Active Adapter Protects Against Transients

For the PDF version of this article, click here.

Many battery-powered products, such as cell phones, digital camcorders and portable DVD players, have the capability to run from an external power supply, such as a car adapter or an ac-dc converter. This poses some challenges for the product designer and requires additional study to understand the transient behavior when the product is plugged and unplugged from ac power adapters, car adapters and other power sources. This article describes from a system level what considerations and factors must be understood and analyzed by the designer to avoid creating unexpected damage from transient power conditions.

Causes of Transients

Transients can and do occur when an adapter is hot-plugged into its battery and system load. This is due to parasitic inductance in the adapter cabling, contact fingers, printed-circuit board traces in the current path, the bulk output capacitance found in the adapter and the input capacitance of the battery and system load.

If the adapter is plugged in first, its output capacitor (C_O) charges up to the open-circuit voltage of the adapter (V_ADP). When the adapter is connected to the load, the input capacitor (C_IN), which is at 0 V, draws current through the cabling inductance. This energy transfer creates a transient condition until steady-state equilibrium is reached, where the voltage at the adapter is approximately equal to the input capacitor minus any resistive losses in the cable and contacts.

The adapter cabling exacerbates these transients because the cabling inductances store energy, which is then transferred into C_IN. This raises the voltage across C_IN further above V_ADP. The voltage across C_IN eventually peaks, falls back to V_ADP, and may ring for a while, depending on the circuit. This voltage rise can easily exceed the maximum ratings of the components in the battery and system load.

Transients in Action

Fig. 1 shows how a typical adapter cable with C_IN values of 1-µF, 10-µF and 100-µF can introduce voltage spikes well above the output voltage of the adapter. Simulations are shown for a 5-V adapter. The adapter was modeled as a simple voltage source with a 2-Ω series impedance. The output capacitor for the 5-V adapter is assumed to be 220 µF. The cable impedance was modeled as an equivalent series resistance (ESR) of 0.75Ω and an equivalent series inductance (ESL) of 0.8 µH.

The tradeoff between C_IN and transient behavior shows that a smaller C_IN allows voltage transients to be twice as large as the adapter voltage (Fig. 1a). Larger capacitance lowers these transients but increases in-rush current, which can cause a droop in V_ADP (Fig. 1b). One could add a series resistor or increase the cabling resistance to limit in-rush current, but the cost would be added power dissipation and a constant voltage drop after the adapter. Zener diodes can limit the voltage transient, but they are not very accurate, so it is difficult to set the protection threshold to be higher than V_ADP but still low enough to protect the system load.

Fig. 2 shows the NCP346 overvoltage supervisory IC for adapter input protection placed between the V_ADP and V_IN to limit in-rush current while protecting the system load from high-voltage transients and other high-voltage faults, such as an incorrect or damaged adapter. The low-side driver of the NCP346's OUT pin has a resistance of typically 100 kΩ. This limits the turn-on gate current of the high-side p-channel MOSFET such that the transistor will turn on in a controlled manner. This reduces the ringing that can occur from the sudden turn-off of excessive in-rush current.

The waveforms of Fig. 3 show a typical off-the-shelf adapter hot-plugged into a bulk input capacitance with and without the protection circuit. (Ch1 shows measurements with the IC installed; Ch2 shows waveforms without that circuit.) The measured inductance and resistance of the cable was 1.3 µH and 0.1 Ω. The output capacitance of the adapter was 330 µF. The optional resistor divider was used per the NCP346 datasheet to adjust the overvoltage threshold to 6 V for the protected circuit. The input capacitance at VIN was 10 µF, which is a common input capacitance (C₁) for a dc-dc converter that could be drawing power from an adapter input. For this configuration, the unprotected transient voltage reached 6.84 V. In the protected case, the transient is not detectable in the oscilloscope trace.

Fig. 4 shows that adding the NCP346 circuit (from Fig. 2) both limited the ringing at the capacitance node and added active overvoltage protection. In Fig. 4, an incorrect voltage of 12 V was applied and the V_IN-protected input never experienced the fault (Ch1 in the figure). The proposed circuit effectively protected its system from the fault. On the otherhand, the unprotected input (Ch2) experienced a voltage transient of more than 16 V with ringing.

Protection-Circuit Design

The first step in designing the protection circuit requires selecting the proper overvoltage lockout (OVLO) point. The IC comes in two versions with a nominal overvoltage threshold of 4.45 V and 5.5 V. If levels are too low, they can be adjusted upward with a resistor divider between the V_CC, IN and GND pins. However, the IN input is not an ideal high-impedance node, so some care must be taken in adjusting the OVLO to still be accurate:

Making R₂>>R_IN, the effects of R_IN are minimized. A design procedure for this step is as follows. The equation to select the adjusted overvoltage threshold is:

By following these steps, one can adjust the OVLO and still maintain a good tolerance:

Use typical R_IN and V_TH values from the electrical specifications provided on the NCP346 datasheet.
Minimize the R_IN effect by selecting R₁<< R_IN.
Let X=R_IN/R₁=100 to minimize RIN effects while keeping the resistor ladder as high an impedance as possible.
Identify the required nominal overvoltage threshold based on the adapter's voltage tolerance. One should design the input supply such that its maximum supply voltage in normal operation is less than the minimum desired overvoltage threshold.
Calculate nominal R₁ and R₂ from the nominal values and pick standard resistors close to these values:
Use min/max electrical specification data from the NCP346 datasheet and resistor tolerances to determine the OVLO tolerance.

For instance, the circuit used in Figs. 3 and 4 had a 6-V OVLO. The NCP346 used has an OVLO of 5.5 V +/- 200 mV. From Eq. 2, using 1% resistors, R₁=549 Ω and R₂=6650 Ω. This equates to an OVLO of 6.01 V with a tolerance of ±290 mV, so the resistors only added 90 mV of inaccuracy. Tighter tolerance resistors improve this even further.

Next, choose an appropriate MOSFET and Schottky diode. The MOSFET and Schottky must first withstand the overvoltage fault, so the worst-case devices should have voltage ratings that match the NCP346's 30-V rating. Since the R_DSON and V_F of a MOSFET and Schottky, respectively, depend on the maximum voltage they can withstand, it is worthwhile to note if the protected application actually needs the full 30 V of protection.

For a 30-V system, the MOSFET should have a drain-source break-down voltage (V_DSS) of 30 V. The gate-source breakdown should be at least 20 V if the product could use a vehicle car charger/adapter. This would provide immunity to a steady-state failure of the car charger applying the adapter voltage directly to the portable product. A 12-V V_GSS should be adequate for most other low-voltage adapters.

The same sort of analysis should be applied when selecting a Schottky. Schottkys come in ratings of 10 V, 20 V, 30 V and higher. While the Schottky will mostly clamp itself when V_IN rises, there could be a finite time before the Schottky conducts where it sees the entire OVLO condition. So, to maintain the safety of the system, a 20-V or 30-V diode should be used.

The definite possibility of these faults occurring cannot be overstressed. End users have more products that require different adapters than ever before. The faults illustrated in this article could have been from an end user mistakenly plugging a portable DVD player's adapter into an MP3 player's charger jack. Or, a user could have plugged an adapter for an older cell phone into a newer cell phone that requires a lower adapter voltage. Or, there could be an aftermarket universal adapter that is not actually compatible with a PDA. These examples illustrate how the solution described in this article protects unsuspecting end users from damaging their products.