Understand and Apply Supervisory ICs to Avoid Low-Voltage Power-up Glitch Headaches

Experienced engineers know that one of the riskiest times for a system is when power is applied. Depending on time constants and how smoothly and quickly the power rail comes up to nominal, the different ICs and parts of the system may start, lock up, or start in an incorrect mode as they attempt to work with each other. Adding to the challenge is that the timing and slew-related performance of the ICs on power-up can be a function of temperature, associated capacitors, mechanical stress, aging, and other factors.

The potential problem is aggravated as operating voltage rails drop to low single-digit values, reducing the amount of “slack” or headroom for functioning with the nominal rail value. All of these factors can lead to inconsistent startup performance and frustrating debug sessions.

For these reasons, analog IC vendors have devised specialized ICs that offer supervisory management features that eliminate the uncertainty and inconsistency of power-up. This article will define and characterize the glitch problem, and then show how it can be avoided through the addition of some small, specialized ICs from Analog Devices.

What is a glitch?

As with many engineering terms such as “buffer” or “programmable,” the word “glitch” has different meanings depending on the context. A glitch can be:

• A noise-induced spike on a signal or power line
• A sudden, brief drop in a power supply rail due to a load transient
• A microsecond period when both upper and lower MOSFETs in a bridge are inadvertently turned on simultaneously, as a result of different turn on/off times in their gate drivers (a very bad occurrence)
• A momentary indeterminate signal and race condition due to timing tolerances and differences between components

This article looks at the glitch that can occur during the “power-up” period when power is turned on, and the ICs are transitioning to their normal operating condition, especially in low-voltage systems. Such power-on glitches are especially frustrating because they can cause intermittent, hard-to-debug problems that have no apparent correlation or consistency. As the glitch-inducing conditions are often “on the edge,” their occurrence can vary with temperature, power-rail tolerance (while still within specification), individual component variations in a batch of the same device, and other hard-to-determine factors.

What is this glitch, and what is its source? Consider a system with a microcontroller and an associated supervisory/protection reset IC. The role of the latter IC is simple and focused: to maintain reliable system operation during power-up, power-down, and brownout conditions (Figure 1).

Figure 1: Understanding a glitch source begins with a look at a simple, typical arrangement of a microcontroller and its associated supervisory/protection reset IC, both powered by a battery and its regulator. (Image source: Analog Devices)

In a typical battery-powered application, the DC-DC converter generates the supply rail from a small, low-voltage battery. The supervisory IC is generally added between the DC-DC converter and the microcontroller to monitor the supply voltage and enable or disable the microcontroller.

The supervisory IC ensures reliable operation by accurately monitoring the system power supply and then asserting or de-asserting the microcontroller's enable input. The enabling and disabling of the microcontroller is managed via the supervisory IC’s reset output pin. This pin is typically an open-drain that is connected to a 10 kilohm (kΩ) pull-up resistor. The supervisory IC monitors the power supply voltage and asserts a reset when the input voltage falls below the reset threshold.

After the monitored voltage rises above the threshold voltage to its nominal value, the reset output remains asserted for a reset timeout period and then de-asserts. This allows the target microcontroller to leave the reset state and begin operating.

But what happens to the reset line before the supervisory IC turns on and pulls it low? The answer is found by looking closely at a typical power-up sequence (Figure 2). As supply rail V_CC begins to power up, both the microcontroller and the supervisory IC are off. As a consequence, the reset line is floating and the 10 kΩ pullup resistor causes its voltage to track V_CC.

Figure 2: In a typical power-up sequence, the reset line is floating, so its voltage tracks the rise in supply rail V_CC. (Image source: Analog Devices)

This voltage rise can be anywhere between 0.5 to 0.9 volts, potentially causing system instability. Once the supervisory IC turns on, the reset line is pulled down to prevent the microcontroller from inadvertently turning on. This glitch is common to all previous generations of supervisory ICs.

Low-voltage systems magnify the problem

This glitch scenario becomes a major concern with the trend toward low-power devices that are operating at ever-lower voltages. Consider systems with three logic levels of 3.3 volts, 2.5 volts, and 1.8 volts (Figure 3). For the 3.3-volt system, the output low-voltage threshold (Vol) and the input low-voltage threshold (Vil) are between 0.4 volts and 0.8 volts. If a glitch occurs at 0.9 volts, it would potentially cause the processor to become unstable by switching it off and on.

Figure 3: Logic levels have shrunk from 3.3 volts down to 1.8 volts, and so have associated voltage thresholds. (Image source: Analog Devices)

The situation for a nominal 1.8-volt system is more sensitive. Now, Vol and Vil are much lower at 0.45 volts and 0.63 volts. A 0.9 volt glitch in this system represents a larger percentage, giving it a higher potential for error.

How does this situation play out with the glitch impacting system operation? Consider a power supply voltage V_DD which ramps up slowly to 0.9 volts and “lingers” there for a short period of time (Figure 4). Although this voltage is not enough to turn on the supervisory IC, the microcontroller could still be enabled and running in an unstable state. Since the 0.9-volt value is in an indeterminant state, the glitch can be interpreted by the microcontroller RESET input as either a logic 1 or 0, which would erratically enable or disable it.

Figure 4: As the power supply voltage V_DD ramps up to 0.9 volts and lingers there, the microcontroller can be turned on and off erratically. (Image source: Analog Devices)

This causes the microcontroller to execute partial instructions or incomplete writes to memory, as just two examples of what might happen, likely causing system malfunction and possible catastrophic system behavior.

Solving the glitch problem

Overcoming this problem does not require a return to higher voltage rails, or demand complicated system-level architectures to eliminate its occurrence or minimize its impact. Instead, it requires a new generation of supervisory ICs that recognize the unique aspects of the problem and prevent glitches from forming, regardless of the voltage level during power-up or brown-out conditions.

Achieving this result requires a proprietary circuit and IC such as the MAX16162, a nanopower supply supervisor with glitch-free power-up. With this tiny IC—available in four-bump WLP and four-pin SOT23 packages—the reset output is held low whenever V_DD is lower than the threshold voltage, preventing a voltage glitch on the reset line. Once the voltage threshold is reached and the delay period is completed, the reset output de-asserts and enables the microcontroller (Figure 5).

Figure 5: The MAX16162 holds the reset output low whenever V_DD is lower than the threshold voltage, preventing a voltage glitch on the reset line. (Image source: Analog Devices)

Unlike conventional supervisory ICs that are unable to control the reset output state when V_CC is very low, the MAX16162 reset output is guaranteed to remain asserted until after a valid V_CC level is achieved.

The MAX16161 is a close sibling of the MAX16162 with nearly identical specifications, but with one functional difference and some redefining of pin assignments (Figure 6). It features a manual reset (MR) input that asserts a reset when it receives an appropriate input signal, which can be either active-low or active-high, depending on the option selected. In contrast, the MAX16162 has no MR input but instead has separate V_CC and V_IN pins, allowing threshold voltages as low as 0.6 volts.

Figure 6: The MAX16161 and MAX16162 are similar but with a small functional and pinout difference: the MAX16161 has an MR input that asserts a reset when it receives an appropriate input signal, while the MAX16162 has separate V_CC and V_IN pins. (Image source: Analog Devices)

Sequencer versus supervisor

Another pair of terms that have some overlap and ambiguity are supervisor and sequencer. A supervisor monitors a single power supply voltage and asserts/releases reset under defined circumstances. In contrast, a sequencer coordinates the relative resets and “power OK” assertions among two or more rails.

The MAX16161 and MAX16162 can be used as simple power supply sequencers (Figure 7). After the output voltage of the first regulator becomes valid, the MAX16161/MAX16162 insert a delay and generate the enable signal for the second regulator after the reset timeout period. Because the MAX16161/MAX16162 never de-assert reset until the supply voltage is correct, the controlled supply is never incorrectly enabled.

Figure 7: A circuit using the MAX16161 can be configured so the device not only ensures glitch-free power-up but also manages power-rail sequencing between two rails. (Image source: Analog Devices)

There are also many designs that have multiple rails and more complex sequencing needs. In these situations, the Analog Devices LTC2928 Multichannel Power Supply Sequencer and Supervisor offers a solution (Figure 8).

Figure 8: The LTC2928 power sequencer manages power-up and power-down sequencing among four independent rails, and enables user control over key parameters. (Image source: Analog Devices)

This four-channel cascadable power supply sequencer and high-accuracy supervisor allows designers to configure power-management sequencing thresholds, order, and timing using just a few external components. It ensures that power rails are enabled in the desired order. In addition to power-on sequencing, it can manage the complementary and often equally critical power-down sequencing.

The sequence outputs are used to control supply-enable pins or N-channel pass gates. Additional supervisory functions include undervoltage and overvoltage monitoring and reporting, as well as microprocessor reset generation. The type and source of faults are reported for diagnosis. Individual channel controls are available to exercise the enable outputs and supervisory functions independently. For systems with more than four rails, multiple LTC2928s can be easily connected to sequence an unlimited number of power supplies.

Conclusion

Glitches are present in every application, but they have not posed a significant issue for higher voltage applications which dominated until recently. Now, power supply voltages are moving lower, making system turn-on less reliable due to 0.9-volt glitches.

As shown, designers can improve reliability using newer supervisory ICs that offer glitch-free operation to provide the highest degree of system protection for low-power/low-voltage applications.

Recommended Reading

1. Analog Devices/Maxim Integrated Products, Design Solution 7550, “Is Your Application Protected from Glitches?”