If your circuit design has requirements to be tolerant to certain faults and to report their occurrence, you’ll need to test the fault detection and protection features of your design during the test and bring-up process. To make this testing easier, consider building fault-injection fixtures and reporting capability into the prototype printed circuit board.
Overvoltage and undervoltage monitoring
You can build circuitry into your design to force an overvoltage or undervoltage event for a voltage supervisor to detect. For a power supply with a digital interface, you may need only to expose the communication pins. For a power supply with no digital interface, you would need to expose the feedback pin.
A simple way to modulate the output voltage is to inject current into the feedback node, thereby artificially depressing the output voltage. It is much easier to design this circuit into the PCB than adding it to the built PCB. Figure 1 is an example of this type of circuit design for the TPS628502-Q1 automotive step-down converter from Texas Instruments (TI) that provides 1.8 V and up to 2 A from a 5 V input. Applying a pulse-width modulation (PWM) signal at TP80 varies the output voltage between approximately 1.1 V and 2.5 V. Designing this circuit such that the output voltage when no current is injected sets the output voltage above the designed 1.8 V enables overvoltage fault detection testing.
Figure 1 An output voltage modulation circuit example. Source: Texas Instruments
For more details on selecting component values for this circuit, see the “Analog voltage input” section of the Analog Design Journal article, “Methods of Output-Voltage Adjustment for DC/DC Converters.”
This test fixture makes it possible to verify the overvoltage and undervoltage detection capability of voltage supervisors such as TI’s TPS389006-Q1. There is a GUI available for developing circuits with the TPS389006-Q1 that plots the monitored voltages and any detected faults. Figure 2 shows an undervoltage and subsequent overvoltage event created with the circuit in Figure 1. Channel 2 on the supervisor monitors the 1.8-V rail. The undervoltage threshold was set to 1.5 V and the overvoltage threshold set to 2.1 V. Modulating the duty cycle of the PWM signal allowed for artificial modulation of the output of the TPS528502-Q1, both below and above the thresholds.
Figure 2 Undervoltage and overvoltage fault detection in the TPS389006-Q1 GUI. Source: Texas Instruments
In addition to the fine-grained control that the circuit in Figure 1 allows, it may also be useful to test the power-supply response to a short to the supply or ground. Bare headers on the board itself make it easy to test momentary or continuous shorts. Onboard indicators such as an LED connected to the power supply’s power good pin provide an easy indicator of a successful short injection and subsequent recovery. Figure 3 illustrates an example of these test fixtures, once again using the TPS628502-Q1 as the example.
Figure 3 Short to VIN and ground test fixtures with power good indicator. Source: Texas Instruments
The circuitry facilitates injecting the faults. Then the TPS389006-Q1 supervisor GUI indicates a short to 5V, a recovery, a short to ground, and a final recovery has occurred (Figure 4). The fault register remains set until cleared.
Figure 4 A 1.8-V short to ground fault detection in the TPS389006-Q1 GUI. Source: Texas Instruments
Another common fault that you may want to detect is incorrect sequencing—rails not coming up in the correct order, rails not coming up at all, or rails coming up with incorrect timing are all examples of incorrect sequencing. Just as with overvoltage and undervoltage fault injection, there are two methods to induce sequencing faults: digitally and manually.
Some sequencers, such as the TPS39700-Q1, are run time-configurable. During development, you can use a GUI to configure the sequence order and timing using I2C. Configuring a purposeful mismatch in the sequence defined in the TPS39700C-Q1 and the expected sequence in the supervisor makes it possible to test the fault detection capability of the system.
If the supervisor is not configurable through software, then breaking the connection between the sequencer and rails will modify the sequence, which has the added benefit of allowing you to enable and disable a device independent of the sequence. Figure 5 shows one method of device enable control in a 5-V to 1.8-V/2-A TPS628502-Q1 power supply. If no jumper is installed on J52, the enable pin will pull down. I advise designing selection headers in such a way that that input pins are placed in a known state if no jumper is installed.
Figure 5 Enable pin circuit to enable sequence testing. Source: Texas Instruments
Defining a correct sequence of a 3.3 V to 1.8 V to 1.2 V and then removing the J52 jumper results in a sequencer fault reported by the TPS389006-Q1 supervisor, shown in Figure 6. MON3 monitors the 1.8-V rail.
Figure 6 The TPS389006-Q1 sequence fault detection. Source: Texas Instruments
It’s important to consider how to perform fault testing at all stages of the circuit design process. Incorporating fault injection and indication into the circuit schematic and layout can help streamline subsequent testing. I hope these examples inspire you to help make your own design verification easier.
Matt Griessler is a systems engineer for the Power Supply Design Services team at Texas Instruments.