Plug and play your way to balancing supercapacitors

BY ROBERT CHAO, President and CEO,
Advanced Linear Devices
www.aldinc.com

Supercapacitors are unique because they can work in conjunction with or as an alternative to batteries in applications needing many rapid charge/discharge cycles, and they are popular in applications such as short-term energy storage, regenerative braking, and static RAM memory backup.

But balancing a supercapacitor in a design involves more than connecting a few wires. When a supercapacitor has an operating voltage of 2.5–2.7 V, most applications use an array of devices connected in series for higher voltage.

Many supercapacitors have a leakage current that varies based on numerous factors, including:

• initial leakage value
• leakage variation with charging voltage, charging current, and temperature
• operating temperature range
• chemistry, material, and construction
• aging

Preventing this imbalance requires a balancing circuit that keeps the voltage across each supercapacitor within limits by automatically correcting the effects of changing leakage currents, all with minimal added leakage current or power consumption.

For the module designer, this represents an additional layer of complexity. A simple resistor-based design dissipates power and doesn’t respond to aging and temperature variations; an op amp-based design has higher performance but is complicated, wastes power, and consumes excessive board space. Developing a solution also requires layout expertise and analog or power design skills; the result may be added development time or even making the problem worse and adding unwanted leakage of its own. 

Overview of a new PCB

A new printed circuit board (PCB) offers a “plug-and-play” solution by combining all of the circuitry required to accomplish these tasks for any size of supercaps. Fig. 1 shows the board, which is designed to be used in conjunction with precision MOSFET arrays.

Each board measures only 0.6 x 1.6 in. and, when fully populated, can balance up to four supercapacitors connected in series; in addition to the unpopulated, populated boards are available with different combinations of MOSFETS to reach required voltages. For larger arrays, any number of boards can be connected in series, subject only to a maximum voltage per board of 15 V.

Fig. 1: Four supercapacitors plus external components.    

The board can accommodate supercapacitors from 0.1 F to 3,000 F and beyond. The average additional power dissipation due to the use of the boards is zero, which makes this method of supercapacitor balancing very energy-efficient. It is especially suited for low-loss energy harvesting and long-life battery-operated applications.

Fig. 2 shows the schematic of the board used with four supercapacitors C1–C4. The MOSFET is an array of four identical chips, which contains two devices in an identical configuration. Depending on the board selected, each supercapacitor can have a single MOSFET, or two devices in parallel, connected across it. Connecting two devices in parallel gives an equivalent MOSFET with twice the output current and twice the sensitivity to voltage change.

Fig. 2: Schematic of a fully populated board with three MOSFET arrays. The design ensures that the voltage across each supercap (VA – VB, for example) does not exceed the maximum allowed value.     

The “XX” in both part numbers denotes the voltage across the MOSFET that corresponds to a leakage current of 1 μA at 25°C, as discussed in the next section.

Each MOSFET is rated for a reverse bias diode current of up to 80 mA maximum. Any reverse-bias condition as a result of changing supercapacitor voltages, especially during fast supercapacitor discharge, could lead to some internal nodes becoming reverse-biased with surge current in excess of this limit.

The board has TO277 pads for adding external Schottky rectifiers (power diodes) to clamp such current transients. 

An optional resistor can also be connected across each pair of supercapacitors. Normally, this is left open, but if those two supercapacitor locations are unpopulated, a zero-ohm jumper in the corresponding location should be installed.  

Circuit operation

How does the circuit automatically balance leakage? A MOSFET can be considered as a three-terminal voltage-controlled resistor over a limited operating range, with the ON resistance R_DS(ON) an exponential function of the input voltage V_GS . When the gate and drain are connected, V_DS is also equal to V_GS and called the threshold voltage V_t . Fig. 3 shows the data for the quad MOSFET array.

Fig. 3: The chip's relative ON resistance vs. input voltage.    

The “XX” in the ALD part numbers denotes the threshold voltage V_t that gives an output current I_OUT of 1 µA at 25°C. For example, the ALD810025 quad array has V_t = 2.5 V, so I_OUT = 1µA at 2.5 V and 25°C. This corresponds to an R_DS(ON) of 2.5 MΩ. Table 1 shows the change in I_OUT as V_t varies.

Table 1: Equivalent ON resistance at different input voltages and output currents for the ALD810025 at TA = 25°C.

When a MOSFET is connected across a supercapacitor in an array, the exponential relationship between V_t and R_DS(ON) means that any small increase in the supercapacitor voltage caused by leakage current from another supercapacitor will cause a large decrease in the R_DS(ON) of that MOSFET. This will increase I_DS(ON) and reduce the voltage. As the voltage decreases, I_DS(ON) will also decrease.

Depending on the PCB part number selected, each supercapacitor can have a single MOSFET or two devices in parallel connected across it. Connecting two devices in parallel gives an equivalent MOSFET with twice the output current and twice the sensitivity to voltage change.

The leakage current balancing mechanism is fully automatic and works for supercapacitor arrays with different leakage current profiles. In a stack with many supercapacitors in series, multiple MOSFET arrays function to balance the various leakage currents.

For example, if leakage current I_C1 is greater than I_C2 in the four-capacitor array in Fig. 2 , then both M1s (in U1 and U2) are turned off and the M2 pair becomes the active circuit element to balance C1, the supercapacitor with the greatest leakage.

Without the auto-balancing circuit, the voltage across C2 can keep rising until it destroys the part. The MOSFET balancing solution contributes virtually no additional current drain and adjusts for temperature, time, and other environmental changes that may contribute to out-of-balance conditions.