Designing a Buck Converter
Designing a buck converter isn’t easy. Most users want a box that works—taking in one DC voltage and delivering another. This box can take various forms. It can be a step-down to generate a lower voltage or step-up to generate a higher voltage. There are also plenty of special options, like step-up/down, flyback and single ended primary inductor converter (sepic), which is a DC-DC converter that allows the output voltage to be greater than, less than, or equal to the input voltage. For a system that will run on AC power, the first AC-to-DC block will probably create the highest DC voltage level needed by the system. Therefore, the most widely used devices are step-down converters.
The step-down converter using a switching regulator delivers the highest efficiency of converter types. High efficiency means less energy is lost in the conversion, simplifying thermal management.
Figure 1 shows the basics of a type of step-down switching regulator, a synchronous buck converter. The term synchronous buck indicates that a MOSFET is used as the lower switch. Comparatively, a standard buck regulator has a Shottky diode as a lower switch. The main benefit of a synchronous buck regulator compared with a standard buck regulator is better efficiency due to a lower voltage drop of the MOSFET versus a diode.
The timing information for the lower and upper MOSFETs is provided by a pulse-width modulation (PWM) controller. The input to the controller is a voltage fed back from the output. This loop allows the buck converter to regulate its output in response to load changes. The output of the PWM block is a digital signal toggling up and down at the switching frequency. The signal drives the MOSFET pair. The duty cycle of this signal determines the percentage of the time that the input is directly connected to the output. Thus, the output voltage is the product of input voltage and this duty cycle.
Choosing the IC
The control loop noted above allows the buck converter to maintain a steady output voltage. That loop can be implemented in a number of ways. The simplest converters use either voltage or current feedback. These converters are rugged, straightforward and cost-effective. As buck converters began to be used in a variety of applications, a weakness was found. Consider the power circuitry for a graphics card. As the video content changes, so does the load on the buck converter. The system can handle a wide range of changes, but the efficiency rapidly degraded for light load conditions. If efficiency is a concern, it’s time for a better buck converter solution.
One such improvement is called hysteretic control. An example is the Intersil ISL62871. The efficiency versus load is presented in Figure 2. These converters are designed for worst-case conditions, so light load is not a permanent situation. These DC:DC converters are better at coping with changes in load variations without drastically affecting system efficiency.
Choosing the Switching Frequency
Although it is sometimes fixed for a device, it is still worth discussing switching frequency. The chief trade-off is efficiency. In the simplest terms, the MOSFETs have a certain turn-on and turn-off time. As the frequency increases, the transitional time increases as a percentage of the total time. The result: efficiency is reduced. So if efficiency is the most important design goal, consider lowering the switching frequency. If the efficiency of the system is adequate, then allow a higher switching frequency. The higher frequency will allow the use of smaller external passive components, namely the output inductor and capacitors.
An alternative challenge is creating a discrete solution, which requires about 40 components, a complex task that requires significant additional effort. In the voltage-mode buck converter design, the external components and their parasitics play a large role in the performance of the system. These will be detailed as we address each component.
With this particular buck converter, we must select five additional components, the input capacitance, the output capacitance, the output inductor, and the upper and lower MOSFETs. The output inductor is selected to meet the output ripple requirements and to minimize the PWM’s response time to a transient load. The lower bound of possible inductor values is set by the ripple requirement. Before running out to find the smallest (and possibly cheapest) inductor that will work for you, remember that they are not ideal. Real inductors have a saturation level. That saturation level must be higher than the peak current in the system to create a successful design. Experienced designers also know that the inductance isn’t constant versus current. In fact, the value of inductance drops as you pull more current through the component. Check the inductor datasheet to ensure that your chosen value is sufficient with the peak current in your system. It seems that erring on the larger side might be the best inductance choice. Care is required, though. Larger values of inductance will reduce output ripple, but they will also limit the slew rate. Eventually, a large inductance will limit response time to a load transient. So, in selecting the inductor there is a clear trade-off between a quieter output due to lower peak-to-peak ripple or needing the system to respond quickly to a transient event.
The input capacitance is responsible for sourcing the AC component of the input current flowing into the upper MOSFET. Therefore, their RMS current capacity must be sufficient to handle the AC component of the current drawn by that upper MOSFET. It is common to use a mix of input bypass capacitors at this point. For quality and low temperature coefficient, ceramic capacitors can decouple the high frequency components. Bulk capacitors supply the lower frequency RMS current, which is tied to the duty cycle. (More RMS current when the system is operating further from 50% duty cycle.) The bulk capacitance can be several multi-layer ceramic capacitors. In lower cost applications, however, several electrolytic capacitors are typically used in parallel. In surface mount designs, solid tantalum capacitors may be chosen for the bulk capacitance, but be careful to note the capacitor’s surge current rating. (Surge currents are common at start-up.) When choosing any capacitor in the buck converter system, look for small equivalent series inductance (ESL), small equivalent series resistance (ESR) and finally the total capacitance required. As always, optimize the choice based on budget. There is one final note in regard to capacitor voltage ratings. To minimize hard-to-find failures, choose capacitors with ratings 1.2 to 1.3 times greater than the input voltage, that is, the voltage across them.
The output capacitor must filter the output and supply current to the load during a transient event. Interestingly, the equivalent series resistance (ESR) and voltage rating have more effect on the choice of capacitor than the actual capacitance value. Notice that the peak-to-peak current ripple from our inductor is transformed into peak-to-peak voltage ripple by the ESR of the output capacitor. Since the system probably has a limit on output voltage ripple, it is important to choose a capacitance (or set of parallel capacitors) that will minimize the ESR. Of course, capacitors must have sufficient voltage rating. With this combination of requirements, approach the capacitor tables from vendors to find a suitable solution. One final caution, pay extra attention to the ESR data; it might not be given in the table at the same frequency as your switching frequency. Check the component datasheet for adjusted values of ESR.
The MOSFETs are typically chosen for Rds(on), total gate charge and thermal management requirements. Review several manufacturers’ datasheets. Choose something like the Infineon BSC050N03LS with 35nC of gate charge and Rds(on) of 5 milliohms for the upper MOSFET. Complement that with the Rds(on) of 1.6 milliohms for the lower MOSFET (BSC016).
Closing the Loop
As discussed earlier, the output is fed back to the input. This connection creates a compensation loop. There are various types of compensations, such as Type I, Type II, and Type III. Type I compensation is a single-pole solution, Type II is a two-pole solution with one zero and Type III is a three-pole solution with two zeroes. Each type increases in component count from the previous one, yet also allows for greater flexibility in design. For performance, set the bandwidth of this loop to be approximately a quarter of the switching frequency. The higher the crossover frequency of the loop with regard to the actual switching frequency yields a faster loop response. In addition, make sure the phase margin is greater than 30 degrees and less than 180 degrees, a typical stability criterion.
The design process is similar with a hysteretic buck converter in place of a voltage-mode converter. Luckily, the high quality hysteretic-mode control helps overshadow the parasitics of the external components. Otherwise, the process is similar.
To summarize, here is the process of designing a buck converter. After choosing a controller IC, select accompanying external components. There are different parameters that are important for each selection. Once the MOSFETs, output inductor, input and output capacitors are chosen, finish with compensation.
Plenty of work goes into designing a good buck converter -- and more integrated versions are now available. Some designs have integrated MOSFETs. Some designs integrate the compensation. A select few have integrated the output inductor as well. One such offering is Intersil’s ISL8201M. All that is needed is a resistor to set the output voltage, an input capacitor and an output capacitor. That is good news for busy system designers.