|
Computer Audio Asylum Music servers and other computer based digital audio technologies. |
|
|
|
Computer Audio Asylum Music servers and other computer based digital audio technologies. |
|
In Reply to: You know full-well that music is different as timing is important. posted by Norm on January 16, 2012 at 06:07:19:
Your are under two misapprehensions.
Here are the facts:
1. The timing of data transfer across the USB is irrelevant. The only timing that is relevant is the timing of the clock at the DAC chip that controls the time at which the analog signal changes. The time at which the data is transferred across the USB is no more relevant than the time at which the CDR copy you are playing happened to be burned.
2. Faster components have nothing to do with the accuracy of timing. That is controlled by the quality of the clock architecture and implementation. Lack of speed in data transfer is of relevance only if the transfer is delayed so long that it does not arrive in time. It is perfectly possible to play music transferred over a connection that operates slower than the data transfer rate if sufficient buffering is accorded and the play list has finite duration.
It is interesting to see where these myths originate and what their actual relationship to reality might be.
1. Poor interconnect designs transmit the clock across the interconnect. A prime example of this is SPDIF or AES/EBU where the clock is provided by the transport. With this poor clock architecture any noise on the clock arriving at the DAC will affect the clock that operates the DAC chip, unless heroic measures are employed to prevent these. These measures are seldom completely effective and they come at a price, namely that the local oscillator has to be variable which means it can not be the highest possible quality without further costly and heroic measures. With SPDIF/AES there is a simple solution to this, namely to run an extra wire and send a clock from the DAC back to the transport. This eliminates the entire problem if done correctly, but of course at some extra cost. With USB the situation is far worse, because the streaming equivalent (adaptive USB) has tremendous jitter because the clock is completely unrelated to the audio speed, a characteristic of the packet system. However, it is possible to run USB in such a way that the clocking in the DAC is independent. This requires sending control information from the DAC to the Transport telling it when to send packets. There are two ways of doing this and there are products that use both approaches (block mode or "asynchronous").
2. Faster components may allow for sloppier design practices, so if a mediocre engineer (or a competent engineer whom circumstances force to feed his family by shipping products on command of bean counters before their design has been completed) may get better results by working with faster components. However, if a competent engineer does a good design, it will not matter how fast the components are, assuming that they are fast enough. I have designed real time hardware/software that ran reliably where the components were less than one percent faster than the theoretical minimum and where the presence of a single extra instruction or machine cycle in the inner loop would have caused the system to fail. USB 2.0 is fast enough for all possible two channel sample rates with a huge margin to spare. The extra speed of FW 800 or USB 3.0 is irrelevant for audio applications. (These high speed interconnects were developed for bulk data movement applications such as file backup.)
Another reason why faster components may sound better is related to why they are faster. It's not the speed itself, it's the characteristics of the underlying semiconductor technology. The speed of newer computers comes in two ways: faster circuitry and more circuitry (more parallel operations). This becomes possible because the transistors get smaller and as a result can operate faster (less capacitance). This means there is less power consumption for a given task. The resulting reduction in power consumption means less electrical noise to disrupt analog signals. The benefit comes from the smaller transistors, not their speed. In many cases even more benefit can be realized by slowing down the clock speeds which permits lower voltages and even lower power consumption and hence lower noise.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- RE: You know full-well that music is different as timing is important. - Tony Lauck 07:59:14 01/16/12 (1)
- Everything you said is what I said. - Norm 20:11:20 01/16/12 (0)
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2024, Audio Asylum, all rights reserved ]
