Fault protection is a critical function of all power supply controllers. Nearly all applications require some kind of over-load protection. For peak current-mode controllers this is easily accomplished by limiting the maximum peak current. In a discontinuous flyback, placing a limit on the peak current ultimately limits the power that the supply can draw from the input source. However, limiting the input power does not limit the output current of a supply. When the input power is held constant in an overload fault, as the output voltage drops, the output current rises (P=V*I). In the event of a short circuit, this can place unacceptably high losses in output rectifiers or system power distribution. With a little creativity and a few extra components, this article shows you how a simple peak current limit can be modified to turn the power supply into a constant current source, rather than a constant power source.
Fig. 1 shows the ideal output voltage versus current for both constant power and current limits. In both cases, the overload fault protection is set to occur at 120 percent of the maximum rated load. In a system with power limit, the output current rises as the inverse of the voltage as the load is increased. In a real system, the flyback controller with power limit may shut off at some point due to loss of bias voltage on the controller. By comparison, the system with current limit immediately shuts down once the overload threshold has been crossed. Current limit can be accomplished by directly sensing the load current on the secondary side of the isolation boundary. However, doing so requires significantly more circuitry, degrades efficiency, and is usually cost prohibitive.
Fig. 2 displays the schematic of a 5V/5W discontinuous flyback power supply used in a charger for mobile devices. In our example, we used the UCC28C44 controller, which is typical of most economy peak-current mode controllers and implements a power limit function. In a discontinuous flyback, neglecting the effect of efficiency, the power delivered to the load (P) is given by Equation 1.
Because both the transformer inductance (L) and switching frequency (f) are fixed, the output voltage (VOUT) is regulated by controlling the peak primary current (IPK). As the output current (IOUT) is increased, the voltage begins to decrease, but the feedback loop demands a higher peak current to maintain voltage regulation.
Inside the flyback converter, the feedback voltage on pin 1 (COMP) is compared to the peak current, which is sensed by R15 and filtered by R13 and C12. A separate over-current comparator terminates a pulse, if the current sense voltage ever reaches 1V. This limitation of peak current is how the power limit is realized in most pulse-width modulation (PWM) controllers. With the power held constant, Equation 1 can be rearranged as shown in Equation 2. In this equation, it is easy to see mathematically how the output current is inversely proportional to the output voltage during power limit.
Some controllers also contain a second comparator that is tripped by peak currents that are higher than the first level comparator. This second level comparator triggers a complete shutdown of the controller and initiates a restart cycle. This extra level of protection is designed to catch catastrophic failures within the power supply itself, for example, a shorted transformer winding or shorted output diode. But most situations that involve shorted loads typically never cross this threshold.
Fig. 3 shows the output and bias voltages versus load current for the circuit in Fig. 2. The output V-I characteristics very closely follow the ideal case shown in Fig. 1. Power limit is incepted when the load current reaches approximately 1.3A. As the load increases, the output voltage begins to fall. Because the bias voltage is a reflection of the output voltage, it also begins to fall. The PWM controller shuts off when the bias voltage drops below the turn-off level of 9V.
In this example, although the peak current limit is engaged when the load exceeds 1.3A, the load current can be more than twice the rated load before the converter shuts off. This may be unacceptable in some applications. Instead, a more square-shaped V-I curve is desirable. This can be accomplished very easily by leveraging the fact that the bias voltage decreases as the load increases past the power limit point. By adding a few components, the decreasing bias voltage can be used to fold back the switching frequency during power limit. By doing so, the switching is forced to be proportional to the output voltage, as shown in Equation 3. Substituting Equation 3 into Equation 2 reveals that, theoretically, the output current is no longer dependent on the output voltage during power limit, see Equation 4.
The components added to create this improved current limit are highlighted in the schematic shown in Fig. 4. The switching frequency of the flyback converter is set by R10, R8, and C11, which program the internal oscillator. An internal 5V source charges C11 through R10 and R8. As the bias voltage falls, the resistor divider of R7 and R11 turns on Q1 and over-rides the internal 5V source, decreasing the switching frequency. The bias diode (D4) must now be a dual, series-connected diode, so that R7 and R11 do not divert current from the controller during startup. The values of R7 and R11 are selected so that Q1 is off during normal operation, and only turns on when the bias voltage has dropped below approximately 12V.
The results of adding these components are shown in Fig. 5. Just as before, both the output and bias voltages begin to drop as the supply enters power limit. Once the bias voltage has dropped low enough to begin to turn on Q1, any further increase in load current causes the switching frequency to decrease, which in turn decreases the available power to the load. This accelerates the over current shut down process. Notice that there is still some correlation between the output current and output voltage. This is due to coupling of the bias winding inside the transformer and limited gain of Q1. Despite these imperfections, the V-I characteristics are sharply improved with the added circuit. In fact, now the supply will not provide more than 1.5A into a faulted load.
In summary, power supplies that provide power limit protection can still source large amounts of current into an over-loaded output. As shown here, implementing a precise current limit can be done easily and inexpensively just by adding a few components around the primary-side controller. Although presented for a flyback converter, this scheme can also reduce the current tailing seen in buck-derived converters.
References
Download a datasheet for the UCC28C44.
Learn more about power solutions from TI: www.ti.com/power-ca.