uraki

我知道audio transfer係用isochronous，我講緊嘅比較唔係用block好似send file咁去比較。因為如果有錯，佢 ...
ntony 發表於 2020-8-4 07:06

首先唔知你有冇聽過USB線係有巨大分別.
利申我用緊條USB古河GT2 PRO就係削高頻. 一耳朵分別不用AB.

由於我親身感受過呢個分別, 所以我覺得其實係無可能係來自於DATA的完整性. 點解? 因為就當係有DATA LOST, Digital domain data 對於線來講等於encrypt左, data lost 都係100% RANDOM的, 即係得出來的聽感係會係100% RANDOM. 跟本無可能做成特定風格. 因為一條線, 無可能只撰擇丟失高頻的Digital DATA. 亦無可能選擇到保留底頻等等.

所以從我自己的經驗理解到, 做DATA CHECKING跟本唔會有意義. 區別不是源自DATA的完整性. 所以從一開始我就叫你當條線係100%SHIELDING, 永遠不會有DATA LOST的大前提下作討論. 但你仍然想討論DATA LOST.

如你想討論, 可以從實驗結果得知.
假如第一貼入面的測試, 如果真係因為DATA LOST引致有不同的FFT特性. 咁樣代表住同一條線每一次的測試出來的FFT都有好大的不同(DATA LOST100% RANDOM). 測十次有十次巨大的差異. 係絕對形成不了"特性"的.

ntony

首先唔知你有冇聽過USB線係有巨大分別.
利申我用緊條USB古河GT2 PRO就係削高頻. 一耳朵分別不用AB.

由於 ...
uraki 發表於 2020-8-5 12:40

有試過USB線係有巨大分別，不過係出起一條好低質嘅USB線上面，明顯係條線有問題。至於好貴嘅線，我無乜點去試，尤其音響名牌出嘅USB線，我去試音室聽過，唔覺有乜好。

至於比對嘅data完唔完整，我覺得呢方面係好有用。正如師兄所講，無咗音頻。好定唔好？呢堆音頻如果係干擾而造成，咁係好定唔好呢？但似乎又有批人好鍾意呢類可能係雜訊導致嘅高頻。

我唔係話師兄當時聽嗰條古河削走嘅係雜訊高頻，我只係打個比喻。當比對signal integrity係identical的話，咁我哋可以斷定條線係唔係做啱佢嘅工夫。至於其於嘢就係後面處理嘅DA同Amp嘅問題。

我哋玩digital嘅高音質，係唔係應該由呢個方向去玩呢？如果唔係，玩死自己都未必玩得好。當然，有人鍾意好似研發菜色咁乜都試一餐，睇下邊一個配搭有好嘅效果，但我只能夠話呢種玩法唔係有天份嘅人嘅玩法。

因為就當係有DATA LOST, Digital domain data 對於線來講等於encrypt左

呢part我唔係好明講乜。

data lost 都係100% RANDOM的, 即係得出來的聽感係會係100% RANDOM. 跟本無可能做成特定風格. 因為一條線, 無可能只撰擇丟失高頻的Digital DATA. 亦無可能選擇到保留底頻等等.

我哋研究digital analogue processing呢個係要視乎distortion。Loss又係另一個命題。而家講distortion，因為師兄講個signal受到EF同circuit noise產生出嘅distortion。的確，呢類distortion係比較隨機。但有好多時我哋都因為我哋了解我哋用嘅component而掌握足夠嘅spec去知道Signal Noise Ratio同埋邊個frequency受影響較大。Engineer亦係用呢堆數據決定點砌成套嘢。

至於耳聽無咗某段嘅高頻，可以係因為因為條線傳送時嘅distortion，亦可以有好多其他原因。師兄試下用返digital去諗，高音頻嘅訊號，同中低音頻嘅訊號，分別起邊？你要記清楚，USB傳送緊嘅係digital訊號，唔係analogue！佢係一堆01連同個時間資料咁去對面。究竟點先削得走高頻，起我諗到嘅情況只有係佢條線個bandwidth有問題。

uraki

你未明我點解要強調首先假設100%唔會DATA LOST. 點解唔需要理DATA的完整性.

因為聲音的改變唔係來源於DIGITAL DOMAIN(Logical data). 唔係來自於101010 有冇收錯料. 而係來自始於接受101010時的波動(電壓電流微波動)影響DAC的電路(請參考同軸果篇文, 結語甚至係USB的來源質量比同軸AES等要求更嚴苛), 甚至影響 DAC 的THD. 你現在的想法是線材干擾會令101010變000001 錯左而變聲. 想證實/否定呢樣野, 但係其實完全唔係講緊呢樣野.

你最尾果段其實就等於我所講嘅 encryption. 條線係無可能從DATA上削走特定頻段的010101. 呢D010101 logical data對於設計一條導線來說相等於加密左咁. 無可能Address到邊D打邊D. 正正係咁. 就算發生DATA LOST 引致的聽感改變. 都會係100% RANDOM, 係唔會有特色唔會有PATTERN.

咁問題來啦, 既然01010101唔會引致削/增加某些頻段. 咁我聽緊D乜? 明明有聽感的分別.
所以就有3方面. PWR, GND. 同埋 PC 信源的質量.

其實你只要認真睇同軸果篇野(相等於PC USB質量的區別), 就知我講乜...

同埋討論個TOPIC, 唔係線引入幾多干擾. 而係杜絕 PC USB的干擾.

諗下SOUND GALLERY, 幾十萬的SOUND SERVER. 號稱可以PK 同價位黑膠轉盤.
佢都係輸出USB, PWR,GND,V+,V-
我地的爛PC又係輸出一樣. 又係4條線, 有無可能通過100%電隔離, 優化光纖CONVERTER輸出USB的質量, 拉近兩者的差距呢?

ntony

因為聲音的改變唔係來源於DIGITAL DOMAIN(Logical data). 唔係來自於101010 有冇收錯料. 而係來自始於接受101010時的波動(電壓電流微波動)影響DAC的電路(請參考同軸果篇文, 結語甚至係USB的來源質量比同軸AES等要求更嚴苛), 甚至影響 DAC 的THD.
uraki 發表於 2020-8-6 20:37

呢樣就係我一直咁諗法。
不過我覺得用encryption去形容digital analogue processing唔太合適。定係你指encoding？其實Forward Fourier transform同encoding又好唔同。FFT成件事複雜好多。將digital用FFT去轉成analogue好考粒chip功力，不過呢個命題亦唔係我哋討論內容。但FFT嘅chip同circuit有幾受電流影響，個FFT做得好唔好，就係影響嘅原因。但講到尾，呢一part都係唔多關條USB線事。

用Coaxial去傳送signal同USB係完全兩回事。。。
Coaxial真係用analogue waveform去載住signal。。。直頭係analogue，佢用作digital data transmission係因為可以DFT將analogue轉成digital。佢真係本身係analogue㗎喎！
USB係用voltage加埋時間去用digital嘅概念去載signal。佢本身帶嘅signal係digital來㗎喎。睇下圖：

我唔知係唔係我又get錯你喇。

諗下SOUND GALLERY, 幾十萬的SOUND SERVER. 號稱可以PK 同價位黑膠轉盤.
佢都係輸出USB, PWR,GND,V+,V-
我地的爛PC又係輸出一樣. 又係4條線, 有無可能通過100%電隔離, 優化光纖CONVERTER輸出USB的質量, 拉近兩者的差距呢?

本身黑膠嘅frequency range同digital format就無得比。黑膠本身analogue，個frequency range好高好闊，但起碟上面嘅noise超多亦唔準確。乜嘢PK同價位黑膠，拉埋兩者用來比，都已經令我恥笑佢哋。唔識嘅比佢嚇死，識就比佢整到笑死。
我唔知佢哋用乜嘢digital format，用乜嘢freqency sampling rate同bit depth啦。就算用好高frequency同bit depth嘅digial container去裝住個media data，睇返本身digital收音錄音器材個靈敏度同容確度就已經好難同黑膠比。
https://en.wikipedia.org/wiki/Co ... d_digital_recording

我睇法係，個分別好多時未必係起條線用乜方法傳輸信號嗰度。

uraki

其實一切數碼都是源自模擬. 你上示波器一樣可以睇到WAVE.
算啦... 有D雞同鴨講. 唔解釋太多啦.
簡單D或者可以直接參考人地個結論. "所以对USB输入的质量要求极高，是明显高于同轴，光纤与AES等SPDIF协议的"
為咩佢要咁講呢. 如果覺得唔同意, 自己又有D咩野反駁的理據等等.

上面篇文個作者就係大陸果D DAC/轉盤的Designer來.

pallee

回覆 54# ntony

Some info

FFT, fast fourier transform is the math model to convert signal from time domain to frequency domain. It has nothing to do with digital signal.

PCM digital signal to analog wave can be simply done by R2R (series of stepwise RC low pass filters converting pulses to sinc waves). It was the method before MCU type DAC chip launched. Sorry that I have not studied DSD to analog (because I am an old man).

About Vinyl

Vinyl system was the best recording and playback system 50 years ago. But it has many con. Frequency range coverage is very limited.

For low frequency, needle movement is big (more surface area, wider groove size). This limits the storage time of a record. To maintain a reasonable storage time (45 minutes) by a plastic record, vinyl starts from 60 Hz usually. In the past, it was not a problem. Most old speakers started from 60 Hz.

Low frequency signal should be attenuated too (RIAA frequency response curve). Preamp and normalization circuit is required to bring the frequency curve back to a straight line.

For high frequency, plastic material is not a good storage media. Needle skipping occurs easily. For example, the best record manufacturer in Germany produces 16 K Hz records regularly. Upon anniversary, she produces 18K or 20 K Hz records but the yield is very low (< 10 %).

Mid frequency is the pro of Vinyl. Basically analog signal has unlimited bits.

S/N of Vinyl is as low as 60 dB. Signal convergency and hence sound separation are not good. Without good sound separation, sound positioning in stereo system cannot be achieved.

Nowadays, 60 - 16 K Hz audio and bad sound positioning are unacceptable.

pallee

回覆 55# uraki

It is a good video about Vinyl system pro and con.

https://www.youtube.com/watch?v=lzRvSWPZQYk

pallee

回覆 54# ntony

FYI. For long distance RCA cable is co-axial (e. g. arena equipment). It is because co-axial cable has a very low line inductance. It prevents signal group delay.

ntony

回覆  ntony

Some info

FFT, fast fourier transform is the math model to convert signal from time d ...
pallee 發表於 2020-8-7 21:16

可能係我表達能力唔好，又或者我表達錯。我知道FFT係個algorithm去轉個domain。我係講緊DSP過程用FFT同encoding（uraki話encryption!?其實我唔知係唔係指緊encoding）係好唔同。R2R起讀書時都有上堂聽過。

至於黑膠，黑膠近年興起，我見其實佢又有與時並進。因為而家科技可以刻到幾nm。至於支唱針有無科技上嘅躍升我就唔太了解。但其實黑膠都有潛力可以再發展再釋放。但佢嘅缺點就唔係咁易整走。

ntony

回覆  ntony

FYI. For long distance RCA cable is co-axial (e. g. arena equipment). It is because co ...
pallee 發表於 2020-8-7 22:15

Thank you