Any info about what's inside Naim DAC?
Posted by: goldfinch on 15 June 2009
Hi chaps,
It has been told Naim new DAC will be based on the developments involved in the Bentley proyect, could anyone elaborate more on this?
Particulary, how Naim DAC will approach jitter issue?
Since there isn't any Naim white paper (this would be great to see one day), I can't get any info about it.
Jose
It has been told Naim new DAC will be based on the developments involved in the Bentley proyect, could anyone elaborate more on this?
Particulary, how Naim DAC will approach jitter issue?
Since there isn't any Naim white paper (this would be great to see one day), I can't get any info about it.
Jose
Posted on: 03 July 2009 by Eric Barry
On 96 v192 see the third post here:
Eric I have no problem if you borrow and paraphrase from the explanation given on the site you linked to, however, to avoid this turning into a debate about another DAC company, I thought the link best not be posted. Thanks, Richard
Eric I have no problem if you borrow and paraphrase from the explanation given on the site you linked to, however, to avoid this turning into a debate about another DAC company, I thought the link best not be posted. Thanks, Richard
Posted on: 03 July 2009 by Eric Barry
From Dan L****:
"Hello all,
I am new to this forum.
I want to buy a high end converter (I am between 3 brands) and I was wondering whether it is safe to assume that a converter running at 192 KHz can handle the 96 KHz sampling rate better than the 96 KHz unit? Dan said that the bigger the sampling rate the more chances of error there are, so I would like to know what these chances are and whether they are related to the clock per se or to the AD converter. Or, in other words if I get a 192 KHz unit and run it at 96 Khz are the chances of error less than if I had the 96 Khz unit and run it at 96 KHz?
Same question is for lower sampling rates: Dan's Gold AD is 96 KHz, does that mean that it's more accurate at 48 KHz?
Regards,
Yannis"
An AD converter that can convert at 192KHz, can be made to operate at 1/2 the rate (96KHz) by a process called "decimation by a factor of 2". In fact most converters these days, the converter front end (the modulator) operates at very high rates, such as 64, 128, 256, 512 or even 1024 times faster then the sample rate. Bringing it down to a final rate (the sample rate) of 48KHz, or 96Khz or 192KHz is done by that decimation process.
But a 100MHz front end modulator can not yield good audio results. You can keep decimating down, but even a theoretically perfect decimation will not save you from high distortions and noise. The limitations are already there in the modulator.
The designer faces a tradeoff regarding the basic converter architecture. The question is: shell I optimize the design for 96 KHz? Or for 192 KHz? This is a conceptual and theoretical decision, before you even begin the hardware design.
Take a sigma delta architecture (as in most converters today), and you have some basic parameters to play with: Modulator front end speed, modulator order (how many bits in the front end), filter order (the feedback loop). So for any given set of such parameters (fixed resources), one can trade off the final desired bandwidth against accuracy (distortions and noise). In other words, “holding all things equal”, one can trade off speed against accuracy. This is a “block diagram” tradeoff right from the get go.
Say you want to dig a hole for a swimming pool. You have the resources to dig 100 cubic feet. One can dig a 1 square foot area to 100 feet of depth. Or on can dig a 100 square feet area but only 1 foot deep. Or one can dig a 20 foot area at 5 feet depth… That is an analogy for the modulator design. One needs to plan for an optimized outcome, before you even started digging. You want enough area to swim, but it also needs to be deep enough. In the case of audio, one wants to have enough bandwidth, which is dictated by how much the ear can hear, but the more bandwidth is not free – it reduces accuracy. It is a tradeoff! If you believe that your ear can hear 96KHz then you may opt for 192KHz. If you think that the ear can hear up to 48KHz, then 96KHz may be of value. The fact is – 96KHz sampling is already too high.
Not less important is the real hardware design. A circuit that operates faster, will yield larger error. Say you need to charge a capacitor, if you have more time, you can reach better precision. If you apply a step into an opamp, it will settle to a more accurate value when you take more time. And so on… Note that charging caps and settling opamp IS at the hart of convertors…
I talked about AD’s but DA’s are similar. They too have a modulator, and the tradeoff is also there right from the start, as well as the limitations due to circuit limitations due to speed – accuracy tradeoff (charging caps, settling opamps and much more).
A 192KHz AD and DA are device that can capture or reproduce signals that occupy up to 96KHz signals. But you simply do not need that for audio. That is a ridicules bandwidth in a world of mics and speakers (mostly 20KHz devices). It is a bad tradeoff.
A converter that allows you to process signals you do not need such as over 50KHz has a BUILT IN traded off, and one can not fixed the accuracy issues after the fact, by decimation from 192KHz to 96KHz. The accuracy is already reduced before the decimation.
A converter that does a good job at 192KHz will likely do as good job at 96KHz (not better). But given the same “fixed resources” a designer can get better results when aiming at 96KHz. Of course a converter optimized for 96KHz can not accommodate signals above 48KHz. Is that a good trade-off? My dog does not hear 48KHz…
"Hello all,
I am new to this forum.
I want to buy a high end converter (I am between 3 brands) and I was wondering whether it is safe to assume that a converter running at 192 KHz can handle the 96 KHz sampling rate better than the 96 KHz unit? Dan said that the bigger the sampling rate the more chances of error there are, so I would like to know what these chances are and whether they are related to the clock per se or to the AD converter. Or, in other words if I get a 192 KHz unit and run it at 96 Khz are the chances of error less than if I had the 96 Khz unit and run it at 96 KHz?
Same question is for lower sampling rates: Dan's Gold AD is 96 KHz, does that mean that it's more accurate at 48 KHz?
Regards,
Yannis"
An AD converter that can convert at 192KHz, can be made to operate at 1/2 the rate (96KHz) by a process called "decimation by a factor of 2". In fact most converters these days, the converter front end (the modulator) operates at very high rates, such as 64, 128, 256, 512 or even 1024 times faster then the sample rate. Bringing it down to a final rate (the sample rate) of 48KHz, or 96Khz or 192KHz is done by that decimation process.
But a 100MHz front end modulator can not yield good audio results. You can keep decimating down, but even a theoretically perfect decimation will not save you from high distortions and noise. The limitations are already there in the modulator.
The designer faces a tradeoff regarding the basic converter architecture. The question is: shell I optimize the design for 96 KHz? Or for 192 KHz? This is a conceptual and theoretical decision, before you even begin the hardware design.
Take a sigma delta architecture (as in most converters today), and you have some basic parameters to play with: Modulator front end speed, modulator order (how many bits in the front end), filter order (the feedback loop). So for any given set of such parameters (fixed resources), one can trade off the final desired bandwidth against accuracy (distortions and noise). In other words, “holding all things equal”, one can trade off speed against accuracy. This is a “block diagram” tradeoff right from the get go.
Say you want to dig a hole for a swimming pool. You have the resources to dig 100 cubic feet. One can dig a 1 square foot area to 100 feet of depth. Or on can dig a 100 square feet area but only 1 foot deep. Or one can dig a 20 foot area at 5 feet depth… That is an analogy for the modulator design. One needs to plan for an optimized outcome, before you even started digging. You want enough area to swim, but it also needs to be deep enough. In the case of audio, one wants to have enough bandwidth, which is dictated by how much the ear can hear, but the more bandwidth is not free – it reduces accuracy. It is a tradeoff! If you believe that your ear can hear 96KHz then you may opt for 192KHz. If you think that the ear can hear up to 48KHz, then 96KHz may be of value. The fact is – 96KHz sampling is already too high.
Not less important is the real hardware design. A circuit that operates faster, will yield larger error. Say you need to charge a capacitor, if you have more time, you can reach better precision. If you apply a step into an opamp, it will settle to a more accurate value when you take more time. And so on… Note that charging caps and settling opamp IS at the hart of convertors…
I talked about AD’s but DA’s are similar. They too have a modulator, and the tradeoff is also there right from the start, as well as the limitations due to circuit limitations due to speed – accuracy tradeoff (charging caps, settling opamps and much more).
A 192KHz AD and DA are device that can capture or reproduce signals that occupy up to 96KHz signals. But you simply do not need that for audio. That is a ridicules bandwidth in a world of mics and speakers (mostly 20KHz devices). It is a bad tradeoff.
A converter that allows you to process signals you do not need such as over 50KHz has a BUILT IN traded off, and one can not fixed the accuracy issues after the fact, by decimation from 192KHz to 96KHz. The accuracy is already reduced before the decimation.
A converter that does a good job at 192KHz will likely do as good job at 96KHz (not better). But given the same “fixed resources” a designer can get better results when aiming at 96KHz. Of course a converter optimized for 96KHz can not accommodate signals above 48KHz. Is that a good trade-off? My dog does not hear 48KHz…
Posted on: 03 July 2009 by David Dever
Not sure I agree, as there are some exceptional tweeters out there (Dynaudio Esotar, Morel)-which can easily show up differences in sample rate.
Sennheiser MKH series RF microphones can handle signals well above 40kHz, when paired with the right mic preamp electronics-great for cymbals, even with the requisite analog distortions that must still be considered "information", even if flawed.
Interestingly, none of these "designers" ever seem to discuss bandwidth in terms of octaves-a cursory examination would reveal harmonic spectra that easily exceeds 60kHZ-even on 32' labial organ pipes.
Sennheiser MKH series RF microphones can handle signals well above 40kHz, when paired with the right mic preamp electronics-great for cymbals, even with the requisite analog distortions that must still be considered "information", even if flawed.
Interestingly, none of these "designers" ever seem to discuss bandwidth in terms of octaves-a cursory examination would reveal harmonic spectra that easily exceeds 60kHZ-even on 32' labial organ pipes.
Posted on: 03 July 2009 by js
Same here. I don't think the 'because it can't be made as well' is a good reason and I've heard positive differences up to 96k but of course I can't be sure that this is only due to sample rate and not the specific kit. I really didn't expect a dif between 88 and 96 but it may have been DAC optimization as I don't know the circuit. 96 has been the sweet spot so far and like I said, 192 hasn't shown me benefit but I don't think it's a very difficult freq to deal with in current tech. Perhaps a given price point or unavailability of a certain type of device may make it an issue or not the best solution but I don't see a tech problem. Look at the frequency these DACs oversample at and it makes 192k look pretty insignificant.
I do think that perhaps Dan is right and 192 is just making things more complicated for no benefit which is never a good thing but I just don't think that there's a technology wall causing it.
I do think that perhaps Dan is right and 192 is just making things more complicated for no benefit which is never a good thing but I just don't think that there's a technology wall causing it.
Posted on: 03 July 2009 by fixedwheel
quote:Originally posted by GFFJ:
I never liked LP as a medium, and would never return to it now, as to get it to sound nearly so fine as decent digital transfers costs many thousands.
Plus you have so much wasted space housing all that deadful card-board and vinyl, with the associated artwork of so often dubious quality compared to the music ...
OOOhhh!!! You little pyromaniac!!
John
Posted on: 03 July 2009 by Guido Fawkes
I'll just hang another Roger Dean poster on the wall.quote:Plus you have so much wasted space housing all that deadful card-board and vinyl, with the associated artwork of so often dubious quality compared to the music ...
Posted on: 05 July 2009 by james n
I'm hoping to nip out of work for a few hours on Wednesday to go to the Summer Sounds at my local dealer. Looking forward to finding out what this DAC is all about 
James
James
Posted on: 07 July 2009 by goldfinch
Excellent James, let us know your findings!
Posted on: 08 July 2009 by james n
Will do 