|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
138.207.213.28
Trying to catch up, and the idea of a fully blown up DAC fascinates me.
What is going on on this? I see AN is working on one, which would temp me, but no idea on price or ETA.
I looked at Lampizator, but it seems to be an afterthought on their TOTL line and their product descriptions and offerings are.. well. Byzantine? KISS?
KP
Follow Ups:
If I were going to attempt a discrete DAC, I'd really want to try out the dCS ring dac concept.
Outside of dCS the only vendor I know who had licensed it was Arcam, and that was a very short lived product.
I recall reading that dCS' "ring DAC" is a 5-bit sigma-delta converter. Not a full native resolution converter.
_
Ken Newton
It would be hard for a discrete R2R DAC to outperform an R2R that has been miniaturized on a chip in terms of technical performance. The laser trimming of resistors in a chip based DAC is a precision that would be tough to match with "macro" scale resistors. I also notice that the chip based ones could handle higher clock speeds. I think a BB PCM1704 or PCM63 could acutally handle up to 16X oversampling...this translates into speeds up to 768Khz, which is far above 192, 256 or even 512Khz that everyone is talking about.
The one argument that I have heard that might account for a better sounding discrete R2R is that you can use "better" sounding resistors. This would of course would be difficult if not impossible to determine from measurements.
However, there might be something to it. Why, when the performance was very similar, was one 20-bit DAC chip preferred over others when their performance numbers were very similar? I am thinking also of the AD1862N, and the UA D20400A DACs that were the direct competitor of the PCM63PK. The UA was very good as was the BB and for some reason the AD didn't catch on. HOwever, the AD1865N did catch on and is beloved and used to this day in some very high end designs? Was it how the resistors sounded?? Doubtful but who knows for sure. Perhaps the artifacts of their operation gave the color to their sound as each had signficant differences in their architecture.
Interestingly, the first Accuphase DACs were discrete 16 bit implementations that were optically isolated...
Don't know if they sounded good but maybe they are overlooked gems today??
Personally, I still love my PCM63PK based Monarchy M24 DAC and my UA D20400 based Kinergetics DAC (now Lampizated for the output stage). Although, I have now heard the Lampizator Atlantic with a discrete R2R and that was very nice sounding indeed.
Agree. It would be painstakingly difficult and expensive to source extremely high precision discrete resistors especially for the lower significant bits.... AND have them all track precisely over temperature. Laser trimmed on a chip would be better all around. Going fully 'discrete' is more of a novelty in this day and age, IMHO.
Spinning Schiit:
"Forget everything you know about DACs. Yggdrasil is the world's only closed-form multibit DAC, delivering 21 bits of resolution with no guessing anywhere in the digital or analog path. We've thrown out delta-sigma D/As and traditional digital filters to preserve the original samples all the way through from input to output. Let's take a closer look:
21 Bits, No Guessing: Mission-Critical D/A Technology
When doctors are trying to diagnose whether you have gas or cancer from MRI results, or when the military is trying to ensure a missile hits an ammo dump and not a nunnery next door, they don't use "24 bit" or "32 bit" delta-sigma D/A converters. Instead, they rely on precision, multibit ladder DACs , like the Analog Devices AD5791. This allows them the bit-perfect precision they need for critical applications, rather than the guesswork of a delta-sigma. We chose this same critical technology for Yggdrasil. Following these unique D/A converters are sophisticated discrete JFET buffers and summers."
I know I am dumping my BB PCM63K based DAC anytime soon...nor my Ultranalog ones either. Interestingly, The UA dacs were discrete for the first 8 bits and then used a BB 12 bit industrial ladder DAC for the lower 12 bits.
This highlights some of the issues of a discrete, or a hybrid discrete, DAC approach. That industrial IC DAC utilized for the lower 12bits undoubtedly required hand trimmed gain scaling so to linearly match the upper 4bits. Otherwise, there would be a significant discontinuity (producing distortion) every time the signal crossed the -24dBFS level.
_
Ken Newton
Edits: 11/21/16
Hi Ken,
It was the first 8 bits discrete and the last 12 IC. There was also a significant amount of supporting cicuits for deglitching etc. measurements were SOTA at that time.
Hi, morricab,Yup, the Ultra Analog DAC modules were widely regarded as state of the art at the time. However, as for every industry, the IC industry is largely driven by pressure for cost reduction. That drove the release of full resolution 20bit and 24bit DAC ICs, such as the PCM1702, AD1862 and PCM1704 over discrete or hybrid DAC modules. Cost later drove the wide scale transistion to sigma-delta DAC technology that rules today.
_
Ken Newton
Edits: 11/21/16
Have you ever heard one of the old Accuphase DACs that were discrete? First they had a discrete 16 bit one and later I think a 20bit discrete DAC. Interesting that was one of the first really high end approaches and now it comes back again...kind of like tubes...and SET in particular :)
Big J
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
I have my eyes on a DC-81L, which is discrete 20bit.
with the matching DP-90 transport. Lots of PCM63s!
I guess most examples of these will be 100V, meaning step-downs and so on, but it would be lovely to find either an 81 or a 91 and renovate/revitalise them.
Big J
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
Yeah one of those would be nice!!
I am looking seriously now at the Aries Cerat Kassandra II DAC. It is using 16 AD1865N per channel with a tube/transformer coupled analog stage. 70KG of DAC!! Enormous power supply for max stability on the tube stage.
I never had opportunity to hear one of those Accuphase DACs.
_
Ken Newton
"Spin", is the correct word for it. Most of that Schiit consumer sales argument is all too familiar from the audio industry. The performance needs of a missile guidance system, or of an MRI machine are different than those of an audio DAC. It's not simply a matter of those other applications having more stringent requirements, but that they have DIFFERENT requirements. Specifications good for one application are not necessarily good for the others.The notion of 'closed-form' anti-image filter is another example. Audiophiles are always trying to incorrectly apply intutive analog domain notions to the non-intuitive digital signal processing domain. The notion of the importance of retaining all the original digital sample values is an analog domain notion. All that's required for perfect signal reconstruction is the removal of all of the ultrasonc images. Retaining all of the orginal sample values is not required for perfect signal reconstruction.
_
Ken Newton
Edits: 11/18/16
Maybe I misread, but the close-form isn't about anti-image but about asynchronous sample rate conversion which is used for jitter rejection, not image rejection.
Still, I can say I really like the sound. :)
Best,
Erik
Schiit's spin is pretty convincing IHMO for those not wanting delta-sigma, for whatever reason.
Like audiophile cables, DACs are overcrowding the market so turn up the SPIN !
The problem is that sales spin is often complete misinformation. On the plus side, most Schiit products don't require you to take out a second mortgage on your home to purchase, so no great harm done if a Schiit product turns out to be less satisfying than the spin had led one to hope.
_
Ken Newton
A $50 wall wart switching power supply incorporating fighter jet tech!
Oh, well, I guess that no real harm can be done for $50. One thing that I will say about such ads is that they do sometimes provoke intrigue regarding the details of what exactly they are so lathered up about - LOL.
_
Ken Newton
What's yours?
One with the components are blown up' on a board, rather than embedded on a silicon chip.
KP
Maybe try the Schiit Modi Multibit to get a taste of a non-delta/sigma DAC for cheap?
I'm sold on this implementation, currently running a Bifrost MB, and soon upgrading to the Gungnir MB (Gumby).
I've had an AN 3.1x for about 15 years, so very much sold on the nos approach.
Curious as where (if anywhere) to go next, and discrete looks so interesting.
I think the Schiit's use digital filters?
KP
There is the Holo Audio Spring Dac that is getting some attention on the net.
It is designed, manufactured, and sold in China, also available from a US dealer:
Big J
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
Discrete (full native resolution/non-sigma-delta) core DACs are driven by the lack of such cores in commerical D/A chips in current production. Commercial audio D/A chip vendors have transitioned to sigma-delta conversion some years ago. There really isn't an objective performance argument in favor of full resolution converters (also commonly, and sometimes incorrectly, referred to as ladder or R-2R converters) in audio applications. The argument usually centers around subjective performance. Some DAC component vendors, such as Audio Note, have been utilizing new old stocks of full resolution converters, such as the AD1865. No doubt, dwindling stocks are driving those same vendors to develop discrete core D/A converters.
That said, producing consistently high performance discrete D/A converters are not a trivial design or manufacturing endeavor. Pretty much, all such discrete solutions will perform objectively worse than the out of production commercial IC versions. Those that do perform well are relatively costly.
_
Ken Newton
...do these discrete/NOS DACs pass high resolution data untouched or do they down sample the incoming data/music to 16/44.1?
Edits: 11/12/16
The old dac chips can handle quite high sample rates. For instance, the AD1865 and PCM63 can do 8 x 44.1kHz and the TDA1541a can do 4x or 8x depending on implementation. In their original cd players of the 80s, there were oversampling chips built in so that the 44.1kHz data was oversampled before it got to the dac chip.
If the dac is set up for NOS and the input/interface device allows, a dac with one of the old dac chips can handle high resolution, i.e. max. of 8x 44.1kHz = 352.8kHz or 8 x 48kHz = 384kHz.
The input/interface device (USB to I2S, USB to SPDIF, etc) is usually the limiting factor and that is usually governed by the conversion chip (i.e. XMOS, etc) and the speed of the clocks installed.
" ...do these discrete/NOS DACs pass high resolution data untouched or do they down sample the incoming data/music to 16/44.1?"
Non-oversampling (NOS) DACs transition the data off the media directly to analog, without any other digital conversion or processing taking place. All the filtering is analog, after the D/A transition takes place.
(Downsampling to 16/44 would degrade the signal, if the native rate off the media is of higher resolution.)
Most R2R dacs handle directly up to 24/192. They will not do DSD. My dac does not down or oversample. It also has no digital or analog filters
Alan
In have an Audio-GD Master 7 which uses the old 1704 R2R dac. Sounds better to me than and Delta/Sigma dac I have ever heard. What audio note is trying to do is create a chip with the 16 resistors not on the dac chip. They are having a very hard time doing this
Alan
From what I've read, using resistors of sufficient precision to obtain 16 bits when crafting discrete DACs is an/the(?) issue. Replicating it again and again must be even harder.
Like you, I'm quite happy with my 1704 DAC any will hang onto it for as long as possible.
Big J
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
True, and in addition to the resistor accuracy challenge is the issue of maintaining that accuracy over temperature variation. There are other challeneges as well, such as minimizing glitch energy. These are challenges both of design and of manufacture.
_
Ken Newton
This is also what I admired about the Philips TDA1541A, and whose execution in designs like the AMR CD77 is so noteworthy.
Do you know if such implementations were required/possible/executed with Burr Brown's DACs, by any chance?
DEM described.
Big J
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
Not the older Burr-Brown full resolution DACs, such the PCM63 or PCM1702/1704. However, the newer Burr-Brown/T.I. sigma-delta based PCM179x series DACs utilize a type of DEM called data-weighted-averaging (DWA) to convert the quantizer element inaccuracy (which otherwise produce distortion) in to random noise. See the below youtube video for an explanantion on the differences.
https://www.youtube.com/watch?v=BYLo2zRJ7FU&sns=em
_
Ken Newton
Big J
"... only a very few individuals understand as yet that personal salvation is a contradiction in terms."
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: