sound a bit forward in the soundstage. Soundstage width seems unchanged, but imaging height and depth seem taller and deeper. In general, it sounds like a good feature to pair with the Ray Samuels Emmeline HR-2 house sound. Bass attack seems deeper with added energy and but shorter decay. Midrange is a bit forward and grainy. I'll have to keep trying it on and off, before I render a final verdict. So far, so good!
post #5 of 37
I'm not convinced about upsampling/oversampling. I think 16/44.1 is just fine. Now if only the assclown recording engineers would just properly master the CDs...
post #6 of 37
Thread Starter 
To a certain degree, I agree with you, but I'm still experimenting not necessarily with oversampling as per my source component's capabilities: I am trying to get a handle on how well oversampling contributes to total system synergy and harmony along with balance. So far, it's good with the standard Legato PRO filter which doesn't use a high pass filter. HiBit is quite nice in terms of increasing low-level information extraction, micro-dynamics, and dyanmic shading is improved in combination with the oversampling frequency increase. These are just notes you see...
post #7 of 37
you cannot turn oversampling off, you're just switching between different oversampling filters - DAC's internal and Pioneer's proprietary..
post #8 of 37
Thread Starter 
Originally Posted by Glassman
you cannot turn oversampling off, you're just switching between different oversampling filters - DAC's internal and Pioneer's proprietary..
Ummm...I am not certain that you are correct, but I am of an open mind.

Pioneer's HiBit and Legato PRO are completely defeatable and the triple Burr Brown 1738E DACs do not invoke oversampling by default unlike the Denon's Burr Brown 1792 DACs which do invoke oversampling and upconversion by default (which are not defeatable through either an option within the proprietary Denon OS or through the company's firmware updates). I am pretty certain of this, but I can allow that I may be incorrect.
post #9 of 37
you should check the theory of operation of modern delta sigma DACs.. they all oversample, because they have to.. the signal that goes into the 6MHz sigma delta modulator has to be clean from noise at least upto a couple of hundred kHz in order for the modulator to work right.. usually the internal filters oversample 8x so from 44.1 you get 352.8, that means noise free band up to some 352.8-22.05 ~ 330kHz and thats okay for the delta sigma modulator, it eats this signal and produce usually 5bit @ 128x44.1=5.6448MHz signal that is presented at the analog outputs..

any oversampling, upsampling or resampling done before the data enters DAC are basicaly overriding the DAC's internal filters.. even in case of upsampling to 24/192 the DAC's filters do oversample 2x to 384kHz before it enters the modulator..

also you should know that DSD = 1bit/64x44.1 delta sigma modulated signal
modern ADCs and DACs work natively with 5bit/128x44.1 sigma delta modulators..

the words UPsampling UPconversion are quite misleading and marketing instead of technical terms.. increasing the samplerate is unavoidable, it's not a special feature, snake oil kind of thing but absolute must.. as with everything this can be done various ways having impact on the resulting sound..
post #10 of 37
Wow Glassman, you are right.
I did some Google search and realized 5-bit is the state-of-the-art right now.
I wonder why no one ever told me before.
But the sigma-delta architecture is way complex and far beyond my apprehension.
The datasheet of BB 1738 does not say how many bit is the actual converter. But I assume it is multi-bit. It has built-in decoding interface for DSD signal. I guess it still does not decode DSD with 1-bit sigma delta DAC. But I could be wrong about this.
Welly, don't get too excited about redbook. It is invented by Phillips and why in the world would they invent something worse twenty years later?
post #11 of 37
Welly, playing devil's advocate, I think it could be argued that converting from PCM to DSD must require A LOT more processing than upsampling PCM to high-rez PCM. It could be that this process introduces so many additional artifacts that PCM converted to DSD is compromised to the point it can't ever sound very good. Or it could be that 16/44.1 upsampled to DSD just doesn't sound very good where a 24/96 source signal converted to DSD might sound much better. When they make SACDs from PCM digital masters, they are typically making them from 24/96 masters, not 16/44.1. I know that of the SACDs I have that are digital recordings (and therefore PCM), they almost always sound superior to the original 16/44.1 CD. So, I think it *is* possible to convert from a "hi-rez" PCM source to DSD and still end up with an SACD that spanks 16/44.1 in the butt.

You're also relying on the one chipset in your source, it may be that that particular set is just not that adept at DSD/SACD or at the conversion process. I think it would be more interesting to take an analog tape master, make a DSD version, a 16/44.1 version, and then a DVD-Audio version so there's common source material for each version, but then again, the mastering engineer would likely be using different equipment with different signal paths to create each version, so that's not reliable either. IMO, it's very very hard to make proper comparisons of the various formats. Cheers.
post #12 of 37
DSD is basicaly PCM just with slightly different properities.. delta sigma equals noise shaping, you can hear about noise shaping related to PCM..

you cannot do anything with DSD, not even stupid volume control, without increasing bit depth, low pass filtering and running through the sigma delta modulator again.. and that precisely is what happens to PCM when converting to DSD - low pass filtering, then running through delta sigma modulator.. no snake oil again..

PCM always is a product of decimating DSD-like sigma delta signal.. these two areas overlaps in praxis..

there is very few SACDs recorded at 1bit/64Fs natively and even much less if any D/A converters working at that frequency natively.. the main difference between 24/96 and DSD is that DSD has theoreticaly incredible resolution at lower frequencies, however at higher frequencies it can be easily worse that 16bit PCM, also DSD has greater bandwidth, because of the fact that it does not have to be low pass filtered with as steep filters as 96kHz PCM.. 24bit PCM offers theoreticaly 144dB precision over the whole frequency spectrum, the only possible >problem< is the need to attenuate everything under at least 120dB at half the samplerate, that means using steeper filters with more ringing.. but the move from 44.1kHz where you had to attenuate by at least 96dB (16bit) in just 2.05kHz (44.1/2-20) to nowadays 96kHz, where you need to attenuate by 120-144dB (20-24bit), but you have a wide band of 28kHz (96/2-20) to do that, is probably almost ultimate cure..

I suggest everyone checking this paper:
post #13 of 37
Interesting thoughts, Glassman! So are you really sure there's no pure 16/44.1 or 24/192 PCM DACs around? What a terrible world with all this complicated sampling-rate conversion thing, which certainly isn't something the accuracy of the original signal will benefit from! If that's really the case I'm somehow going to lose my respect for digital music reproduction.

Originally Posted by Glassman
You cannot do anything with DSD, not even stupid volume control, without increasing bit depth, low pass filtering and running through the sigma delta modulator again...
How about «pure DSD» indications? Untrue, or is it possible to record an SACD in DSD without any processing?

...the main difference between 24/96 and DSD is that DSD has theoreticaly incredible resolution at lower frequencies, however at higher frequencies it can be easily worse that 16 bit PCM...
I wouldn't call the equivalent of barely more than 17 bit at 20 Hz (according to my calculation) «incredible resolution». Whereas 7.46 bit at 16 kHz is simply lousy indeed. Or is my calculation wrong?

post #14 of 37
Originally Posted by JaZZ
I wouldn't call the equivalent of barely more than 17 bit at 20 Hz (according to my calculation) «incredible resolution». Whereas 7.46 bit at 16 kHz is simply lousy indeed. Or is my calculation wrong?
I am not a technical expert in these things.
But I think PCM and DSD are just different math forms to store waveform information. Since DSD has 4x information content than 44/16 based in bits, I would surprised if DSD does not encode more information at least in some frequency band, at least mathematically. If DSD encoding has less information and takes up more space, I am sure all the DSP experts in Philips and Sony would never put it to the market in the first place.
post #15 of 37
Jazz, there certainly are DACs that can work at the stated PCM samplerates/bit depths.. they are called R-2R DACs, because they use 16-24 internaly calibrated switches each enabling half the current than the previous.. these are the original DACs that were used when digital was being put into praxis.. their precision is limited, their cost is very high because the need for precisely trimming the steps.. TDA154x, AD1865, PCM170x are a few examples of such DACs, however they still were usually used paired with oversampling filters, although they can work without them as well..
the vast majority of DACs used todays (since 2nd half of 90's, that is) are based on sigma delta modulators running at 128Fs, Fs being 44.1kHz.. originaly they were single bit, however nowadays they are mostly multibit - but just 5bits.. those five bits are interpreted differently than in R-2R DACs.. there is 2^5 equal switches being switched pseudorandomly.. it's rather easy to precisely calibrate these switches given they all need to be same and the randomisation completely decolerates possible errors from the audio signal.. they're much cheaper to manufacture and provide better measured performance..

yes, there certainly is a way how to record directly in DSD.. you have to have the material prepared/mixed before the actual A/D conversion though.. remastering of analog tapes is a good candidate for recording directly in DSD, that is 1bit/64Fs sigma delta modulator ADC.. current ADCs are running 5bit/128Fs modulators, 1bit/64Fs is less demanding.. the question is wheter there is any DAC that would run at 1bit/64Fs, I don't know any..

well in fact on low frequencies, good 5th order sigma delta modulator achieves greater than 160dB precision I believe.. theoreticaly, that is.. in praxis we are limited to approx. 125dB at max at current state of technology.. the high frequency precision is limited though, but it certainly ain't 7bit.. your calculations are wrong.. I suggest taking a look into PCM4202 datasheet - true single bit A/D converter with 64Fs/128Fs DSD output capabilities..
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Dedicated Source Components
Head-Fi.org › Forums › Equipment Forums › Dedicated Source Components › 44.1kHz/16bit & 192kHz/24bit versus 2.822MHz/1bit!