Many cost and engineering improvements are realized by integrating the current pass element and currentsense function into a hot-plug controller integrated circuit (IC). Controlling the passelement power is the safest and fastest way to charge the load capacitance. Pass-element protection is maintained while boosting efficiency and reliability.
A circuit controlling current to a load usually consists of three components—an IC controller, a MOSFET switch through which current passes, and a current-sense element. A simple but powerful concept is to integrate these two parts into one monolithic IC.
At first glance, it may seem that the primary benefits are reducing circuit board space and some of the cost of the external components. This is all true. But if that integrated product also manages the load by controlling the power dissipated in its MOSFET, it then becomes a much more robust solution— and ideal for high-reliability systems.
Many high-reliability systems must stay powered on through equipment upgrades, failures, or repairs. A current-management solution, sometimes known as a “hot swap” or “hot plug” controller, keeps systems operating in two ways. First, it controls power turn-on to avoid excessive inrush current when applying power to a module with discharged capacitors. Second, it’s a fast-acting circuit breaker on an overcurrent fault. In both cases, module components are protected, and the local failure doesn’t cause the system voltage to drop out of specification and disrupt other operating modules on the same power bus.
SAFE OPERATING AREA
Some power systems use dv/dt control, which limits inrush current by slowly ramping the voltage to the output. The issue here is that the MOSFET must be oversized because operating at constant current may take the MOSFET out of its safe operating area (SOA).
Remember, the MOSFET SOA data is published for the device operating at 25°C. It must be de-rated for real-world conditions. For dv/dt control, changing the load capacitance from the designed value changes the inrush current. Increasing load capacitance increases the inrush current.
Other systems use di/dt controllers, which hold a constantcurrent ramp to charge the load capacitance. The current required to charge the load capacitor is unchanged with capacitance, but generally has a longer charge time than the dv/dt method.
Methods that charge the load capacitance by limiting the power dissipation in the MOSFET will protect the MOSFET and charge the load capacitance in the shortest possible time within the maximum current limit selected.
Figure 1 defines the general SOA curve, which shows the operating limits of a typical MOSFET. Sections 2 and 4 are the current and voltage limits of the MOSFET. The RDS(ON) for the MOSFET determines the Section 1 limit. Section 3 is the thermal limit, which is important because excessive heat is the most common reason for MOSFET destruction. Not surprisingly, the plot of the thermal limit is also a plot of constant power.
The typical MOSFET datasheet SOA curve shows the family of constant power curves for a fixed time (Fig. 2). The red line demonstrates the problem with current-limiting controllers. The MOSFET must be over-rated from its normal operating point to survive startup and fault when operating at point C.
MOSFET PROTECTI ON
In contrast, the blue line sets the constant-power operating point for reliable operation of the MOSFET under fault conditions. At startup, the controller operates at point B. By charging the capacitor and decreasing the voltage across the MOSFET, the control returns to current limit when the product of current demand and the voltage across the MOSFET is less than the power limit of the MOSFET.
The TPS2420 and TPS2421 from Texas Instruments incorporate this power-limiting feature. These power controllers with integrated MOSFET control the power turn-on and overcurrent events by limiting the power dissipated in the package. The power limiting is a controlled way to manage the load and guarantee that the MOSFET will always be operated within its SOA. (TPS2420 and TPS2421 datasheets and technical documents are available at www.ti.com/tps2420-ca and www.ti.com/tps2421-ca.)
A scope waveform shows the voltage and current starting up into an 800-µF, 60- load (Fig. 3). The MOSFET power dissipation is constant as calculated by the scope in the yellow math trace. The voltage across the MOSFET is the constant 12-V input (not shown) minus VOUT. The MOSFET power is the product of the MOSFET voltage and IOUT. Maximum power is fixed at 5 W. As VOUT increases, the MOSFET voltage decreases and IOUT can increase to maintain constant power dissipation.
Continue on Page 2
Another protection for the integrated MOSFET is over-temperature shutdown. In the case of the TPS2420 and TPS2421, the recommended operating junction temperature is less than 125ºC. Since excessive heat destroys MOSFETs, all controls are suspended and the part turns off when the junction temperature exceeds 160ºC.
High junction temperatures occur when operating at high ambient temperatures or repeated high-stress events. After a cool-down hysteresis of 10ºC, the part resumes its normal operating function. An external MOSFET has no shutdown protection and the heating problem continues until the component is destroyed.
The low 30-m RDS(ON) of the internal MOSFET helps minimize power loss in the package. To help dissipate 5 W of power in a tiny quad flat no-lead (QFN) package or small-outline IC (SOIC), each package contains a PowerPad, which is an IC package. In PowerPad packages, the substrate of the chip comes in contact with the metal leadframe, which is exposed at the package bottom side. Compared to standard packages, the PowerPad offers greater power dissipation, and when soldered to a circuit board plane, dissipation improves even further. The plane can be connected by vias to other planes to form a large heatsinking area.
To read and control current in any power-management scheme, the sense element is a necessary component of hot-swap circuits. A discrete resistor can dissipate 0.25 W in controllers operating at equal power to the integrated controller. Sense resistors are usually oversized to avoid errors due to self-heating in components with lower power ratings.
Most current-limiting hot-swap controllers have a current limit threshold of 50 mV across the sense resistor. This source voltage isn’t driving the load. Rather, it becomes wasted power that requires cooling. FETs are used for integrated current sensing. When the current-sensing function is integral to the controller, no power losses are incurred.
Highly integrated parts minimize the engineering effort. Just select the current trip point, the maximum current limit, and the time that the maximum current is allowed to start discharged capacitors. These parameters are easily defined and then programmed with one or two common resistors and a capacitor. In contrast, component selection for controllers with external components requires a careful MOSFET selection based on safely de-rated parameters, calculation of the timing capacitors to determine optimum load capacitor charge time, and specification of currentsense resistors.
Hot-Pluggi ng tHose Drives
One of the important applications for the power manager is the hot plug of a disk drive in a computer system or redundant array of independent drives (RAID). The 12-V specification for a disk drive is approximately 1-A operating current and 2-A typical spin-up. To safely power the drive:
• Estimate the surge current requirement. Select a 2.5-A setting for IFAULT to allow some margin for the operating current and satisfy the start current requirements.
• Calculate RISET:
RISET = 200,000 / 2.5 = 80 kO, use 80.6 kO (1)
• Since ISET satisfies the spin-up current, the timer can be set for the additional load to charge the capacitor. Estimate approximately 20 ms. Calculate the timer capacitance:
CT = 20 × 10–3 / 38.9 × 103 = 0.514 × 10–6 F, use 560 nF (2)
These are the values of RISET and CT that are used in Figure 4.
Layout engineering for a discrete solution is a complicated matter. The sense resistors need to be wired in the flow of power to the load. Thus, they are wired in a Kelvin connection to the controller sense pins. When the sense resistor shares the power input pin, care must be taken to power the controller package properly. The controller VCC pin isn’t simply connected to an internal power pin.
Like the sense resistor, the MOSFET is in the proper flow of power to the load. Most layout concerns are avoided with integrated components. Simply place each with sufficient current-carrying capability at VIN and VOUT and place a ground pour under the IC to form a heatsink to the PowerPad package. Locate support components that set the current limit and the fault time close to the IC pins.
Continue on Page 3
Figure 4 shows the simple wiring and RC support components. Fewer components and reduced complexity help simplify the overall test engineering effort for integrated controllers.
Low-voltage powermanagement applications requiring less than 5 A can benefit by integrating the MOSFET and currentsensing function into the controller. Such a combination will deliver lower cost, higher performance, smaller size, simpler engineering, and fewer components to test.
The Power Good status signal (PG output) indicates the internal MOSFET is fully enhanced. PG connects to downstream dc-dc converters to turn them on when the output is at the proper level. By delaying turn-on of the resistive load until the MOSFET is fully enhanced, more power is available to charge bulk capacitance at startup, and heat dissipation decreases in the package.
The Fault status signal (FLT output) indicates an overcurrent condition. The FLT signal returns data to the power controller or system front end. Power managers either latch off or retry on fault. A latching controller can only be turned on again by cycling the enable or the main power. A retry controller also turns off on fault, but tries to restart periodically. When the fault is cleared, the controller automatically powers the load.