All trademarks are the property of their respective owners.
A boost converter circuit such as TPS61022 is shown in Figure 1-1. With alternate turning on and off of the two integrated MOSFET, the inductor stores energy and then released to VOUT which is higher than the input voltage. The average value of VOUT is set by the FB pin and R1, R2 resistors.
The ideal operating waveform at heavy loading condition are shown in Figure 1-2. In the image:
In this equation:
However, one may observe much larger output voltage ripple than the calculation result in real circuits. This application details the root cause of the observation and proposes solution to solve the problem.
The Equation 1 assumes that the output capacitor is ideal and capacitor ripple is perfectly measured by the voltage probe. However, the actual ripple waveform would be related to the setting of oscilloscope bandwidth and voltage probe grounding method.
The bandwidth of a Tektronix oscilloscope can be set to 20 MHz or full bandwidth, which could be 500 MHz or higher depending on the oscilloscope model. The voltage probe grounding loop could be as in Figure 2-1. The left probe has long grounding wire and the test point is not closed to the capacitor. The right probe has very short grounding wire and the test point is on the terminals of the capacitor.
Taking TPS61022 as an example, the output ripple waveform with different setting are shown from Figure 2-2 to Figure 2-4. From the waveform, three conclusions can be derived :
The voltage spike shown in Figure 2-3 is caused by the ESR (effective series resistance) and ESL (effective series inductance) of the output capacitor. The Figure 3-1 shows the impedance of a 0805-package ceramic capacitor over frequency. Point A is the resonant frequency of the ESL and the capacitor.
The impedance of the capacitor is equal to its ESR which is approximately 3.5mΩ. The impedance of the capacitor increases linearly once the frequency is higher than 10 MHz. This is behavior of the ESL. From the impedance of point B and point C, the ESL is estimated to be 0.27nH.
The voltage crossing the ESR is in proportion to the current through capacitor. A ceramic capacitor has only several milliohm resistances and normally there will be several ceramic capacitors in parallel. Thus, this voltage ripple caused by ESR is small and can be neglected.
The voltage crossing the ESL is in proportional to its di/dt, which could be very large. Because the switching frequency of a boost converter always tends to be as large as possible to reduce the solution size. The di/dt will also increase with the switching frequency to reduce the power loss in the IC.
Considering the ESL and the parasitic inductance of the PCB trace and IC package, the schematic of the boost converter power stage is shown in Figure 3-2 where LPAR1 is the sum of pin-out and PCB parasitic inductor and LPAR2 is the parasitic inductor of the GND pin;
When Q1 turns off and SW node voltage is higher than VOUT, the inductor current starts to transit from Q1 to Q2. During the current transition period, the equivalent circuit is shown in Figure 3-3, where
The Figure 3-3 can be further simplified to a LC circuit with initial inductor current IL1. The inductor value is the sum of all the parasitics inductor in the circuit. The capacitor is CQ1. So the voltage spike of the ESL is defined by Equation 2. The voltage spike is in proportinal to inductor current.
And the oscillation frequency is defined by Equation 3.
When Q1 turns on, the inductor current firstly transits from Q2 to Q1, then voltage Q1 quickly declines to zero. The simplified schematic during this period is shown in Figure 10, where CQ2 represents the parasitic capacitance of Q2 when it turns off.
The maximum voltage spike at ESL is defined by Equation 4. The circuit can also be simplified to a LC circuit. The VESL voltage level is in proportion to the boost output voltage, and the oscillation frequency is the same as Equation 3 because the CQ2 and CQ1 is similar.
Taking TPS61022EVM as example, CQ1 and CQ2 are approximately 0.4 nF. LPAR1 is closed to 0.3 nH; LPAR2 is approximately 0.05 nH; ESL is approximately 0.09 nF as there are three capacitors in parallel. Thus, the LC oscillation frequency is up to 300 MHz.
At VIN=3.6V, VOUT=5V, IOUT=3A condition, the IL1 is about 4.4 A. The VESL spike at both Q1 turning on and turning off condition are approximately 0.75 V. however, the actual voltage spike from Figure 2-3 is lower than this calculation result, especially at Q1 turning-on condition. The main factor leading to the gap is the turning on/off time of the Q1. The turning on/off time of Q1 is at 5ns level. The greatly impacts the voltage spike caused by a 300-MHz LC ringing. It is hard to have an equation to calculate the voltage spike if considering the turning on/off time. A better method is simulation with PSPICE-FOR-TI or other tools.