Innovations to reduce design time/complexity improving efficiency
Electronic designers often met with the challenge of meeting the most stringent EMI specification in a very short period. This is particularly true for today when design cycles are decreasing and fast ramp to market is seen as on the key differentiating factor.
To add to the complexity of EMI optimized design, the complete system (be it for automotive, industrial, or consumer applications) is tested mostly at the end of the design cycle i.e. when the PCB and BOM are all finalized. Hence any last-minute changes in PCB or BOM are not only time consuming but very costly. The low-frequency emission(less than 30MHz), as discussed in the earlier segment, is generally easier to deal with but requires a bulkier passive filter. But what about high-frequency emission? Could it be dealt with just the changes in the PCB layout? The answer is probably not if the high-frequency emission is a result of switch-mode power supply device packages, its slew rate, its high switching frequency, etc.
If the high-frequency emission is related to the device chosen for the switch mode power supply in the system, can a filter be used to filter out the same? The answer again is probably not. And why because due to PCB and EMI filter parasitic, most EMI filters are rendered ineffective beyond 50MHz or so. So the only solution left is to change the device but since BOM and design are already approved, this option means more delay in the project, piling up extra cost, and possible impact on the reliability of the design if the new design doesn’t go through thorough the testing.
Are all switching mode power supplies created Equal as far as high-frequency emission is concerned? To answer this question, it is important to understand the root cause of high-frequency emission in switch-mode power design.
What causes High-Frequency Emission?
There are three main causes of high-Frequency emission in switch-mode power supply:
Slew rate on switch waveform:
Power designers continue to choose higher and higher switching frequency for their switch mode power supply design due to size advantage. For example, a few years ago the preferred switching frequency for automotive power solutions used to be sub AM bands i.e. 400KHz or so. But the general trend now is to choose above AM band switching frequency and hence is the need to shorten the rise and fall times of the switch node to reduce the switching losses. However, a switch node with very short rise and fall times maintains high energy content even at high frequencies close to its 100th harmonic, as shown in Figure below. The switch-node waveform of converters that offer short rise and fall times will cause much higher emission.
Figure. EMI plots of square waveforms with different rise and fall times
Also, the high-frequency emission of an SMPS, switching at a higher frequency, is higher when compared to an SMPS switching at a lower frequency. Here is the comparison of conducted EMI performance of 400KHz switcher vs 2.1MHz switcher (same device and layout) in 30MHz to 108MHz range. A 400KHz switcher has much better attenuation at a higher frequency. Why is it so?
The answer lies in the energy of the harmonics of a switcher falling in a high-frequency domain.
For example, if the emission of both the switchers is measured at 42MHz, the energy of 20th harmonics of 2.1MHz switcher will be higher than 105th harmonics of 400KHz switcher. The same effect can be seen in the figure showing FFT of a 2MHz and 400KHz square waveform.
The emission of a 2MHz converter is higher than 400KHz by 15dB or so.
In conclusion, higher switching frequency SMPS has higher emission at higher frequency bands and faster rise and fall time further increases the emission. When choosing a device in higher switching frequency application, look for additional features in the device that can help mitigate EMI effects.
Switch node ringing:
Switch-node ringing is caused by the parasitic inductance in the power loop resonating with the overall capacitance present in the switch node of the DC/DC converter and emissions from ringing may range close to 100 MHz. The switch node ringing can parasitically couple as near E field to the board, environment, the power source, and the return line. This switch node coupling is the main source of common-mode noise in the SMPS. At such higher frequencies, the bulky and costly common mode filter at the input can provide some attenuation but different techniques are needed to avoid the emissions.
The Package, PCB, and layout could be the source of parasitic inductance in the power loop of the switch-mode power supply. For example in the case of Buck converter, the loop from high side FET’s drain to the high-frequency bypass Capacitor to low Side FET source(Ldrain + L source) could have parasitic inductances and that contributes to the Switch node ringing.
Here is the typical example of the effect of the input loop on switch mode power supply.
The FET package inductance and capacitance, PCB inductance, and capacitance as well as device’s package inductance, etc are the other sources of Switch node ringing. Conventionally the switch node ringing is addressed by adding a lossy snubber which results in inefficient power design.
How to solve the high-frequency EMI issue and reduce time/complexity:
Hotrod package Technology: One primary technique is to inherently minimize the power-loop inductance. Newer products from TI like the LM53635-Q1, LMS3655-Q1, LMR33630-Q1, LM61460-Q1, LM62480-Q1, etc move away from wire-bond packages to lead frame-based packages (Figure 10), which help lower the power-loop inductance and in turn reduce the switch-node ringing. In the lead frame-based packages, the silicon die is flipped and placed on the lead frame directly which minimizes the parasitic inductance caused by wire Bonds on SW, VIN, and GND pins.
Figure 10. A flip-chip on a lead frame quad flat no-lead (QFN) package helps reduce the power-loop inductance
Here are the benefits of using hotrod packaged devices :
- Lower inductance => dramatically lower switch node ringing (see above image)
- Lower Rds_on and hence higher efficiency
- Smaller solution size
One other very important benefit of hotrod packaged devices is that it facilitates parallel input path pinouts i.e. the layout arrangement of the input capacitors of the DC/DC converter. By optimizing the pinout of the DC/DC converter so that there is symmetry in the layout of the input capacitors, the opposing magnetic fields generated by the input power loops are contained within the symmetric loops, thereby minimizing the emissions to systems nearby. Hence the parallel input path further minimizes the high-frequency EMI emission particularly in the most stringent FM band as shown in the figure below.
Enhanced Hotrod package Technology: The Enhanced Hotrod Package technology offers all the EMI benefits of Hotrod package and has an added advantage of even lower SW node capacitance resulting in much lower ringing. The RLC parasitic on VIN/GND pins is also lower in Enhanced hotrod package devices compared to Hotrod package devices.
Figure from SNVA935
LM60440-Q1 comes in an enhanced hotrod package and the details of pinout and board layout are shown in Figure. The enhanced Hotrod package not only improves efficiency but also includes a footprint that has a large DAP at the center of the package. The DAP felicitates better thermal dissipation through PCB and thereby reduces the rise in Junction temp by more than 15% when compared to the previous generation. On top of it, lower RLC parasitics on VIN, GND, and SW pin also result in better efficiency.
Integrated Input Bypass Capacitor: As mentioned in the switch node ringing segment, the large input loop results in higher emissions at higher frequency bands due to increased switch node ringing. Integrating high-frequency decoupling capacitors inside the device’s package results in minimization of input loop parasitic and thereby reduces EMI. This technique is used in LMQ61460-Q1, as shown in Figure.
Figure from Power Density White paper.
Figure 6 shows the improvements in the radiated EMI performance of LMQ62440 when compared to the same device with no integration of the Bypass capacitor. Both (pin to pin replacement devices) were placed in identical conditions on identical boards for the EMI comparison. The results show an improvement of 9db in the most stringent TV band (200MHz – 230MHz) in the radiated EMI set up which helps in meeting the EMI standards without the need for any additional components on the board.
True Slew rate control: Despite these techniques, there may be designs where the high-frequency (60- to 250-MHz) EMI may still not be within specified standards. One common way to mitigate and improve margin in the high-frequency range is to use a resistor in series with the boot capacitor of the switching converter. This helps slow down the switching edges leading to lower EMI but comes with a penalty of lower efficiency and potential Undervoltage lockout issues with the boot voltage.
Switching converters like the LM61440/LM62440-Q1 are designed to allow a resistor to select the strength of the high side FET’s driver during turn on. As shown in Figure 12, the current drawn through the REBOOT pin, the dotted loop, is magnified and drawn through from CBOOT, the dashed line. This current is then used to turn on the high-side power MOSFET without putting a resistor in the series path. With REBOOT short-circuited to CBOOT, the rise time is rapid; switch-node harmonics will not roll off until above 150 MHz. If CBOOT and REBOOT are connected through 700 Ω, the slewing time increases to 10 ns typical when converting 13.5 V to 5 V. This slow rise time allows the energy in switch-node harmonics to roll-off near 50 MHz under most conditions.
Figure 12. True slew-rate control implementation in the LM62440(SNVSB70D Figure 15)
Figure 13. Reduction in Switch node ringing using true slew rate control
Effective modeling capabilities to reduce cycle times and validate IP techniques
Modeling of any circuit is an important way of evaluating the performance of the design in the early stages and thereby plays a critical role in reducing the design cycle time. EMI modeling is a complex process that involves time-domain circuit analysis as well as frequency-domain Electro-Magnetic simulations of the PCB. The intent here is to help Power designers in reducing design iterations to meet EMI standards faster.
Improve low-frequency EMI designs using WEBENCH simulation – Input filter designer tool helps the switching power supply designer to automatically design a proper input filter to mitigate lower frequency (<10MHz) conducted EMI noise for a given compliance standard like CISPR 22 and CISPR25. The tool optimizes the filter sizes at the same time makes sure that the design complies with the given EMI standard. It also ensures filter stability and converter loop stability while designing the filter. This online tool supports over 100 TI power parts.
One common mistake in designing an Input EMI filter is undamped input filter inductors which can have negatively affect the overall stability of the design. The Webench designer performs impedance analysis on the input filter and SMPS and suggests the dampening component required for ensuring the overall stability of the design.
Figure: Input EMI suggestion with impedance analysis on WEBENCH online tool
Conducted and Radiated EMI results published in datasheets: The evaluation module of switch-mode power supplies are tested against industrial and automotive EMI standards and results are published in the datasheet to help the power designers understand the EMI performance of the device.