My Buck-Boost Converter is Unstable: What Should I Do?

Buck-boost converters are designed to take a wide range of input voltage values while providing a regulated output voltage or current. The output can be adjustable or fixed, but regardless of the adjustability, the regulator always attempts to adjust the output to provide a target output value. When the input, output, or both quickly vary, there is a possibility that the regulator stops providing a stable output. This regime is known as instability, and when it occurs the regulator fails to track an output voltage.

When we refer to “instability” we are not referring to the superimposed ripple on top of a nominal DC output voltage. Instead, instability refers to the case where the output voltage oscillates, fails to start up, or diverges slowly from nominal. There are various reasons this might happen, and sometimes you might not realize instability is occurring without a direct measurement with an oscilloscope.

Buck-boost controllers tend to have wide input voltage allowances, so they can have instabilities arising due to the control loop function as well as variations in input voltage. For example, we can have instabilities arising from any of the following factors:

• Dropout near the low-end of the input range
• Unintentional tripping of protection mechanisms
• Excitation of poles in the control loop or output interface
• Inability to track changes in output voltage

When instability is located, what can be done to solve the problem? There are several options, such as redesigning portions of the circuit. In the case of integrated PMICs or gate drive controllers, it is possible the component is being used outside of its intended operating range. The first step is to spot instabilities through measurement and narrow down to a short list of causes of instability.

How Do You Spot Instability?

Before getting into what causes an instability in a power converter, it’s important to know what instability would look like if you were monitoring output voltage/power in a measurement. The best case is to monitor both the output voltage from the power converter as well as power delivery to the load. Monitoring both takes two instruments: an oscilloscope can be hooked to the output port to monitor the output voltage waveform, and an electronic load can be used to monitor DC power output.

This RIGOL oscilloscope is an excellent option for monitoring power supply output, including for digital systems with fast edge rates.

When a power system has multiple rails operating at once, multiple DC loads will be needed, and it’s a good idea to monitor multiple rails simultaneously in each channel on the oscilloscope.

The various reasons for instability that could be observed with this measurement setup include:

• The input voltage is too close to the low-end input range
• Protection mechanisms tripping:
  • Transient on output is tripping the current limiting feature on the regulator
  • Excessive current (input or output) trips the temperature limiting feature
• Transient excitation on the input or output filter due to presence of a pole
• There is a violation in the regulator’s switching ON time, causing the voltage to drop out
• Underdamped transients are excited at the switching node and observed on the output
• The control loop cannot respond fast enough to track changes in the output voltage

Some of these instabilities will appear very similar to each other, so an observation of an instability on the output will typically require further investigation to determine the root cause.

Typically on the oscilloscope, one would observe some type of oscillation on the output voltage reading. An example is shown below.

A simple sine wave oscillation indicates some type of instability, driven by an excited pole or repeatedly excited transient during switching.

The corresponding reading on the electronic DC load will show a variation in the output power around some nominal power output. The two measurements can be observed to be coincident; when the source of the instability is removed, the output will again stabilize.

Power Dropout at Low V-in

Wide-input boost converters will start to drop out if the input voltage drops too low. Typically you would first observe the power conversion efficiency begin to decrease as you start to bring the input voltage closer to the lower limit, regardless of the power output capabilities of the sourcing DC supply. Eventually, when you get very close to the low-end of the input voltage range, and any transient on the input will cause the input power to oscillate around the minimum input voltage threshold.

While monitoring the electronic load power and scope, you will typically observe the power output begin to oscillate between zero and the target DC output.

Continuous resetting of the regulator may occur near the low end of the input voltage range.

A simple solution here is to create a turn-on circuit that toggles the regulator’s enable function. The threshold for this can be set just above the minimum input voltage range. A typical type of circuit would be to use a comparator circuit with hysteresis.

Protection Circuit Tripping

In some cases, one of the protection mechanisms built into the central regulator, controller, or gate driver may get triggered during operation, which might result in the appearance of an instability. In these cases, the regulator could be engaging a limiter circuit (in the case of current limiting), an overvoltage shutdown circuit, or a UVLO trigger that briefly suspends the regulator. These factors are not so easy to distinguish from each other just by looking at an oscilloscope.

While monitoring a scope and electronic DC load will tell you there is a problem, these points can be distinguished by monitoring any indicator pins on your regulator IC. Some PMICs will have UVLO pins or shutdown indicator pins that will briefly toggle when certain protection mechanisms are triggered. Monitor these with a multimeter or a scope to determine if these problems are occurring.

The simplest set of pins to monitor power output is the power good (PG) pins. Some regulators, such as the dual rail output regulator shown below, have multiple PG pins as shown below.

Check your datasheets for shutdown indicator pins that you can use to monitor your regulator.

If a certain protection feature is tripping on a regulator or circuit, then there is no single solution that will solve the problem. You will need to read about that particular pin to get information as to why a particular protection feature is being triggered.

Filter Circuit Oscillation

An oscillating filter circuit can be excited by transient current draw on the input of a power regulator, or on the output during switching. When filter circuits are implemented, the intent is to reduce noise on the output, but a filter can add a new low-frequency pole into the regulator’s output transfer function. If this pole is excited by the switching element, then a driven oscillation can be observed on the output voltage. This oscillation occurs in addition to inductor ripple.

An oscillation being excited by a 50% duty cycle pulse stream on a switching node.

To solve the problem, you must either redesign the filter to be stable, or adjust the switching parameters so that the oscillation is not excited. The simpler path is to change the filter circuit. Typically, these filters are LC filters, and a few Ohms of damping on any capacitors can be enough to damp a pole in the filter transfer function. If the system will only be driving DC loads in operation, then a ferrite filter could also be used.

Switching Node Transients

Finally, there could be a transient response observed on the output due to switching, which tends to appear as a high-frequency underdamped oscillation. The oscillation will have two important characteristics:

• The underdamped oscillation has higher frequency than the PWM repetition rate
• The underdamped oscillation repeats at the PWM repetition rate
• The same oscillation can be observed at the switching node

The solution here is typically very simple: apply a snubber circuit across the switching node to ground. The purpose of the snubber circuit is to filter the oscillation but without creating new losses in the switching node.

Failure to Track Output Voltage Fluctuations

This will typically occur when the control loop is unable to respond fast enough to a transient that is excited on the power rail. The most common instance where this occurs in on VRMs that supply high current/low voltage to large digital processors. When the processor is running, large transients can be excited at high frequencies. As the control loop attempts to respond by adjusting the output voltage, it is unable to respond fast enough and could be driven into instability.

Control loop exhibiting an oscillation while attempting to compensate for high frequency transients on the output rail.

This could be identified with the comparator (COMP) output for the rail that is exhibiting the instability. It would need to be monitored on a scope alongside the power rail output at high frequencies. This can be difficult in digital systems as the output rail needs to be measured close to the device power pin, and this is usually buried under a BGA package. However, it is possible to simulate this as long as the VRM control loop’s transfer function is known.

Whenever you want to design a wide-input buck-boost converter from discrete components, or you need to simulate your wide input voltage regulator IC performance, use the complete set of simulation tools in PSpice from Cadence. PSpice users can access a powerful SPICE simulator as well as specialty design capabilities like model creation, graphing and analysis tools, and much more.

Subscribe to our newsletter for the latest updates. If you’re looking to learn more about how Cadence has the solution for you, talk to our team of experts.