Voltage Supervisors Pull Multiple Duties
Whether implemented using digital or analog technologies, voltage supervisors mitigate the conflict between performance and low power in multiple power rail devices.
The market drives digital signal processor (DSP), microcontroller and field programmable gate array (FPGA) manufacturers to continually increase clock frequencies for higher performance while, at the same time, also demanding lower power consumption. These two opposing criteria led to the development of multiple power rail devices.
A typical multiple power rail device has one voltage, termed an I/O voltage, to power input and output functions such as driving system busses, communicating with legacy logic devices or illuminating LEDs. The I/O voltage, usually 3.3V or 5.0V, is one of the higher voltages on the board. One or more lower voltages, termed the core voltage, are used to power the high-frequency logic within the device. A low-core voltage allows the logic to switch quickly, while reducing power consumption by lowering the switching losses in the transistors when compared against higher voltages.
Figure 1: Analog Voltage Supervisor
These high-performance, multi-voltage devices require tight tolerances on the core voltage in order for the internal logic to function correctly and properly execute software code. Additionally, with multi-voltage devices, there are typically power-up voltage sequencing requirements that must be used to avoid latching-up and damaging the device during power-up and power-down. Voltage supervisors are used with these high-performance devices to ensure correct power-up and code execution under varying power conditions.
In its simplest form, a voltage supervisor contains a voltage detection circuit that triggers a timer that drives a reset output signal. Either or both of the stages, the voltage detector or the timer, can be realized internally with either digital or analog electronics. Figure 1 shows a typical voltage supervisor circuit. The voltage detection circuit is a comparator with hysteresis. The voltage to be monitored and supervised is divided down, typically via an external resistor divider, and compared against a reference voltage. When power is applied to the supervisor, the /RESET output is pulled low. A retriggerable one-shot timer is triggered when the divided down SENSE voltage rises above the reference voltage. The timer holds the /RESET signal low for an additional fixed amount of time, providing a proper RESET signal to a DSP, microcontroller, FPGA or some other complex logic device.
Figure 2: Supervisor with voltage monitoring and watchdog timer
Some supervisors use a watchdog timer (WDT) to monitor code execution of a DSP or microprocessor, in addition to simply monitoring the power rail. A watchdog timer is a timing device that triggers a system reset, if the DSP or microprocessor fails to periodically reset the WDT. The block diagram in Figure 2 shows a supervisor with both a voltage monitor and watchdog timer. Typically, a DSP or microprocessor output pin is used to drive the supervisor’s watchdog input. The DSP must periodically change the output’s logic state, which produces a pulse train. The watchdog timer monitors this pulse train. If the DSP stops producing the pulse train, which occurs if the software is in an infinite loop or hung up in a process, then the watchdog timer times-out and issues a reset signal to restart the DSP and recover from any software error. The length of time from a missing pulse to a reset is determined by the oscillator and counter that comprise the WDT circuitry and depends on the device. WDT times are typically in the 0.5 to two second range.
Figure 3: Digital Voltage Supervisor
Figure 4: Example power up and down sequence
Figure 5: Eight-channel voltage sequencer
Figure 6: Supervisor used for switch debounce
Figure 7: Timing of switch debounce circuit.
Figure 8: WLED flash timer using voltage supervisor
Figure 9: Locked motor rotor detection using watchdog timer
Figure 10: Locked rotor timing
The same supervisor functions can be implemented digitally as well. Figure 3 shows the same voltage supervisor implemented using digital circuits. For the digital supervisor, the voltage to be monitored is still divided to provide a sense voltage (SENSE) that is within the input range of an analog-to-digital converter (ADC). The ADC then converts the SENSE voltage to digital signals that a microcontroller can read. An I/O pin on the microcontroller is used to drive the open drain MOSFET to provide a reset signal. A digital supervisor can provide features that an analog supervisor might not be able to provide.
The digital supervisor contains a microcontroller that makes it more functional and flexible than its analog counterparts. Some digital supervisors provide a serial communications channel that can communicate with other power management units or master microprocessors. Key parameters and configuration information can be stored in the supervisor, which then can be read or modified via the serial channel. This allows the reset thresholds to be modified via software without the need to change the external voltage divider. Additionally, the ADC allows the microcontroller to read a wide range of SENSE voltages where the analog supervisor voltage detection comparator only provides a signal that determines if the voltage is above or below the reference voltage. The wide range of SENSE voltage allows the digital supervisor to set variable under-voltage supervision, as well as variable over-voltage supervision, and brown-out detection. Moreover, most microcontrollers used in digital supervisors have a large number of I/O pins that allow for multiple voltage rails to be monitored by the same device. Digital voltage supervisors are available that supervise up to 12 different power rails, making them suitable for use in large, complex systems.
The voltage sequencer is a derivative of the voltage supervisor. A voltage sequencer is used to provide power-on and power-off sequencing of multiple voltage rails by driving the enable pins of the power supplies in the system. Figure 4 shows an example power up/down profile for a system with multiple voltage rails. The voltages rise and fall at defined times in relation to the other system voltages. The voltage sequencer is used to control and monitor the power-up and power-down timing.
The sequencer provides enable signals to power supplies instead of a system RESET signal. Sequencers are typically digital circuit- or microcontroller-based devices that can be programmed for the wide range of possibilities and combinations of power-up and power-down sequences unique to each individual system. The voltage sequencer enables power supplies based on time or on the status of other power supplies. After a power supply is enabled, a voltage sequencer monitors the supply for undervoltage or overvoltage events and takes appropriate actions, if those events occur. Appropriate actions can include disabling the monitored supply and other supplies, attempting to retry a faulty supply, sequencing-off all of the power supplies or taking other programmed actions. Voltage sequencers are available that can control up to eight power supplies with a single sequencing IC. The sequencer ICs can be cascaded to provide sequencing for more than eight power supplies, if needed. Figure 5 shows a typical configuration of an eight-channel voltage sequencer.
The basic functionality of a supervisor as a voltage level triggered one-shot means that it can be used in several applications outside of the intended use. For example, one common application of a voltage supervisor is to provide de-bounce of push button switches. When a normally open, pushbutton switch is depressed, the contacts do not simply make contact to complete the circuit. Instead, the mechanical portions bounce quickly, making and breaking the connection for several milliseconds before settling to their final mechanical position and completing the circuit. If the switch is used to trigger edge-triggered logic, there will be several false triggers of the logic during the switch bounce. Switch bounce also occurs when the switch is released and the circuit is opened.
Most supervisors have an input for connection to a manual reset switch. A switch is connected to a pin (typically labeled MR). When the switch is pressed, it activates a reset in the same manner as if the monitored voltage had fallen. A supervisor can be used to debounce a pushbutton switch by connecting it to this pin. The reset time needs to be longer than the make or break debounce time of the switch. Figure 6 shows an example of a voltage supervisor used as a switch debounce circuit. Here a pushbutton switch is connected to the MR pin of a TPS3808. The switch specifications state that there is a maximum of 8ms of make bounce, and 10ms of break bounce. An external 2.2nF capacitor sets the length of the reset time to 13ms, which is longer than either the make or break debounce. Figure 7 shows the expected waveforms of the pushbutton input and the reset output.
Figure 8 shows another example of a supervisor being used as a simple timer to drive a high-current white light LED (WLED) in a camera flash circuit. In this example, a high-power WLED is driven with 1A of current to produce a bright flash used for digital photography. However, the WLED can not continuously sustain this current level due to thermal limitations of its packaging. The supervisor is used to provide a 570ms flash of the LED when the ‘flash button’ is depressed.
The watchdog functions of supervisors also find uses outside of their intended applications. A watchdog is useful for detecting when a series of pulses stop. For example, a watchdog can monitor serial communication lines to determine when a serial stream of data or clocks stop, or for sensing a stalled or locked rotor of a motor. Figure 9 shows a block diagram of a watchdog timer being used to detect a locked motor rotor. In this example, a microprocessor is used to drive a motor. An optical slot tachometer is attached to the end of the motor shaft, which provides a pulse train as long as the motor is turning. The phototransistor of the tachometer drives the watchdog timer input, WDI, with the tachometers pulse train. If the motor rotor stops rotating, then the pulse train stops and, thus, the watchdog is not reset. After 1.4 seconds of no pulses from the tachometer, the TPS3128 drives its reset pin low. This active low signal drives an input pin to the microprocessor so that the microprocessor can adjust its drive output to compensate for the locked rotor condition. Figure 10 shows the timing diagram for a locked rotor condition.
Voltage supervisors and sequencers have been in the market for many years. Newer devices are implemented digitally, which allows for greater flexibility, increased number of monitored or sequenced channels, variable voltage thresholds and variable timing parameters. However, even with these newer high-performance digital supervisors, the older analog supervisor still provides functionality for a low number of monitored channels, or to provide alternate functions within a system such as switch debouncing or used as a timing element.
To learn more about voltage supervisors, download datasheets or other technical documents, visit:.
About the Author
Scot Lester is an Applications Engineer in the High-Performance Analog Group at Texas Instruments. He has over 19 years of hands-on board level design experience and specializes in low-power DC/DC converters for battery operated and portable applications. Scot is a member of TI’s Group Technical Staff and holds a BSEE from Montana State University, and an MBA from George Mason University, Virginia.
Texas Instruments Inc.