Widlar Current Source

In the world of precision electronics, the performance of complex circuits like amplifiers and sensors often depends on something deceptively simple: a tiny, unwavering flow of current. Generating these minuscule bias currents is a fundamental challenge in integrated circuit design, as traditional methods using large resistors are physically impractical on a microchip. This leaves designers with a critical knowledge gap: how can one create a stable, microampere-level current source efficiently? This article demystifies one of the most elegant solutions to this problem, the Widlar current source. Across the following chapters, you will gain a deep understanding of its foundational principles and unique operational advantages. The first chapter, "Principles and Mechanisms," will dissect the circuit itself, revealing how a single resistor enables a logarithmic scaling of current and provides exceptional stability. Following that, "Applications and Interdisciplinary Connections" will explore how this ingenious circuit is used as a cornerstone building block for biasing critical components like differential amplifiers, cementing its role in modern analog design.

Imagine you're trying to build a very precise clock. Not just any clock, but one at the heart of a sophisticated electronic device, like a medical sensor or a precision amplifier. The "tick-tock" of this clock isn't a pendulum; it's a steady, unwavering flow of electric current. This current, known as a ​​bias current​​, is the lifeblood of the circuit. It must be minuscule, perhaps just a few millionths of an ampere, and it must hold its value with heroic stability, ignoring the jostles of changing temperature or a slightly wavering power supply.

How would you create such a tiny, stable current? Your first instinct might be to use a resistor. Ohm's law, $V = IR$, tells us that for a given voltage $V$, a large resistance $R$ will produce a small current $I$. But in the microscopic world of an integrated circuit, or "chip," this presents a problem. A resistor large enough to create a microampere-level current from a typical on-chip voltage would be physically enormous, consuming precious real estate and costing a fortune to manufacture. It's like trying to build a delicate wristwatch using a dam to control water flow. There has to be a better way.

A much cleverer approach is the ​​current mirror​​. In its simplest form, it uses two identical transistors, let's call them $Q_1$ and $Q_2$. We create a "reference" current, $I_{REF}$, and feed it into $Q_1$, which is cleverly wired to act like a template. Transistor $Q_2$ is set up to "watch" $Q_1$ and mimic its behavior. Because the transistors are identical twins, the output current from $Q_2$, called $I_{OUT}$, becomes a nearly perfect copy of $I_{REF}$.

This is a fantastic trick, but it doesn't solve our original problem completely. To get a tiny output current, we still need to generate a tiny reference current, and we're back to wrestling with large, impractical resistors to create $I_{REF}$. What we really want is a way to use a reasonably sized, robust reference current—say, 200 microamperes—and "step it down" to the mere 15 microamperes our delicate circuit needs.

This is where the genius of Bob Widlar enters the scene. The Widlar current source is a subtle but profound modification of the simple mirror, and it all hinges on one tiny addition: a single resistor.

The magic of a transistor lies in the exponential relationship between the voltage across its base-emitter junction ($V_{BE}$) and the collector current ($I_C$) that flows through it. The relationship is described by $I_C = I_S \exp(V_{BE}/V_T)$, where $I_S$ is a constant for the transistor and $V_T$ is the "thermal voltage," a small voltage of about 26 millivolts at room temperature. This exponential behavior means that a tiny change in $V_{BE}$ produces a huge change in $I_C$.

A simple current mirror works by forcing the $V_{BE}$ of both transistors to be equal, which in turn makes their currents equal. The Widlar source deliberately breaks this symmetry. It places a small resistor, $R_E$, at the emitter of the output transistor, $Q_2$. Now, as the output current $I_{OUT}$ flows through this resistor, it creates a voltage drop equal to $I_{OUT} R_E$. This voltage pushes back, effectively reducing the base-emitter voltage available to $Q_2$ compared to $Q_1$.

The two transistors share a common base connection, so the base voltage $V_B$ is the same for both. For the reference transistor $Q_1$, we have $V_{BE1} = V_B$. For the output transistor $Q_2$, its emitter is lifted up by the voltage across $R_E$, so its base-emitter voltage is $V_{BE2} = V_B - I_{OUT} R_E$. The crucial voltage difference is therefore created by the resistor itself: $\Delta V_{BE} = V_{BE1} - V_{BE2} = I_{OUT} R_E$.

Because of the exponential nature of the transistor, this small, simple voltage difference leads to a ratio of currents:

$\frac{I_{REF}}{I_{OUT}} = \exp\left(\frac{\Delta V_{BE}}{V_T}\right) = \exp\left(\frac{I_{OUT} R_E}{V_T}\right)$

If we turn this equation around, we get the master formula for designing a Widlar source:

$R_E = \frac{V_T}{I_{OUT}} \ln\left(\frac{I_{REF}}{I_{OUT}}\right)$

This equation is the heart of the Widlar source. It shows that the current isn't copied 1-to-1 anymore; instead, it's scaled down by a factor related to the natural logarithm. This is what we call the "logarithmic squeeze." To achieve a 10-to-1 reduction in current ($I_{REF}/I_{OUT} = 10$), we only need to create a voltage difference across $R_E$ of about $59.9$ millivolts at room temperature. This is a tiny voltage, easily generated by a small current flowing through a modestly sized resistor. For instance, to generate a $15 \, \mu\text{A}$ output from a $200 \, \mu\text{A}$ reference, a resistor of only about $4.49 \, \text{k}\Omega$ is needed—a value easily fabricated on a chip. This elegant technique allows designers to generate very small, stable currents from larger, more manageable ones, solving the "tyranny of the resistor" problem beautifully.

The benefits of that little resistor don't stop there. One of the hallmarks of a perfect current source is an infinitely high ​​output impedance​​ (or output resistance). In simple terms, this measures how "stubborn" the current source is. If the voltage across the source changes, does the current it supplies change as well? A perfect source would say "no," delivering its specified current no matter what. A high output impedance means the current changes very little, which is exactly what we want.

A simple current mirror has an output impedance equal to the transistor's own internal resistance, $r_o$. This is good, but not great. The Widlar source is far better. The emitter resistor, $R_E$, introduces a powerful form of local negative feedback. Imagine the output current $I_{OUT}$ tries to increase for some reason. This would cause the voltage drop across $R_E$ to increase, which in turn reduces $V_{BE2}$, immediately throttling the transistor and pulling the current back down. This self-regulating action dramatically multiplies the effective output impedance.

The output resistance of the Widlar source is approximately $R_{out, Widlar} \approx r_o (1 + g_m R_E)$, where $g_m$ is the transistor's transconductance. This "impedance boosting" factor, $(1 + g_m R_E)$, provides a substantial improvement over a simple mirror. When compared to other current source designs, the Widlar source offers a significant improvement in output impedance over the simple mirror, although more complex circuits like the Wilson mirror can achieve even higher values. This high output impedance is a key reason for its superior performance.

This rock-solid output impedance directly translates into another desirable trait: excellent ​​power supply rejection​​. If the main voltage supply of the chip fluctuates—a common occurrence—this "noise" can affect the voltage across the current source. Because the Widlar source is so stubborn about maintaining its output current (thanks to its high output impedance), it is largely immune to these supply variations, leading to a much cleaner and more stable bias for the rest of the circuit.

Perhaps the most beautiful demonstration of the circuit's elegance is its behavior with changing temperature. Transistor properties are notoriously sensitive to temperature. The saturation current $I_S$ and the thermal voltage $V_T$ both change significantly as the chip heats up or cools down. You might expect the output current of the Widlar source to drift wildly.

But here, the ratio-based design works its magic. When we analyze the temperature coefficient of the output current, a remarkable thing happens. Because both transistors are matched and experience the same temperature, the complex and highly variable term for the saturation current, $I_S(T)$, appears on both sides of the current ratio and simply cancels out. The final expression for the temperature sensitivity depends only on $V_T$, $R_E$, and $I_{OUT}$ itself. This cancellation is not an accident; it's a profound consequence of building a circuit based on the matching and ratio of components, rather than their absolute values. It's a cornerstone of robust analog design.

Of course, no circuit is perfect. Our analysis so far has assumed ideal transistors with infinite current gain ($\beta$). In reality, a tiny amount of current is needed to control the base of each transistor. Including these base currents modifies the core equation slightly, but the fundamental principle remains the same.

Another real-world concern is electronic noise. Every component generates a tiny, random hiss. In a simple mirror, the noise from the reference transistor is mirrored directly to the output. The Widlar source has a more complex noise profile, with contributions from both transistors' shot noise and the thermal noise of the emitter resistor. Detailed analysis shows how these sources combine, and while a Widlar source is not necessarily "quieter" than a simple mirror in all cases, understanding its noise behavior is critical for designing ultra-sensitive electronics.

From a single, strategically placed resistor, an entire cascade of desirable properties emerges: the ability to generate tiny currents, exceptionally high output impedance, excellent immunity to power supply noise, and remarkable thermal stability. The Widlar current source is a testament to the power of understanding the fundamental physics of a device and exploiting its properties with an elegant, minimalist design.

Now that we have taken apart the elegant little machine that is the Widlar current source and understood its inner workings, a most important question remains: So what? What is this clever arrangement of transistors for? If the principles and mechanisms are the grammar of our story, then the applications are the story itself. And it is a fascinating one, revealing how a simple idea can ripple outwards, becoming a cornerstone of modern technology and a beautiful illustration of the interplay between abstract theory and the messy, physical world.

The journey of the Widlar current source begins not in a grand theater, but in a hidden, miniature world where the rules are different. Imagine you are a watchmaker, tasked with building an impossibly small, precise timepiece on the head of a pin. But there's a catch: one of your essential tools is a sledgehammer. This is the predicament of the integrated circuit designer. To make circuits work reliably, they often need very small, very stable electric currents, perhaps just a few microamperes. The textbook way to get a small current from a larger voltage is to use a very large resistor, following Ohm's law, $I = V/R$. But on a silicon chip, where space is the ultimate luxury, a resistor large enough to do this job—on the order of mega-ohms—would be a monstrous, sprawling giant. It's the sledgehammer in the watchmaker's shop: clumsy, inefficient, and utterly impractical.

Here, the Widlar source performs its first and most fundamental magic trick. It offers a way to generate a tiny, stable output current, say $20 \, \mu\text{A}$, from a much more convenient and manageable reference current, like $1 \text{ mA}$, without a gigantic resistor. It does this by exploiting the beautiful exponential relationship between voltage and current in a transistor. The small emitter resistor, $R_E$, doesn't just sit there; it creates a voltage that pushes back against the driving voltage. Because of the transistor's logarithmic nature, a small voltage difference results in a large ratio of currents. The Widlar circuit is, in essence, a "logarithmic lever," where a small, practical resistor can be used to create the effect of a much larger one, giving us precise control over minuscule currents. This principle is not some quirk of one type of transistor. It is a universal strategy. The same architectural idea works beautifully in MOSFET circuits, the workhorses of the digital age, even though their underlying physics follows a square-law relationship instead of an exponential one. The theme is the same: use an active device's own physics against itself to achieve something that would be brutishly difficult otherwise.

Having mastered the art of creating small currents, the Widlar source does not simply rest on its laurels. It takes on a new, more profound role: that of the unsung hero, the quiet enabler of more complex and powerful circuits. A great actor is nothing without a well-lit stage, a stable platform from which to perform. In the theater of microelectronics, many of the star performers, like amplifiers, need a stable "operating point," a quiescent state of currents and voltages from which they can work their magic on incoming signals. This is called biasing.

Consider the differential amplifier, a marvel of symmetry that forms the input stage of nearly every operational amplifier (op-amp). Its job is to amplify the minuscule difference between two input signals while ignoring anything they have in common. Its ability to do this depends critically on the total current supplied to its two transistors—a "tail current" that must be as steady and unwavering as the North Star. Any fluctuation in this tail current would create noise and errors, ruining the performance. The Widlar source is the perfect candidate for this job. It provides a stable, predictable tail current that anchors the entire differential pair, setting the stage perfectly for the amplification to happen. It works silently in the background, but without it, the main act would fall apart.

Yet, in the world of engineering, "good enough" is rarely good enough. The quest for perfection is relentless. For a current source, perfection means being an unshakable, stubborn source of current, no matter what the rest of the circuit does. We can measure this "stubbornness" with a property called output resistance. An ideal current source would have an infinite output resistance—it would supply its set current regardless of the voltage changes at its output. The basic Widlar source is good, but its output resistance is finite. How can we improve it? By being clever and combining our circuit "tricks."

One of the most powerful techniques is the "cascode" configuration, where we stack a second transistor on top of the first. This new transistor acts like a shield. It holds the voltage at the collector of the main current-source transistor nearly constant, shielding it from the wild voltage swings of the outside world. The result? The output resistance of the combined circuit skyrockets, bringing it much closer to the ideal. By adding a cascode transistor to our Widlar source, we create a hybrid circuit that marries the Widlar's low-current capability with the cascode's immense output resistance. This illustrates a deep principle of engineering design: modularity. We can also use a Widlar source simply to generate a stable reference current for an entirely separate, high-performance cascode mirror, creating even more sophisticated and robust systems.

Up to now, our discussion has lived in the clean, orderly world of textbook diagrams. But the leap from an abstract schematic to a physical piece of silicon is a journey into a wonderfully messy reality. On a real silicon wafer, no two transistors are ever perfectly identical. Due to subtle variations in the manufacturing process, there are often smooth, linear "gradients" in properties like current gain ($\beta$) across the chip. If our reference transistor is at one end of the circuit and our output transistor is at the other, this gradient will cause a mismatch, and our current mirror will no longer be accurate.

The solution to this problem is not found in more complex electronics, but in simple, elegant geometry. Imagine you are laying five tiles in a row on a floor that has a slight, imperceptible slope. One tile is red ('R'), and four are orange ('O'). If you place the red tile at one end and the four orange tiles next to it (R O O O O), the average height of the orange tiles will clearly be different from the height of the red tile. But what if you place the red tile in the middle, symmetrically flanked by the orange ones (O O R O O)? The average position—and thus the average height—of the orange tiles now exactly matches the position and height of the red one. The effect of the slope is cancelled out. This is the principle of the "common-centroid layout." By physically arranging the unit cells of our transistors on the chip in this symmetric, interdigitated pattern, we can average out the linear process gradients and achieve a much higher degree of matching and precision. It is a beautiful example of how physics, geometry, and engineering unite to solve a very practical problem.

The physical world imposes other realities, too. Our models assume transistors are perfect, but in reality, a small amount of current always "leaks" into the base terminal, an effect quantified by the finite current gain, $\beta$. This means our calculations for current division must be refined to account for these small, but sometimes significant, base currents. Furthermore, circuits live in the domain of time and frequency. The Widlar source is not just a DC device; its behavior changes at high frequencies. Sometimes we might even add a capacitor to its base node on purpose, to filter out high-frequency noise and improve stability. This, of course, turns its output impedance into a frequency-dependent quantity, a crucial consideration in the design of radio-frequency or other high-speed systems.

Finally, every device has its limits. The Widlar source promises to deliver a constant current, but it can only keep that promise as long as the circuit it's connected to "plays fair." It requires a certain minimum voltage at its output to keep its main transistor in the proper "forward-active" region. If the external load forces this voltage too low, the transistor is driven into saturation. It ceases to be a high-impedance source and starts to behave more like a simple switch with a low resistance. Understanding this operational boundary, or "compliance voltage," is critical for robust design; it is the edge of the map where the contract is broken and the device no longer behaves as intended.

From a clever trick to avoid large resistors to an essential building block in complex amplifiers, and from a theoretical ideal to a physical object whose very geometry dictates its performance, the Widlar current source is a microcosm of analog design. It teaches us that elegance often lies in simplicity, that the most useful tools are often those that turn a device's own nature to our advantage, and that true mastery lies in bridging the gap between the perfect world of ideas and the beautiful, imperfect reality of our physical universe.