|
While the infrastructure for plugging portable devices into the universal serial bus (USB) for datacom-with-power applications is fairly universal today, using the USB as a power source for direct powering or charging a battery isn't necessarily foolproof. You'll generally need over-voltage protection circuitry; here's what to consider in designing your discrete or IC-based circuit.
Characteristics and precautions
Downstream systems that you want to connect to can be powered in several ways. In the typical setup, PCs and peripheral devices plug into a connector that has a VBUS supply pin and D+ and D- data pins. The user should expect to see a VBUS voltage, as defined by the USB spec, of nominally 5 volts (maximum 5.25 volts). Usually the VBUS pin is connected to the supply input pin of a transceiver (sometimes through a LDO, which has a maximum rating of 6 volts) and/or the input pin of a charger when the VBUS supply is used for charging a lithium battery (maximum rating is 7 to 10 volts in most cases). One can also connect various systems to charge their internal lithium battery using a wall adapter option (Fig. 1). In this case, the VBUS pin and GND are connected; D+ and D- are shorted.
(Click on Image to Enlarge)
Figure 1: Basic USB configurations
There are several issues to contend with. System-wise, long cables can induce ringing due to serial inductance. The maximum ripple voltage depends on the mobile system's input capacitance and parasitic inductance, and electrical parasitics of one form or another generally create the most problems.
From the power standpoint, the output voltage from a typical adapter can far exceed the maximum ratings of various electronic components in the portable product. AC/DC power supplies often show poor line regulation. For instance, when an SMPS charger is used and there's loss of optocoupler feedback, the output voltage can soar to 20 volts. Avoid many of these types of problems by adding over-voltage protection (OVP) circuitry.
How to design
Generally, the USB port is rated for 100 mA (unconfigured mode) to 500 mA (configured mode). To save power, the USB generally enters a suspend mode when there is no data, and a bus-powered device such as a transceiver cannot draw more than 500 microamps (Fig. 2). A host can initiate a resume command or a remote wake-up to reactivate the host. The OVP circuitry is placed directly between the VBUS line and the transceiver (Fig. 2).
(Click on Image to Enlarge)
Figure 2: OVP considerations
For best results, the OVP core should use a p-MOS driver; the resulting current draw will generally be very low, as it needs to be. Ceramic caps are needed at the input and output. Note what happens in case 2 of Figure 3, where we have removed the output capacitor. As a result, the pass element stays open when a fast input transient appears on the OVP device input. Thus we see an overshoot, and the output spike can reach 10 volts, damaging equipment downstream. Place a 1-microfarad ceramic capacitor at the output to address the issue.
(Click on Image to Enlarge)
Figure 3: Why you need an output capacitor
Also consider over-voltage (OVLO) and under-voltage (UVLO) thresholds. These thresholds are determined by the OVP's internal comparators, which switch off the pass element when the circuit encounters an under-voltage or over-voltage condition. The OVLO level must be higher than VBUS (5.25 volts) plus the hysteresis of the comparator. Still, the maximum value of the OVLO parameter must be lower than the maximum rating of the first component of the system. Usually the OVLO is set at 5.675 volts to allow robust protection of the downstream system. The downstream system can tolerate 6 volts, and a ripple voltage (VUSB) up to 5.25 volts [1].
Don't neglect the thermal dissipation of the OVP circuitry; the internal MOSFET is often called on to deliver a bit of current. A p-MOSFET is recommended because it doesn't draw too much current in the typical application. A p-FET does have a higher Rdson than an n-FET, so heat transfer has to be optimized in order to avoid thermal damage. Depending upon the power required by application, a package with an exposed pad (such as the NCP360μDFN) is recommended. Consult the device's RθJA specification.
Several levels of protection
As previously implied, inrush current is a root cause of trouble. Include a soft-start sequence in the OVP circuit. This sequence brings the p-FET up slowly (Fig. 4).
(Click on Image to Enlarge)
Figure 4: Limit inrush-current effects with soft-start circuitry
As seen, a 4-ms soft-start function ensures there will be no voltage spike at the output even if VUSB or the output from the wall rises quickly (such as for hot-plugging applications).
The most critical characteristic of such protection is the circuit's ability to quickly detect any overvoltage condition and then open the internal FET. The turn-off time of the OVP circuitry is measured between the OVLO threshold crossing and the fall of Vout. Choose your components wisely. The NCP360, for example, combines a fast turn-off time (700 ns typical, 1.5 microseconds maximum) and low current draw (Fig. 5).
(Click on Image to Enlarge)
Figure 5: Minimize the OVP's turn-off time
To provide more protection, add over-current circuitry (OCP). By providing this additional block, you ensure the charge current or the load current cannot exceed the limit internally programmed. The current limit should be set at least at 550 mA to deal with the maximum transients specified under the USB spec. This OCP function is integrated in such devices as the NCP361.
References
1: Discrete Protection Solutions for Portable Charging, Bernard Remaury, ON Semiconductor.
About the author
Bernard Remaury is currently an applications and systems engineer with the Low Voltage Power Management Group. He specializes in battery management and protection for portable power-management ICs. Bernard previously worked for Motorola in the Wireless Group and was involved in the development of their Power Management Unit (PMU). Bernard can be reached at Bernard.remaury@onsemi.com
|
