Comparison of 3 step-down converters to predict EMC issues

Step-down converters’ switch-node voltage waveform defines the electromagnetic compatibility (EMC) behavior for automotive CISPR 25 Class 5 measurements. The ringing frequency in the switch-node waveform is an important signal on the EMC receiver, where a higher ringing amplitude on the switch node often causes EMC issues. Understanding the switch-node waveform enables predicting the converter’s EMC characteristics as well as optimizing EMC filter design at an early design stage.

This article compares three automotive step-down converters to provide practical advice on using switch-node waveforms to predict EMC characteristics for automotive CISPR 25 Class 5 measurements. This is helpful to optimize EMC filter design and PCB layout to meet CISPR 25 Class 5 standards.

Switch-node measurements

Switch-node waveforms are used to compare the EMC characteristics among three automotive step-down converters. Figure 1 shows the switch-node measurement on an evaluation board using an active voltage probe.

Figure 1 Use an active voltage probe for the switch-node measurement on the evaluation board. Source: Monolithic Power Systems

The switch-node voltage waveform typically has a rising time and falling time between 700 ps and 2 ns. This requires a minimum oscilloscope bandwidth of about 1 GHz on the voltage probe tip, where the voltage can be measured with an active probe or a passive probe that has the necessary bandwidth.

For both variants, the ground connection to the PCB must be as short as possible to ensure that the measured ringing on the switch node does not include the additional ringing from the long probe ground connection.

Figure 2 shows the correct voltage probe tip position for the switch-node measurement on the evaluation board. Connect the GND tip as close as possible to the IC’s PGND pin and connect the probe input tip as close as possible to the IC’s switch-node pin. Solder the active probe tip with a 0.7-pF input capacitance directly to the component pads via removable gold-plated measuring tips.

Figure 2 Position the probe tip correctly for the switch-node measurement on the evaluation board. Source: Monolithic Power Systems

Histogram and time trend

Figure 3 shows a step-down converter’s switch-node voltage (yellow trace), fSW histogram (pink trace), and time trend (orange trace).

Figure 3 The dual frequency spread spectrum of the MPQ4371-AEC1 includes the switch-node voltage, fSW histogram, and time trend. Source: Monolithic Power Systems

The oscilloscope measures the switch-node voltage for each trigger event across a period of 400 µs and calculates the frequency of each switching cycle. Each calculated frequency is accumulated in the histogram. The total duration of this test is about 10 minutes. For the last trigger event, the measured frequencies are represented as time trend fSW vs. time.

The measured frequencies in Figure 3 verify the fSW vs. time relationship from the MPQ4371-AEC1 datasheet. The time trend waveform confirms the specified dual frequency spread spectrum modulation frequencies of 15 kHz and 120 kHz. By verifying proper IC operation, these frequencies provide an overview of the expected fSW values for CISPR 25 Class 5 measurements.

Voltage waveform

Step down converter’s switch-node voltage waveform is measured with an active probe. Figure 4 shows the rising and the falling edges of MPQ4371-AEC1, in which both waveforms are overlaid on the oscilloscope by an alternating rising and falling trigger. The rising edge has a rising time of 922 ps and a step response with a 273 MHz resonance frequency and a 3.2 V peak-to-peak voltage.

Figure 4 The switch-node voltage waveform for MPQ4371-AEC1 has rising and falling edges. Source: Monolithic Power Systems

The MPQ4371-AEC1 step-down converter’s Quiet-FET technology enables combining fast slewing edges without excessive ringing. Quiet-FET technology does not significantly degrade efficiency like a snubber or bootstrap resistor (RBST), and instead uses a minimum two-step sequential switching action to turn on the internal MOSFETs.

The resonance frequency is determined by the parasitic hot-loop inductances and capacitances. The equivalent hot-loop series inductances (ESL) are defined by the following:

ESL of the 100 nF, 0603-sized MLCC (about 800 pH)
ESL of the high-side MOSFET (HS-FET) and low-side MOSFET (LS-FET)
ESL of the package lead frame
ESL of the PCB traces between the MLCC and IC’s VIN and PGND pins (about 700 pH/mm)

The switch-node waveform can also be predicted using a simulation of the PCB hot-loop network.

Frequency domain

Figure 5 shows a fast Fourier transformation (FFT) of step-down converter’s switch-node waveform. The average fSW of 420 kHz is distributed between 384 kHz and 456 kHz (green markers) and corresponds to the measured histogram from Figure 3. The switch-node resonance frequency at 273 MHz is distributed between 250 MHz and 300 MHz (red markers) due to dual frequency spread spectrum modulation and corresponds to Figure 4.

Figure 5 A fast Fourier transformation is applied to the MPQ4371-AEC1’s switch-node waveform. Source: Monolithic Power Systems

Radiated emissions (RE) antenna for CISPR 25 Class 5

The vertical monopole, biconical, and log periodic antenna measurements in CISPR 25 Class 5 can be analyzed. Figure 6 shows the radiating switching inductance at peak CISPR 25 (blue) and average CISPR 25 (yellow), where the analyzer resolution bandwidth (RBW) = 9 kHz, fSW = 420 kHz, input voltage (VIN) = 13.5 V, output voltage (VOUT) = 3.3 V, and load current (ILOAD) = 2.5 A. The dual FSS modulation is helpful to maintain RE below the limits.

Figure 6 The vertical monopole antenna measurement of MPQ4371-AEC1 passes CISPR 25 Class 5. Source: Monolithic Power Systems

Figure 7 shows the radiating objects (for example, the harness or radiating traces on the PCB) at peak CISPR 25 (blue) and average CISPR 25 (yellow), where RBW = 120 kHz, fSW = 420 kHz, VIN = 13.5 V, VOUT = 3.3 V, and ILOAD = 2.5 A.

Figure 7 The biconical antenna measurement of MPQ4371-AEC1 passes CISPR 25 Class 5. Source: Monolithic Power Systems

Figure 8 shows the switch-node resonance frequencies between 250 MHz and 300 MHz (corresponding to Figure 4 and Figure 5) at peak CISPR 25 (blue) and average CISPR 25 (yellow), where RBW = 120 kHz, fSW = 420 kHz, VIN = 13.5 V, VOUT = 3.3 V, and ILOAD = 2.5 A. There is no RE that exceeds the 250 MHz to 300 MHz resonance frequency range.

Figure 8 The log periodic antenna measurement of the MPQ4371-AEC1 passes CISPR 25 Class 5. Source: Monolithic Power Systems

Figure 9 shows the 1.2 GHz switch-node resonance frequency within RE at peak CISPR 25 (blue), average CISPR 25 (yellow), and the noise level (gray), where RBW = 120 kHz, fSW = 2.2 MHz, VIN = 13.5 V, VOUT = 3.3 V, and ILOAD = 2.5 A.

Figure 9 The log periodic antenna measurement of the MPQ4323M-AEC1 step-down converter passes CISPR 25 Class 5. Source: Monolithic Power Systems

Switch-node waveform for MPQ4323M-AEC1

The MPQ4323M-AEC1’s integrated, 100 nF, hot-loop MLCCs reduce the internal parasitic inductances, which shifts the resonance frequency to higher values and reduces the resonance amplitude. Figure 10 shows an example of a fast slewing, switching converter combined with low internal parasitic inductances. This improves the switch-node waveform and reduces RE.

Figure 10 A fast-slewing switching converter combined with low parasitic inductances improves the switch-node waveform of the MPQ4323M-AEC1 step-down converter. Source: Monolithic Power Systems

Switch-node example on a 2-layer PCB

Figure 11 shows two different step-down converters soldered on the same 2-layer PCB. The left curve shows the MPQ4326-AEC1 with frequency spread spectrum modulation on a 2-layer PCB, with a switch-node resonance at 450 MHz. The right curve shows a step-down converter in a suboptimal set-up without FSS modulation and a 320 MHz resonance. The two converters are compared on the same PCB and with the same external components.

Figure 11 Two step-down converters are compared in a switch-node example on a 2-layer PCB. Source: Monolithic Power Systems

The step-down converter with the suboptimal set-up indicates undesirable resonance on the rising edge (red arrow), meaning there is a timing difference between the HS-FET and LS-FET. This resonance is caused by using a 2-layer PCB instead of a 4-layer PCB. Compared to a 4-layer PCB, a 2-layer PCB layout has higher parasitic inductances within the hot loop, which increases the resonance amplitude and changes the location of the switch-node resonance.

The increased amplitude is observed with both converters. In addition, the 2-layer PCB does not have the important solid ground layer directly under the top layer, resulting in a larger resonance amplitude and stronger RE.

FFT of step-down converters on a 2-layer PCB

Figure 12 shows the FFT of the switch-node voltage waveforms for the MPQ4326-AEC1 (with FSS modulation) and step-down converter with the suboptimal set-up (without FSS modulation) from Figure 11.

Figure 12 A fast Fourier transformation is applied to the switch-node voltage waveforms for the MPQ4326-AEC1 (with FSS modulation) and step-down converter with a suboptimal set-up (without FSS modulation). Source: Monolithic Power Systems

MPQ4326-AEC1 uses frequency spread spectrum modulation, while the step-down converter with the suboptimal set-up is set to a constant fSW. Typically, FSS modulation results in lower fundamentals and harmonics. Whether FSS modulation or a constant frequency is more advantageous depends on the requirements of the application. However, FFT shows the differences between the two methods.

MPQ4326-AEC1’s FFT shows the switch-node resonance at 450 MHz, and the step-down converter with the suboptimal set-up shows the switch-node resonance at 320 MHz. These switch-node resonance frequencies can be found in the CISPR 25 Class 5 measurements.

Understand switch-node waveform

This article analyzed the relationship between the switch-node voltage waveform and the frequency domain, using MPQ4323M-AEC1, MPQ4326-AEC1, and MPQ4371-AEC1 automotive step-down converters as examples. Understanding the switch-node waveform enables predicting PCB behavior for CISPR 25 Class 5 measurements. The measured resonance frequency shows up in RE measurements, enabling improved EMC filter design for suppressing the resonance frequency.

Furthermore, it is possible to assess expected frequency range interferences at an early stage by understanding the switch-node waveform. This helps find a suitable step-down converter according to the application specifications, shorten development times, and reduce costs by simplifying component selection for the EMC filter.

Ralf Ohmberger is a staff applications engineer at Monolithic Power Systems (MPS).

<!–
googletag.cmd.push(function() { googletag.display(‘div-gpt-ad-native’); });
–>

The post Comparison of 3 step-down converters to predict EMC issues appeared first on EDN.

Comparison of 3 step-down converters to predict EMC issues

Become a member

Become a member

Become a subscriber

Become a sponsor