|
|
楼主 |
发表于 2016-2-21 22:12:55
|
显示全部楼层
本帖最后由 yamuling 于 2016-3-7 19:11 编辑
Chord DAVE设计师Rob Watts论坛言论节选：
关于电源和噪音：
What people don't understand is that a DAC internally is a RF noise hell (and moreover is signal correlated too) - and this is several orders of magnitude more noisy than the noise you get from a properly designed PSU. That is where the effort goes into - the internal nodes in the DAC, where people don't see, appreciate or understand the difficulties.
大伙不了解，解码内部才是射频噪音的垃圾场——比从外部电源引入的噪音多几个数量级。这就是为什么我们在不断努力去改善解码的内部节点。而大伙看不到，也不能欣赏和理解这其中的难度。
关于DAVE的FPGA核心：
Dave has a single FPGA, with about ten times the capacity of Hugo's. I don't know where 4 came from, that would have been the DAC 64.
It controls everything, including the display, which means that apart from the USB decoder (which is not on Dave's ground as its galvanically isolated), there are no other digital processing devices. This way I get the lowest possible noise, with everything running from the crystal oscillator. That said, it was a pain creating my own font generator to drive the display.
DAVE内部只有一块FPGA，性能为HUGO的十倍。我不知道四块是谁说的，那是DAC64时代的设计。
FPGA控制一切，包括显示屏，这意味着除了USB接收（它不与DAVE共地，因为做了电气隔离），机器内没有任何其他的数字处理元件。这是为什么我做到了最低的噪音可能性，一切根据时钟运转。换句话说，为自己的显示屏设计字体合成器真心痛苦。
关于开关电源、线性电源：
I am sure there is some compromise in using a linear power supply - in fact I know there are compromises.
Perhaps I should explain.
When developing Hugo in 2012 I had to use switching regulators (battery life issues) rather than linear ones, as we all know linear regulators sound better don't they?
Funny thing was, in testing and listening, the switching regulators sounded a lot smoother and darker with much better sound-stage depth.
Also the DAC measured a lot better too, particularly with very small signals. I now know exactly why this happened, and why they sound better and exactly why they measure better within a DAC.
Moral of the tale? Don't make assumptions based on what you think is happening, do the work, do the measuring, and do the rigorous listening tests. Then you find out that things are not as simple as you think, and you find out that previous certainties were mere assumptions.
我可以确信的是，如果换上线性电源，那会是一种妥协——实际上，我知道有哪些妥协。
我来解释下。
12年我设计HUGO的时候，我不得不选用开关稳压（电池持续时间的原因）代替线性稳压，我们都知道线性稳压声音更好不是？
有意思的是，在测试和试听环节，我们发现开关稳压听起来更顺更黑，连带音场深度提升非常明显。
同时，解码测出来的指标也好多了，特别针对细微信号。我现在掌握了到底是什么环节，为什么解码内开关稳压声音更好、测试结果更好了。
寓意？别根据你认为会发生什么去臆想，去做，去测试，谨慎地试听实践。你会发现事情不像你想象的那么简单，然后发现很多先前确信的事，其实只是臆想。
关于耳机模式里的Crossfeed设定：
0 is no crossfeed; 1 is min, 3 is max.
As to cross-feed - I prefer 3, its closer to loudspeaker perspective.
关于耳放crossfeed补偿模式：0为无补偿，1最小，3最大。
我倾向3，最近接听音箱的感觉。
Now the problem is definitely not jitter from the source - my DAC's can tolerate 2uS of jitter and it will have zero difference to the measurements - also the USB is isochronous asynchronous so the timing comes from the DAC clock, so source jitter is not a problem.
So I looked into the issue of different SQ with sources and found two sources of error:
1. RF noise.
2. Correlated current noise.
So the solution to the above problems is galvanic isolation. This means that RF noise from the source can't get into Dave, and small correlated currents can't get in too. And this approach gave two benefits - much smoother sound quality, and a deeper soundstage.
Now with Dave I can no longer hear which source is connected, but before without the galvanic isolation it was easy to hear. Additionally, optical sounds almost identical to the USB, as optical is clearly galvanically isolated too.
关于USB链接不同电脑或数播音质不同：
我的解码可以容忍高达2uS的jitter，而测试结果并无不同。异步USB意味着时钟基于解码，所以音源Jitter不是个问题。
我觉得不同转盘音质不同，主要源于音源产生的两类干扰：射频噪音和相关电流噪音。
解决方式是做电气隔离，之前没有隔离的机器（Hugo等）能轻易听出音源差异。另外，光纤与USB的音质几乎等同，也是因为光纤显然能做到电隔离。（他曾说平衡数字线路是造成音质问题的原因之一，所以DAVE多达四组BNC数字输入，两组光纤输入，而只有一个AES口）。
网友问：为什么其他解码耳放一体机需要额外的耳放部分
为什么其他解码的模拟部分如此复杂：
Just a thought. Why do other DAC/headphone amps have amp sections when Hugo/Mojo get by without one, and many including Chord say it is more transparent? Have Chord got the patent for ampless amps :-)
A：Because they can't using chip based DAC's. Chip DAC's have two current outputs. So you need two I to V converters (amps) then a differential to single ended amp, then a headphone buffer to deliver the current. You also need a lot of analogue filtering wrapped around these amps. So why are normal DAC's so complex in the analogue domain? Two reasons:
因为（这种设计）不能使用基于芯片的解码。解码芯片有两路电流输出。所以需要两组IV转换（放大器），然后是差分转单端的放大器，然后是耳机缓冲。你还需要一堆模拟滤波围绕放大器。（所以增加了干扰和底噪）因此为啥普通解码模拟部分都那么复杂？有两种原因：
1. Silicon DAC's are horribly noisy, as the substrate and grounds are bouncing around due to switching activity. So to counter this, it is done differentially, which means the ground noise is cancelled. It also hides the problems of the reference circuitry, which can't be made with low enough impedance on silicon. This translates to more distortion, and crucially noise floor modulation.
1，解码硅片内部吵得很，因为基板和地会对开关活动造成不断反射。为了对抗这些，它会用到差分线路，因此地的噪音可以被消除。它也隐匿了参考线路中的问题，即是硅片上做不到足够低的内阻。这编译出更多失真、还有关键性的底噪调制。
2. Delta sigma converters run at low rates - best is at 12 MHz - this means that there is a lot of noise that must be aggressively filtered out in the analogue section. This also applies with R2R DAC's too as these have even worse problems due to the very slow switching speed.
2，Delta sigma转换器的速率很低——最高也不过12 MHz——这意味着许多噪音必须靠模拟环节的强力滤波来处理。这也适用于R2R解码，因为切换速率如此之慢，产生的问题只会更严重。
So to run with a single amp section you need the DAC to be single ended and to run the noise shapers at much higher rates to reduce your filtering requirements. Because the analogue section with Mojo is discrete, I can use extremely low impedance and low noise reference supplies - something that is impossible on silicon. This has the other benefit of eliminating noise floor modulation (actually there is a lot more to it than this as there are countless other sources of noise floor modulation in a DAC). To make the filtering easier, the pulse array noise shapers run at 104MHz - over an order of magnitude faster than normal. There are other benefits to running the noise shapers at 104MHz, principally the resolving power of the noise shaper. Now soundstage depth is determined by how accurately small signals are reproduced. The problem with noise shaping is that small signals get lost - any signal below the noise shaper noise floor is lost information. But by running the noise shaper at much faster rates you solve this problem too - indeed Mojo noise shapers exceed 200dB THD and noise digital performance - that's a thousand times more resolving power than high end DAC's.
要同单端放大部分一起工作，DAC必须是单端的，且整形滤波必须处于非常高的频率以减少滤波需求。因为Mojo内部的模拟部分是分立的，我可以使用极端低内阻和低噪音基准的产品——有些硅片完全做不到。另一个好处就是不再需要底噪调制（实际上需要做的远不止这些，解码内还有其他数不尽的底噪调制源）。为了让滤波更简单，脉波阵列噪音整形运行于104MHz——一个数量级地超越一般做法。运行在如此高的频率还有其他方面好处，尤其是噪音整形的分辨率。如今音场深度取决于小信号的还原准确程度。噪音整形产生的问题，就是小信号的损失——信号小于噪音整形本底噪音的就是损失的信息量。但通过把噪音整形运行在极高的速率，你也就解决了问题——实际上，Mojo的噪音整形能超越200dB THD，噪音数字水平——分辨率上成百上千倍地超越那些high end解码。
If I get time today I hope to publish noise floor modulation measurements showing Mojo has zero measured noise floor modulation. This level of performance does not happen on any other non pulse array DAC's at any price, and its the primary reason why Mojo sounds so smooth and musical.
如果我今天有时间，我希望能发一些底噪调制的测试数据来证明Mojo的零底噪调制。这种水平也没有任何价位的非脉波阵列解码可以做到。这也是为什么和弦可以做到如此平顺乐感。
其他帖子还有很多技术讨论，就不一一翻译了，有兴趣可以去原帖看。
解答网友提问，最佳解码耳放一体机大战三选一：南瓜HD，法国解码和DAVE？
So now we need a battle between the 3 pretenders for "Best DAC with a Headphone Out" : NAGRA HD DAC, TotalDAC D1-Dual (or Mono) & DAVE. Who will win ?
Rob:
That depends on how one defines best - if best is defined as transparency and musicality then Dave will win easily. I am prepared to bet a brand new Dave against a bottle of Bollinger that Dave will win.
在于你定义最好的标准是什么——如果看重透明度和音乐性，DAVE轻松胜出。
另备注一下，平衡口因为加了短路保护，不能用来直连耳机。而四个BNC高速输出口，未来CHORD会推出配套的极简数字后级（用于高灵敏度音箱和高级耳机），已达到消除后级个性，系统音色即DAVE音色的目标。
|
|
[复制链接]
发表于 2016-2-21 22:12:09
提升卡
变色卡
