以前的設(shè)計(jì)方案闡明了一個(gè)通常困擾模擬設(shè)計(jì)者的處理方法：討厭的接地回路（參考文獻(xiàn)1）。那個(gè)設(shè)計(jì)方案描述了一個(gè)簡(jiǎn)單而有效的多通道電路。但它是不對(duì)稱(chēng)的CMV（共模電壓）方法，只適用于總線接收端。因此，它只應(yīng)用于信號(hào)輸入，而對(duì)輸出無(wú)效。然而，CMV如果由純交流噪聲組成，一個(gè)不同的CMV修正方案——有源傳導(dǎo)消除——雙向工作，且因此抵消輸入和輸出信號(hào)上的CMV誤差成分。
工程師常年使用無(wú)源CMV感應(yīng)消除（圖1）。有時(shí)被稱(chēng)作“humbucker變壓器”，因?yàn)槟芰扛删€的60Hz“嗡嗡聲”通常為CMV成分，CMV感應(yīng)由信號(hào)源（1）和目標(biāo)（2）之間的初級(jí)線圈與地連接組成，次級(jí)線圈有1到1的轉(zhuǎn)換率。
CMV變壓器依靠初級(jí)和次級(jí)之間的磁耦合原理，例如任何初級(jí)線圈中電壓在次級(jí)線圈中都會(huì)感應(yīng)出同向和反向電壓，從而消除電壓。可以簡(jiǎn)單的通過(guò)增加更多次級(jí)線圈擴(kuò)展原理到多信號(hào)通道——每個(gè)通道都有一個(gè)次級(jí)線圈（圖2）。
然而實(shí)際上，CMV變壓器的Achilles heel將分貝降到噪聲頻譜的低頻端。線圈的感應(yīng)電抗必須遠(yuǎn)大于電纜阻抗，噪
聲抵消導(dǎo)致這個(gè)情形出現(xiàn)。數(shù)百毫亨的電磁感應(yīng)對(duì)滿足低到60Hz頻率標(biāo)準(zhǔn)是必要的。對(duì)多通道應(yīng)用，需要消除低到60Hz頻率，轉(zhuǎn)換到許多銅、核心、體積和重量。然而，如果對(duì)設(shè)計(jì)消耗的功率不介意，可以使用接地：有源驅(qū)動(dòng)CMV核。
圖3中，由LT1797高頻運(yùn)放和MOSFET組成的功率放大器驅(qū)動(dòng)核，精確的抵消對(duì)參考地連接敏感的CMV。結(jié)果這些顯著的線圈感應(yīng)大大增加，可以減少“線圈”到簡(jiǎn)單的單一環(huán)形核通帶。也就是說(shuō)，一旦通過(guò)“圓環(huán)上的孔”，需要多觸電信號(hào)線完成CMV40dB或更抵消，從十幾到數(shù)百萬(wàn)Hz的范圍內(nèi)。
英文原文：
Actively driven ferrite core inductively cancels common-mode voltage
Active inductive noise cancellation cancels common-mode-voltage error components in both inut and output signals.
W Stephen Woodward, Chapel Hill, NC; Edited by Charles H Small and Fran Granville -- EDN, 12/14/2007
An earlier Design Idea illustrated one approach to that traditional headache for the analog designer: the dreaded ground loop (Reference 1). That Design Idea described a simple and efficient multichannel circuit. But it’s an asymmetrical CMV (common-mode-voltage) approach in that it works only at the receiving end of a cable. It therefore applies only to signal inputs and does nothing for outputs. However, in cases in which CMV consists of purely ac noise, a different CMV-remediation method—active inductive cancellation—works bidirectionally and therefore cancels CMV-error components in both input and output signals.
Engineers have for many years used passive-CMV inductive cancellation (Figure 1). Sometimes called a “humbucker transformer” because the power mains’ 60-Hz “hum” is often a dominant CMV component, the CMV inductor comprises a primary winding in series with the ground connection between the signal source (1) and the destination (2) and a secondary winding with a 1-to-1 turns ratio.
The principle of the CMV transformer relies on magnetic coupling between the primary and the secondary, such that any voltage that appears across the primary induces an equal and opposite voltage in the secondary, thus canceling it. You can easily extend the principle to multiple signal channels simply by adding more secondary windings—one secondary for each channel (Figure 2).
However, the Achilles’ heel of the CMV t
ransformer is the fact that the decibels of cancellation fall off at the low-frequency end of the noise spectrum. This situation occurs because noise cancellation depends on the fact that the inductive reactance of the windings must be much larger than the impedance of the cable. Hundreds of millihenries of inductance are necessary to satisfy this criterion for frequencies as low as 60 Hz. For multichannel applications requiring cancellation for frequencies as low as 60 Hz, this fact translates to lots of copper, core, bulk, and weight. However, if you don’t mind if your designs consume a little power, then a work-around exists: actively driving the CMV core.
In Figure 3, the power amplifier comprising the LT1797 high-frequency op amp and MOSFET forces the driven core to precisely cancel CMV as sensed in the ground-reference connection. The result is such a large multiplication of the apparent winding inductance that you can reduce the “windings” to a simple single pass-through of the toroid core. In other words, you need to thread a multiconductor-signal cable only once through the “hole in the doughnut” to achieve CMV of 40-dB or more cancellation, extending from tens to millions of hertz.
Reference
Woodward, W Stephen, “Amplifier cancels common-mode voltage,” EDN, May 10, 2007, pg 82.