Case Study: Texas InstrumentsSafety Considerations

-May 11, 2006

Customer Profile:  Handheld Equipment Manufacturer

The Challenge

In most battery powered applications the system voltages need to be generated with voltage regulation circuits, since battery voltage levels can vary over a wide range. Usually efficient voltage regulation is done by switchmode power supply circuits. Well designed power supply circuits should be able to deal with all possible error conditions specific for their configuration. Faults, for example, can be caused by input voltage dropping too low. Most often this is caused by the battery being discharged below its target end of discharge voltage, which can result in an ‘overload’ or ‘input undervoltage’ condition. Failures in manufacturing such as shorts caused by erratic solder joints, wrong assembly in the application circuit or even failure with application circuit parts, can cause shorts at the power supply output.

The Solution

All circuits require a certain level of supply voltage to operate their internal control circuitry. If the supply voltage drops below that minimum supply voltage, reliable operation no longer can be maintained. The behavior of the converter, as well as the output voltage level, cannot be predicted. For this reason, undervoltage lockout circuits are implemented in well designed power supply circuits. The undervoltage lockout also needs a certain hysteresis should the circuit automatically restart once the supply voltage returns. Without hysteresis the output impedance of the power source can also cause a kind of ‘ringing’ between lockout and startup. This also is an undefined state where the supply voltage for the system cannot be predicted.

Typical inductive DC/DC converters have one controlled active switch and, depending whether synchronous or non-synchronous, one controlled passive switch or just a rectifier. Generally, the current limit is implemented in the active switch which turns off the switch, if the current gets too high. This usually happens cycle-by-cycle. To avoid having noise erroneously turn off the switch, the current sense signal must be filtered or just suppressed at the time when the switch is turned on. This causes delays in turning off the switch. Usually, the main switch connects an inductor to a supply voltage, which means the rate at which the current increases is limited by the inductance. If the current limit is detected, the current still increases until the switch turns off. If enough inductance is still present, the system is protected by the inductor limiting the current increasing before the switch is finally turned off.

Unfortunately, at high currents inductors may saturate—which means their inductance decreases. If the inductor no longer can properly limit the rate of current increase, the main switch will become overloaded and may no longer be capable of turning off. This causes the main switch to break, and the power supply along with the output voltage and input power is no longer controlled.      

Limiting the switch current in a boost converter can be effective only when the output voltage is higher than the input voltage. Under this normal operating condition, the input current and input power are limited by the switch current limit, offering suitable protection. Should the input voltage get higher than the output voltage, a standard boost converter will operate with the rectifier always turned on. This has the effect that there are no means of limiting current in the system.

Figure 1: Normal Switching Operation of TPS61070 Close to Minimum Duty Cycle

Indeed, at high load and narrow duty cycle, the switch current limit may no longer function properly. However, the main switch needs to be turned on in order to detect the current. During this ‘on time’ the current is increasing. In the ‘off state’ there is no voltage across the inductor to decrease the current. This causes the switch current to increase even though the implemented current limit is working properly. Critical operating conditions occur when the output voltage comes close to the input voltage. In the TPS61070, this is addressed by forcing the device to generate a higher virtual output voltage and not using the synchronous rectifying switch anymore. Figure 1 shows standard switching operation close to minimum duty cycle. The blue curve shows the drain source voltage of the main switch and the yellow curve shows the output voltage. Figure 1 shows there is almost no difference between the high voltage level and the output voltage. 

Figure 2 shows the switching operation when input voltage is further increased. Both curves show the same signals as in Figure 1. Now it can be seen that the converter is running with a higher voltage difference between the high voltage level at the drain source voltage of the switch and the output voltage. This ensures the inductor is always discharged to avoid an uncontrolled increase of the inductor and input current.

Figure 2: Switching Operation of TPS61070 in Protected Mode at Minimum Duty Cycle

Should output voltage drop below the input voltage during output overload and output short circuit, other means of protection are necessary. How they can be implemented strongly depends on the basic architecture of the boost converter. If it uses a non-synchronous topology, the boost converter itself can not offer any protection. The rectifying diode is forward biased and allows any current to flow. There is no limiting element to protect the circuit. Short circuit current, or inrush current when applying input voltage, is not limited. If protection is required, this can be done only with additional supporting circuits.

Typically high inrush currents very often create a problem in battery powered applications. They can force the system to turn off due to input undervoltage caused by excessive voltage drop across the battery output impedance. Without using additional circuits, this also can be addressed by using a synchronous boost converter topology–but it only works if there is no diode in parallel to the rectifying switch. Using discrete field effect transistors (FETs), or integrated switches where the backgate cannot be controlled, also does not work. Having a completely controllable rectifying switch allows a precharge mode to be implemented, controlling the current flowing through the converter while the output voltage is lower than the input voltage. In such a precharge mode, the current is controlled like in a linear regulator. Depending on the input to output voltage difference, the possible precharge current or fault back characteristic can be controlled to increase the level of protection.

An example of how such a characteristic can look is shown in Figure 3, measured at the TPS61020. The curves show the maximum possible current in a short circuit condition and at increasing output voltage. With this, a system powered by the TPS61020 is always protected. 


 Figure 3: TPS61020 Short Circuit Current vs. Output Voltage

To view more related stories from this supplement:

Loading comments...

Write a Comment

To comment please Log In