Basic metrics of Linear Regulators

Introduction

Our ICs often have a wide range of supply voltages, while the internal circuitry is designed to operate on a much more narrow range of voltages. This is where LDOs come in.

Linear voltage regulators (or LDOs) are an essential building block of every modern IC. Not only do they stabilize supply voltages, but they are also able to provide a low-noise (quiet) environment for sensitive circuits.

LDO conceptualized

We can conceptualize an LDO through a resistor divider:

A resistor divider is a voltage regulator.

The value of the output voltage is:

At a fixed value of V_IN and R_LOAD, we can calculate the value of R_IN to set V_OUT to a fixed value. The problem appears if V_IN and R_LOAD change over time. Then, we can no longer set the value of V_OUT using a fixed value of R_IN. For every change in V_IN and R_LOAD, we need to re-adjust the value of R_IN. as R_IN will pass a different amount of current to accommodate a fixed V_OUT value, let’s rename R_IN into ‘pass device’.

What properties would the pass device have if we could change its value to keep V_OUT fixed?

• If V_IN would increase/decrease, the pass device would increase/decrease the amount of current passed to the output to compensate
• If R_LOAD would increase/decrease, the pass device would decrease/increase the amount of current passed to the output to compensate

From these properties, we can conclude that a variable the pass device would be an active component that would regulate V_OUT. A component that can change its resistance depending on a specific control value is a transistor.

LDO — From a resistor divider to a control loop

One thing we can observe is that a feedback loop is required to control the value of the pass device. To know if V_OUT is at the right value, the best course of action is to compare it with a stable reference voltage. We can use an Error amplifier (ER) to calculate and amplify the error which we can feed into the pass-device. The loop will keep amplifying the error as long as the error is not 0. Once the error is 0, we have achieved a state where the value of V_OUT is the desired one and we have reached a steady state. A Steady-state describes a state where no values are changing.

Using these concepts, the following is a more expanded schematic of an LDO:

The main concepts which we’ve included so far are:

• The pass device changes its resistance based on the control value
• The Error amplifier (EA) creates the control value by amplifying the error between the output voltage and the reference voltage
• The output voltage can be scaled using a resistor divider
• This regulation loop can keep V_OUT regulated for different values of V_IN and R_LOAD

So far, we haven’t discussed any of the ways how to implement an LDO. The focus of this article is to describe various properties of an LDO. However, we will slowly introduce different implementations over time.

How do we evaluate an LDO?

There are several metrics that we can use to evaluate an LDO. Depending on the role of the LDO in the system the importance of these metrics will vary.

Dropout voltage

The dropout voltage of an LDO is the voltage difference between V_IN and V_OUT at which we can stop regulating V_OUT.

Line regulation graph with annotated regions of operation

When we look at a transfer characteristic of an LDO, we can observe the dropout voltage at the point where the value of V_OUT is no longer fixed. After that, the LDO is in the dropout region, and the value of V_OUT is approximately equal to V_IN. If V_IN is further reduced the LDO turns off and V_OUT = 0 V.

Dropout voltage also has a dependency on the load current. At higher currents, the dropout voltage occurs at higher V_IN values.

(Current) Efficiency

Normally, we define the efficiency of a system by comparing how much input power is used at the output load. Given that the LDO is a series structure that operates by dissipating excess power on the pass device, power efficiency is not a good metric to evaluate an LDO. Instead, we can use current efficiency. This just means that instead of power, we compare currents. We can define the current efficiency of an LDO as:

The difference between the input and output current is the current which we have to spend to properly regulate the system. This current supplies the error amplifier, reference (V_{REF}) circuit, feedback resistors, etc. We can call this current quiescent current I_Q. Then, we can expand current efficiency into:

Static metrics

Line regulation

Line regulation refers to the ability of an LDO to maintain the desired V_OUT for a given value of V_IN. An ideal LDO will have a fixed value of V_OUT irrespective of V_IN while in the regulation region. In reality, V_OUT will vary with V_OUT. The graph we observe will be the same as when we’re looking for the dropout voltage.

Load regulation

Load regulation refers to the ability of an LDO to maintain the desired V_OUT for a given value of load current I_LOAD. An ideal LDO will have a fixed value of V_OUT irrespective of V_IN while in the regulation region. In reality, V_OUT will vary with I_LOAD.

Dynamic metrics

Line transient response

Line transient response refers to the response of the LDO and the shape of V_OUT when there is a change in the V_IN. It is usually tested by applying a step pulse inV_IN after the LDO has entered a steady state. The step pulse amplitude can range from small-signal to large-signal values as testing the line response transient should incorporate all the circuit nonlinearities.

Load transient response

Line transient response refers to the response of the LDO and the shape of V_OUT when there is a change in the I_LOAD. It is usually tested by applying an I_LOAD step pulse after the LDO has entered a steady state. The step pulse amplitude can range from small-signal to large-signal values as testing the load response transient should incorporate all the circuit nonlinearities. A good rule of thumb is to test at least 1–2 current values per decade (e.g. 10 uA, 50 uA, 100 uA, 500 uA, 1 mA, etc.)

Power supply rejection ratio

The power supply rejection ratio (PSRR) refers to the ability of the LDO to suppress disturbances on the supply rails of the circuit and keep V_OUT unaffected. PSRR is usually tested by applying a small AC signal on a supply rail of interest and measuring the ratio between the input and the output signal at that frequency. The frequency can be swept along the frequency band of interest.

PSRR of an LDO.

PSRR resembles the line transient test, however, PSRR is measured for small signal values (as it’s tested using AC analysis).

Stability

Stability is the most critical metric of an LDO. While we can take wide trade-offs with the other LDO metrics, the stability of an LDO is a must and every LDO must be stable. The most common way of testing an LDO is by performing a small signal AC analysis in an open loop over a range of frequencies.

The resulting bode plot can tell us more about stability.

Open loop frequency response of an LDO.

The most notable metric from the bode plot is the phase margin (PM). PM is calculated as the delta between 180 degrees and the phase value at the unity gain frequency (the frequency at which gain is 0 dB). There are different approaches to choosing the optimal value, however, the goal is to keep the value significantly above 0 degrees in the whole operating range of the LDO.

Although stability is most commonly tested using small signal AC analysis, it is often not enough to show for certain that the regulation loop is stable. Additional analysis should be performed, most notably, a transient simulation where a step response of the loop is observed in time. Some metrics in the small signal time domain can be mapped back into small signal frequency domain metrics. This topic is not covered in this article, however, it will be in future ones.

Conclusion

There are several different metrics with which we can evaluate the performance of any LDO. Using these metrics, we can make better trade-offs and maximize the required performance in the system.

It is also important to note that the way we make these trade-offs is much more complex than presented here. Many other parameters will affect our measurements and this article only scratches the basics. I will go deeper into these topics in future articles.

If this article has helped you, please consider supporting me with a coffee :)