Common-Mode Half-Circuit

In the field of electronics, the ability to isolate a faint, desired signal from pervasive background noise is a fundamental challenge. The differential amplifier, a cornerstone of analog circuit design, offers an elegant solution by amplifying the difference between two inputs while rejecting any signal common to both. However, to optimize this noise rejection capability, engineers need a precise method to analyze and minimize the amplifier's response to these unwanted common-mode signals. This is the knowledge gap that the common-mode half-circuit method expertly fills. This article provides a comprehensive exploration of this powerful analytical tool. First, in "Principles and Mechanisms," we will dissect the theory behind the half-circuit model, showing how circuit symmetry allows for this radical simplification and how it reveals the secret to suppressing common-mode gain. Following that, "Applications and Interdisciplinary Connections" will demonstrate how this principle is applied in practical, high-performance circuit design, revealing its limitations and its surprising connections to other areas like control theory and basic network analysis.

Imagine you are in a bustling café, trying to listen to a friend's quiet story. Your ears are bombarded by the clatter of dishes, the whir of the espresso machine, and the chatter of a dozen other conversations. This is the classic signal-in-noise problem. Your brain, however, performs a remarkable trick. It focuses on the subtle differences in sound arriving at your two ears to pinpoint your friend's voice and tune out the uniform background hum. In the world of electronics, we face the same challenge: how to amplify a tiny, meaningful signal while ignoring the loud, pervasive noise that contaminates it. The solution, much like our own hearing, is found in the elegant principle of ​​differential amplification​​.

Instead of building a single amplifier, we build two identical ones and arrange them in a perfectly symmetric configuration. This is the ​​differential pair​​, the cornerstone of modern analog circuits. This pair of transistors, be they BJTs or MOSFETs, is designed not to amplify a single input, but to amplify the difference between two inputs. The signal we care about—our friend's voice—is applied as a ​​differential-mode signal​​: a positive voltage to one input and an equal but opposite negative voltage to the other.

The noise, however, often presents itself as a ​​common-mode signal​​. Think of the 60 Hz hum from power lines that seeps into everything. It appears on both inputs with the same polarity and magnitude at the same time. Our goal is to design an amplifier that is exquisitely sensitive to the differential signal but utterly blind to the common-mode one. The measure of success in this endeavor is called the ​​Common-Mode Rejection Ratio (CMRR)​​, a number that tells us how much more the amplifier loves the signal than it loves the noise. To achieve a high CMRR, we must mercilessly suppress the amplifier's response to common-mode signals, a response we call the ​​common-mode gain​​ ($A_{cm}$).

How do we analyze the circuit's behavior when it's subjected to this unwanted common-mode signal? We could write out all the equations for the full circuit, a tedious but valid approach. But physics and engineering are about finding elegant shortcuts, and symmetry provides a magnificent one.

When a common-mode signal is applied, the gates (or bases) of our two identical transistors receive the exact same voltage. Because the circuit is perfectly symmetric, the response of the left half must be a mirror image of the response of the right half. The currents flowing are the same, the voltages at corresponding points are the same. There's no "difference" for the differential pair to amplify.

This perfect symmetry allows us to conceptually slice the circuit right down the middle. Since nothing crosses this line of symmetry, we can analyze just one half of the circuit, knowing that the other half is behaving identically. This simplified model is the celebrated ​​common-mode half-circuit​​.

But there is a subtle and beautiful catch. The two transistors are not entirely independent; they are joined at their source (or emitter) terminals, and this common point is connected to ground through a "tail" element, typically a resistor ($R_{SS}$) or a current source. What happens to this tail element when we slice the circuit in half?

Let's say a small-signal current $i_s$ flows out of the source of each transistor. In the full circuit, the total current flowing through the tail resistor $R_{SS}$ is $i_s + i_s = 2i_s$. The voltage at the common source node is therefore $v_s = (2i_s)R_{SS}$. Now, consider the perspective of just one of the transistors in our half-circuit. It supplies a current $i_s$ and sees a voltage $v_s$. From its point of view, the effective resistance it's driving is $R_{\text{eff}} = v_s / i_s = (2i_s R_{SS}) / i_s = 2R_{SS}$. It's a marvelous result: in the common-mode half-circuit, the tail impedance is doubled!. Each half of the pair experiences the tail impedance as if it were twice its actual value because they are sharing the burden of driving current through it.

With our powerful half-circuit model, calculating the common-mode gain becomes straightforward. The half-circuit is simply a common-source (or common-emitter) amplifier with a resistor of value $2R_{SS}$ in its source path—a configuration known as "source degeneration."

Let's trace the signal. A common-mode voltage $v_{icm}$ is applied to the gate. This tries to create a drain current. However, as this current flows, it must pass through the effective source resistance $2R_{SS}$, creating a voltage drop that raises the source voltage $v_s$. This increase in $v_s$ counteracts the initial input, reducing the effective gate-source voltage ($v_{gs} = v_{icm} - v_s$). This is a form of ​​negative feedback​​, and it is the key to suppressing the gain.

A careful analysis of this half-circuit reveals the single-ended common-mode gain to be:

where $g_m$ is the transconductance of the transistor and $R_D$ is the load resistor.

Let's look at this formula not as a mere equation, but as a recipe for design. To make $|A_{cm}|$ small, we need to make the denominator large. The term $1$ is fixed. The term $2g_{m}R_{SS}$ is our lever. By making the tail resistance $R_{SS}$ as large as possible, we can dramatically increase the denominator and crush the common-mode gain. This is the central principle of high-CMRR design.

What happens if we take this to its logical extreme? What if we replace the tail resistor with an ​​ideal current source​​? By definition, an ideal current source has an infinite output resistance ($R_{SS} \to \infty$). Plugging this into our formula, we see that the denominator becomes infinite, and the common-mode gain becomes exactly zero!. An ideal current source simply refuses to allow its current to change, no matter what the common-mode voltage does. It starves the amplifier of any ability to respond to the common-mode signal. This is why practical, high-performance differential amplifiers always use sophisticated transistor-based current sources for their tail biasing—to get that $R_{SS}$ as high as possible.

Now we can calculate our ultimate figure of merit. The differential gain, $A_d$, can be found using a similar half-circuit trick (the "differential half-circuit"), which for a simple configuration yields $A_d = g_m R_D / 2$. The Common-Mode Rejection Ratio is the ratio of these two gains:

This beautiful, simple result lays bare the secret to a great differential amplifier: a high transconductance $g_m$ and, most critically, a very high tail resistance $R_{SS}$. A CMRR of $1000$, or 60 dB, means the amplifier is a thousand times more sensitive to the desired signal than to the unwanted noise. Values over a million (120 dB) are achievable in precision integrated circuits.

Of course, the real world is always a bit more complex. In integrated circuits, a phenomenon called the ​​body effect​​ can create an additional path for feedback, slightly altering the gain equations, but the fundamental principle of suppression via a high-impedance tail remains unchanged.

Furthermore, at high frequencies, the tiny capacitances within the transistors come into play. A fascinating consequence of the half-circuit analysis is that the effective input capacitance presented by the amplifier is different for differential and common-mode signals. For a differential signal, the famous ​​Miller effect​​ can dramatically increase the input capacitance. For a common-mode signal, the source degeneration we discovered actually helps to reduce the input capacitance. The amplifier literally changes its "personality" depending on the type of signal it sees!

Even so, these advanced topics all build upon the foundational and powerful idea of symmetry and the half-circuit model—a testament to how a simple, elegant concept can unlock a deep understanding of complex systems.

In our previous discussion, we saw how the wonderful property of symmetry allowed us to dissect the behavior of a differential amplifier. By splitting the input into a "differential mode" and a "common mode," we could analyze the circuit's response to the signal we care about. But the true power and beauty of a physical principle are revealed when it's applied to the messy, imperfect real world. Now, we are going to take that same idea of symmetric and anti-symmetric decomposition and see how it becomes an indispensable tool for engineers, guiding them in the art of building amplifiers that can hear a whisper in a hurricane of noise.

The secret lies in looking at the circuit's response to the part of the signal we don't want: the common mode. This is the noise that pervades our electronic world—the incessant 60-hertz hum from our power lines, the high-frequency chatter from a nearby microprocessor—that couples onto both input wires of our amplifier simultaneously. An ideal differential amplifier would ignore this common-mode racket completely. But how does a real one fare?

To answer this, we turn again to our method. When a purely common-mode signal is applied, both halves of the differential pair do the exact same thing at the exact same time. The circuit's inherent symmetry means we can mentally slice it down the middle and analyze just one side—the "common-mode half-circuit." What we find is remarkable. The differential pair, under common-mode excitation, no longer behaves as a pair at all! Instead, it transforms into two identical, but separate, common-emitter amplifiers working in lockstep. The shared tail resistor, which seemed to tie the two halves together, now plays a new role. Since it carries the current from both transistors, it behaves, from the perspective of our half-circuit, like a resistor of twice its value sitting in the emitter path. This is a classic case of an amplifier with emitter degeneration, a form of local feedback.

This insight is fantastically useful. It tells us that the gain of the amplifier to unwanted common-mode noise is essentially the gain of a simple common-emitter stage with a large emitter resistor. And what do we know about such amplifiers? Their gain is approximately the ratio of the collector resistor to this emitter resistor, $A_{v,cm} \approx -R_C / (2R_E)$. This is the enemy: the common-mode gain, $A_{cm}$. To make our amplifier deaf to noise, our goal is to make this gain as close to zero as possible. The formula points the way: we need to make the effective tail impedance enormous.

How do we achieve an enormous impedance? A simple resistor won't do. To get good performance, we would need a resistor so large it would be impractical to fabricate and would create impossible biasing conditions. This is where engineers get clever. Instead of a passive resistor, we use an active device—a transistor—configured as a current source. A good current source behaves like a very large resistance. How large? A practical design problem might ask for a Common-Mode Rejection Ratio (CMRR) of 80 dB. To achieve this with a standard MOS differential pair, our analysis reveals that the tail current source must have an output resistance not in the thousands, but in the millions of ohms!. This is the first step in our quest for perfection: replacing a simple resistor with an active current source to starve the common-mode signal of any path to ground.

But we can do even better. Engineers are never satisfied. How can we get an even higher resistance? By using a wonderfully elegant trick called a ​​cascode​​. By stacking a second transistor on top of our current source transistor, we can dramatically boost its effective output resistance. The top transistor acts as a shield, holding the voltage across the bottom transistor nearly constant, which makes the bottom transistor behave like a much more ideal current source. The improvement is not just a few percent; it can be staggering. A theoretical analysis shows that replacing a single-transistor current source with a cascode version can increase the CMRR by a factor of thousands. Incredibly, this improvement factor turns out to be related to fundamental device parameters, approximately given by the ratio of the transistor's Early Voltage to the thermal voltage, $V_A/V_T$. It's a beautiful example of how a simple circuit topology change yields a massive performance boost.

Let us imagine for a moment that we have built the perfect current source, one with truly infinite resistance. Our half-circuit analysis tells us the common-mode gain should be zero. Are we finally safe from noise? In an ideal symmetric circuit, yes. However, we have ignored a crucial detail: our two halves of the differential pair are never truly identical. Microscopic variations during the fabrication of an integrated circuit mean that the resistor on the left side, $R_{C1}$, will be ever so slightly different from the resistor on the right, $R_{C2}$.

This asymmetry combines with the finite impedance of a real-world tail source to limit performance. When a common-mode signal is applied, a finite tail impedance allows a small common-mode current to flow in each branch. If these currents then flow through sightly different resistors, they produce slightly different output voltages. A common-mode input has created a differential-mode output. This phenomenon, called common-mode to differential-mode conversion, is a primary limiter of performance in high-precision circuits. In a practical design, a 1% resistor mismatch can become the dominant limitation, capping the achievable CMRR at a value that may be around 40 dB (a factor of 100). This is why a receiver for a high-speed logic family like ECL, which relies on differential signaling for noise immunity, is still susceptible to common-mode noise if its internal components are not perfectly matched. The pursuit of perfection in differential amplifiers is, in many ways, a battle for symmetry.

Our world, of course, is not static. Signals and noise vary in time, often at very high frequencies. What happens to our carefully engineered CMRR then? Trouble arrives in the form of an old friend (or foe): capacitance. Our magnificent, high-resistance tail current source is not just a pure resistance. It has a parasitic capacitance associated with it, a consequence of the physical structure of the transistors. At low frequencies, this capacitance is an open circuit and our large tail resistance dominates. But as the frequency of the common-mode noise increases, the impedance of this capacitor, $1/(\omega C)$, plummets.

At some high frequency, the capacitor effectively becomes a short circuit to ground, completely bypassing our carefully designed tail resistance. The CMRR, so high at DC, is destroyed. This is a universal story in electronics: parasitic effects that are negligible at low frequencies come to dominate at high frequencies, defining the limits of performance. A more complete analysis reveals a rich frequency-dependent behavior, where the CMRR starts high at DC, begins to roll off because of the tail capacitance, and may hit a "floor" determined by resistor mismatches, creating a transfer function with its own poles and zeros.

The principle of analyzing common-mode behavior extends far beyond just rejecting noise. In modern, fully-differential op-amps, which have two outputs instead of one, the average DC voltage of the outputs—the output common-mode level—is not automatically defined. If left uncontrolled, it could drift to either the positive or negative supply rail, rendering the amplifier useless. The solution is a clever application of feedback: a ​​Common-Mode Feedback (CMFB)​​ circuit.

This control system continuously monitors the output common-mode level, compares it to a desired reference voltage, and generates a control signal. Where does this control signal go? To the gate of the tail current source, of course! By adjusting the tail current, the CMFB loop can steer the output common-mode voltage back to its target value. Analyzing the stability and performance of this feedback loop requires us to understand how a change in the tail current source translates into a change in the output common-mode voltage—a calculation that relies directly on the principles of common-mode analysis we have developed. It's a beautiful marriage of amplifier design and control theory.

And lest we think this powerful idea of splitting the world into symmetric and anti-symmetric pieces is a special trick just for transistors, let's consider a humble DC circuit made of nothing but resistors and voltage sources. If the resistor network has a left-right symmetry, but the driving voltages are different, the problem looks complicated. But by decomposing the sources into their common-mode ($V_{cm} = (V_1+V_2)/2$) and differential-mode ($V_{dm} = V_1-V_2$) parts, the solution becomes astonishingly simple. Under the common-mode excitation, the symmetry immediately tells us that no current can flow through the central bridging resistor. It's as if it were cut. Under the differential-mode excitation, the anti-symmetry tells us the center of that same resistor must be at zero volts—a "virtual ground." We can analyze each half-circuit separately and add the results. The complex problem dissolves into two trivial ones. The very same "half-circuit" and "virtual ground" concepts we used for our most advanced amplifiers are right here, in the most basic of circuits.

This is the mark of a truly fundamental idea. The common-mode half-circuit is not just a calculation tool; it is a way of seeing. It is a lens that uses the elegant and powerful concept of symmetry to understand how real-world systems behave, how to design them to be robust against noise, how their performance is limited by frequency and physical imperfection, and how the same core principles echo across different fields of science and engineering.