Audio Physics - Forum

Welcome back to our little technical feature. I do hope you have some fun reading all this stuff.

7)
The VIVACE DPP uses an enhanced input stage structure in order to increase linearity. Just to recap: the first (and very important) stage following the DAC chip is an open loop design using no feedback at all. While an experienced circuit designer should have no difficulties in designing a stage with low distortion or with low noise, it really is a challenging task to design a circuit delivering both at the same time. With the VIVACE DPP, we pushed the limits even further.


The distortion figure of the VIVACE basic is around 0,006% … 0,008% THD, showing some statistically variance due to the used JFET. The enhanced stage has a consistent THD figure of 0,001% @ 0dB. Remember: the usual stage designed around an operational amplifier (OP) like NE5534/32 will give you even better THD than this and reasonable low noise but at the expense of an unsatisfying sound quality due to the overload action in the feedback loop.

So, how does the enhanced stage sound compared to the fundamental structure? Small wonder, the difference is almost negligible due to the fact that the VIVACE DPP circuit is a tweak of the VIVACE basic circuit. While the VIVACE basic sounds sweeter, the VIVACE DPP is clearer and even more revealing. This is especially important when playing high resolution files, where masking effects of the analog stages flattens the sound.

8)
Prior to processing the S/PDIF signal in the actual DAC circuit, the two coaxial S/PDIF inputs pass through a jitter attenuator stage consisting of a PLL and highly stable quartz oscillator. If you see the circuit as black box, S/PDIF comes in on one side and leaves again as S/PDIF on the other site. While the actual sample rate and the bit depth are retained, the timing (jitter) and the stability of the flanks improved by an order of magnitude. Please note, no sample rate conversation is used, as this may alter the sound! I like to mention this fact, because most jitter reduction devices rely on sample rate conversation. As described elsewhere, a sample rate converter “converts” correct samples with the wrong timing into wrong samples (amplitude) with precise timing.

7)
VIVACE dpp 標示用上經過進一步修正的I/V轉換部份, 失真率再低幾倍, 這樣對實際聽感來說, 我們應該預期得到那方面的改善呢 ?

坦白地說, 可聞性差別近乎可以忽略不計, 除非要來實質比較不同音檔格式音質水平。
進一步去修正I/V這段所帶來的改善, 技術上把原來的0.006%-0.008% THD’互調失真’ 降低至恆常的0.001% THD, 代表著把已經十分優異的規格指標再推向另一領域。然而, 在聲音感覺上稍微有加強了一點揭示能力和音色透徹度。

須知道緊接着解碼IC輸出的就是這份I/V轉換電路, 要做到低失真率和低雜噪非常重要 ... 但兩者極難同時兼得, 對富經驗的工程人員也是一項艱難挑戰。

正如上文第4)所指出, op amp 方式的 I/V電路 (例如NE5534/32) 一般也可做出更低的失真率和可接受的雜噪水平, 可惜能夠應付(ESS9018)極高速脉冲的瞬變反應能力很有限, 容易發生過荷。我們在VIVACE dpp 所得出的成果, 相信已推向到新極限。

8)
轉換話題之前還有一個提問, 是關於同軸數據輸入, 它有特別方法去減低jitter嗎 ?

在VIVACE兩個同軸輸入這裡, S/FDIP進入解碼電路之前, 我們利用一個高度精確的PLL電路來穩定數據輸入的時間基準。
大家請特別注意 : 這PLL目的只是從新把數據時基鎖定, 不作任何升頻處理 no sample rate conversation is used 沒有增加或改變採樣頻率 , and the bit depth are retained 並保留原本數位流。

這做法好重要, 我曾經在別處描述過有關升頻的缺點, wrong samples with the wrong timing, 錯誤的時序產生錯誤的信号, 結果完全扭曲。

Let's move onto the Maestro. First may I ask a few general questions about headphone amplifier.

1)
Can we simply say a headphone amp is a line level pre-amp with substantial current output capability ?

I have an impression that the power output specification do not always give a real indication on actual driving capability, especially when driving planar-magnetic headphones. This is due to current capability, or other reason ? Please give us some insight ...


2)
What is the advantage to use balance drive for headphones ? And is it important to use full balance circuitry from input to output ?

By the way, Balance drive and bridged amplifier are not the same, am I correct ?


3)
What is the technical challenge, or objective, when you design the amp ?

Now, after a short break, I am more than happy to continue our technical encounter on audio electronics in general and on LINNENBERG AUDIO in particular.

1) A well designed headphones amplifier has the gain structure and the low noise characteristic of a preamp and ideally the output current and voltage capability of a power amplifier. By this, a headphones amplifier is neither a preamp nor a power amplifier; it is something in between. Many commercially available headphones amplifier suffer from one or the other (or both) limitation (s). To be fair, it has to be said that building an all purpose headphones amp is not an easy task, if not impossible. Headphones differ in their impedance from 4 – 1000 ohms; have sensitivities which differ as much as 30dB and so on. That’s a factor, totally uncommon in the speaker world.

You are well advised to use an amp optimized for the mayor brands and types of headphones like MAESTRO. A perfect match to your favorite headphones is most likely.

2) Historically, headphones seem to be close related to phone jacks (3.5mm, 6.3mm). These TRS (tip, ring, and sleeve) connectors do not allow for balanced drive because of the shared ground. As a matter of course, this severe weakness should be avoided. Balanced drive is the key and this makes me a passionate advocate of this technique. Furthermore, no ground load currents inside the actual amplifier corrupt signal integrity. It is kind of a no brainer, although the amplifier tends to be more costly. In fact, you need a full four channel amplifier to establish full balanced drive. If you can afford the real thing and your headphones allow for balanced connection – go for it.

MAESTRO is not just using a bridged amplifier output – it employs a fully symmetrical signal path from input to output, further increasing the high expenditure. But it is worth it. Just imagine the unnecessary stages that convert a single ended signal into a balanced signal and vice versa. If you see a “balanced” design with a stereo volume potentiometer, you will now know what is going on. Fully balanced needs a 4 times volume control!

3) General amplifier design is independent of the actual application. I design an headphones amplifier with the same care and effort as any other application. My main goals are:

- low intrinsic distortion
- low noise
- high current capabilities
- wide bandwidth and high slew rate

Ivo Linnenberg

讓我們來談談Maestro。 首先, 我想作出幾項耳擴相關的一般性提問 :

1)
我可否簡單地認為耳擴就是一個具備高電流輸出能力的前置呢 ?

根據經驗, 我認為功率規格往往不能真正預示實際推動能力, 由其當匹配平板耳機。
這是受制於電流輸出能力或是什麼其他原因呢 ? 可否給大家解析一下。

首先, 若非耳機電聲特性千種百樣, 要制作一欵來針對特定類型 無非兩個基本要求 :

第一, 須有合適的放大增益倍數和低雜噪水平, 這要求與高質前置無異 ;
第二, 要如功放般具備有穩定的電壓和電流輸出能力 ....

這兩基本看似淺白, 問題是, 既要低雜噪, 又要穩定驅控各式高低阻抗耳機靈敏度, 實非易事, 能乎合這廣泛要求, 已非俗品。

2)
請問閣下, 究竟平衡方式去推動耳機, 甚或採用全平衡電路來制作耳擴, 技術上你有何看法 ? 再者, 平衡推動balance drive 和橋式bridged amp 兩者不是同一樣的吧 ?

耳機傳統上採用 6.3mm TRS接頭 (或者微型的3.5mm), 由於兩聲道共地, 始終不是理想的接通方式, 在許可之情況下須要避免。

要完全解決共地干擾這問題, 我主張採用全平衡電路, 我廠MAESTRO就是這方式制作。 全平衡電路, 代表著從信号的xlr輸入端去到4針耳機輸出, 電路全程都保持正/負相去工作。

市面上有大多耳擴產品, 把平衡信号中途轉換成單端, 再倒相下續平衡放大, 結果自然未夠理想。

話雖如此, 要貫徹實行全平衡電路, 就須要去思考一個誤差和失真都極低的調節音量方法。 這必須是一個4路音控, 而且必要極端精良, 每聲道之正/負相信号才能夠維持在極低損差水平, 否則一切變得徒然 ! 為MAESTRO而設計的高精度音控正正反映出所投入的心思, 和對追求非凡技術成就的熱誠。

3)
關於耳擴設計, 技術上須達成怎麼樣的目標, 和有那些難度呢 ?

各式類別器材 都有幾項廣為通用的技術要點。這幾項指標, 比比皆宜。

至於耳機放大器, 在設計和構思上, 技術目標主要包括 : 固有的極低失真率, 低雜噪, 強而穩定的電流輸出能力, 頻寬廣闊 和優秀的擺動反應slew rate。