Constant on-time control and charge pump combination for portable devices

As demand for longer battery life and the push for “greener” systems increases, the pressure to build power supplies that can respond efficiently from light to heavy loads is constantly mounting. However, improving the efficiency performance of a single phase synchronous buck DC-DC converter from light load to heavy load with wide input voltage has many challenges.

By Kit Nguyen, Sr. Application Engineer and John Lambert, Product Marketing Manager, International Rectifier Corp

Generally wide input voltage variation impacts the stability of controllers and power loss at high loads increase exponentially.  These losses translate into higher power dissipation, which may require complex thermal management to keep the portable device cool resulting in more components and larger board space.

However, employing a single-phase synchronous buck PWM controller with constant on-time control and a gate drive charge pump can minimize losses at both ends of the load. Constant on-time control timing diagram is shown in Figure 1. This type of control scheme, combined with diode emulation, helps to cut power losses when the output current is low.

Figure 1: Constant on-time timing diagram

Along with constant on-time control, IR’s IR3710M single-phase synchronous buck PWM controller also incorporates a gate drive charge pump capability to enable designers to maximize efficiency at higher output current levels. In other words, the PWM controller has features to deliver high efficiency over the entire output current range.  A schematic for wide input voltage 24A synchronous buck DC-DC converter built with this controller is shown in Figure 2.

Figure 2: Circuit schematic for a wide input 24A synchronous buck DC-DC converter using IR3710M

Unlike traditional voltage mode regulators, the constant on-time control method has a faster transient response time. Consequently, when the load current increases, the system requires less output capacitance and reduces the cost and size. Additionally if higher efficiency is desired at the higher current range, the IC features at gate charge pump option which increases the gate voltage of the low side and or the high side MOSFETs. This in-turn decreases the on-resistance of the MOSFET which lowers the conduction losses in the MOSFETs and consequently lowers the total power loss of the converter.

Understanding constant on-time

In the constant on-time control scheme, the circuit uses a portion of the output voltage ripple to compare with the reference voltage, V_REF.  The output voltage ripple, in fact, has a similar waveform as inductor current, and is equal to peak-to-peak inductor current multiplied by equivalent series resistor of the output capacitor bank. The IR3710M PWM controller turns on the upper gate (UGATE) when the voltage at the feedback input pin FB is lower than reference voltage. The feedback voltage is tapped from the output resistor divider and then compared to 0.5V internal reference voltage V_REF As shown in Figures 1 and 2.

The on-time duration for the upper gate is programmed by an external resistor R_FF from input voltage V_IN to feed forward pin FF. This on-time duration T_ON is calculated as depicted in the equations below:

T_ON  =   R_FF *1V * 20pF / V_IN     ……..(1)

R_FF  =  Vout /{1V * 20pF * Fsw}          …….(2)

Where Fsw is the switching frequency

From the above equations, we can see that this on-time is inversely proportional to the input voltage. When input voltage is higher, the on-time is smaller and is larger when input voltage is smaller. This method automates the control of the output voltage ripple as the input voltage changes. In our design, with R_FF  = 180kΩ, V_OUT = 1.1V and V_IN = 12V, typical on-time duration T_ON is 300ns with a 300kHz switching frequency

As a result, the voltage at the FB pin rises above the reference voltage in the slope which depends on the input and output voltages, inductance and programmed switching frequency. The IR3710M controller turns on the lower gate drive (LGATE) after the end of upper gate period with some dead-time between the gates to avoid the shoot through which causes a short from input voltage to ground.  As the result, the voltage at the FB pin decreases until it’s lower than the reference voltage, when the upper gate (UGATE) signal turns on again. 

The rate of decline for the FB voltage depends on output voltage, inductance and the switching frequency.  In the steady state, this process repeats for the next cycle. Unlike traditional architectures using voltage feedback, the IR3710M uses a comparator for this function instead of an error amplifier. Consequently, it eliminates the need for a compensation network, which can be time consuming and complex and calls for additional components.  

Figure 3: Waveform showing frequency boost from 300kHz to 1.5MHz with a 14A load step

Key Attributes

By comparison to traditional voltage mode, another advantage of this architecture is that it provides faster response time during the load transient as shown in Figure 3.  When the load increases with demand, the output capacitor bank supplies the initial charge to the load.  Initially, the output voltage drops as load increases.  The amount of voltage drop depends on the rate of load increase, equivalent series resistor and equivalent series inductor of output capacitor bank. The voltage at FB pin follows the output voltage drop below the reference voltage.  As a result, IR3710M forces the upper gate on to supply the additional charge to the load. Subsequently, the switching frequency increases from 300kHz to 1.5MHz. This scheme decreases the number of switching cycles to increase the inductor current in the shortest possible time, thereby allowing the output voltage to recover faster. This reduces the amount of output capacitance required and therefore saves space and saves system cost.   

In order to reduce power loss and MOSFETs temperature at heavy load conditions, the controller circuit (see Figure 2), uses external diodes (BAT54S) in combination with a ceramic capacitor to convert 5V and 3.3V power supplies into a 7V supply for the gate drive (PVCC).  The concept here is to boost the gate drive voltage of the upper and lower MOSFET gates. Using higher gate drive voltage lowers the on-resistance of the MOSFETs, thus, reducing their conduction losses. In fact, the majority of the power loss for the lower MOSFET comes from the conduction loss. The switching loss is negligible because the voltage across the MOSFET during the dead time is approximately 0.7V from the body diode before the lower MOSFET turns on.  Accordingly, less expensive smaller MOSFETs can be selected here, which in-turn reduces the system cost.

Now, the inductor current determines the two basic modes of operation—continuous and discontinuous. In the discontinuous mode (DCM) operation, defined as the lower peak of inductor current at 0 A, the controller turns off the charge pump feature, allowing the voltage at PVCC pin to drop below 5V supply. Consequently, it reduces the power loss of the driver. And, thereby, improves the efficiency during light load. Also, in this DCM mode configuration, IR3710M turns off the lower gate signal when the IC detects the inductor current at zero level.  Subsequently, it prevents the discharge current flowing from the output capacitor through the inductor, and the switching frequency reduces proportionally to the load. In essence, this feature also cuts conduction loss to improve system efficiency.

Figure 4: System power loss and efficiency comparison with and without the enhanced driver (charge pump) feature 

The impact of this enhanced driver feature on conduction loss and associated improvement in system efficiency is illustrated in Figure 4.  It compares the power loss and system efficiency with and without the enhanced gate drive feature and demonstrates that the efficiency increases by 2% at full-load with enhanced driver.    

In summary, besides cutting power consumption and enhancing overall system efficiency, the constant on-time control method also offers fast load transient response with no loop compensation required, hence, needing fewer external components to simplify the design and offers a solution that ultimately cuts project design time. The controller also provides a gate drive charge pump to improve efficiency at higher output current levels, thus, ensuring high conversion efficiency over the entire output current range. By improving efficiencies at both light and heavy loads, the IR3710M based synchronous buck converter solution maximizes the overall efficiency of the portable system, consequently, enabling the battery to perform for a longer durations. Higher efficiency keeps the MOSFETs cool and reduces or eliminates the need for a cooling system, which in-turn saves space and cost.

Figure 5: PCB design with IRF6720S and IRF6729M DirectFET MOSFETs

With built-in over current protection and under/over voltage shut down, the controller also enables key safety features. The same feature-rich controller IC is also utilized in the IR3870M SupIRBuck™ integrated point-of-load (POL) DC-DC voltage regulator that integrates IR’s high performance control ICs optimized with benchmark HEXFET® MOSFETs in a power QFN package.   

www.irf.com