Dan Tooth co-authored this technical article
Your new design needs to fit twice as much into half the space and cost nothing – sound familiar? You selected the smallest point-of-load regulator and generated the tightest layout you could with the most cost-effective passive components. So far so good. But then you look at the output ripple on your critical rails and it’s not what you expected. What’s going on?
Let’s start by understanding what makes up the output ripple on a buck DC/DC regulator. It is a composite waveform. Traditionally only the three dominant elements shown in Figure 1 have been considered:
Figure 1: Typical output ripple waveforms
However, what you measure has spikes on the edges and a higher square-wave content that changes polarity when you reverse the inductor shown in Figure 2:
Figure 2: Measured output ripple
What has caused these undesirable components? And more importantly, what can you do about it?
When you selected your inductor, the self-resonant frequency (SRF) was above your regulator switching frequency, so all was good. Let’s re-look at that – the inductor has an SRF because it has a parallel parasitic capacitance. Applying the fast edge of the switching voltage to the parasitic capacitance generates a large current spike through the capacitor, which in turn generates a large voltage spike across the ESL of the output capacitor (see Equation 1):
To reduce this spike:
Let’s say that you selected a cost-effective, unshielded inductor. The magnetic field from an unshielded (or resin-shielded inductor) can spread beyond the physical body of the component. The simulation plots in Figure 3 show the field for an unshielded open drum inductor and a fully shielded molded inductor.
Figure 3: Magnetic field for unshielded drum and shielded molded inductors (Source: Courtesy of Coilcraft)
· Select a shielded inductor to reduce the leakage flux that generates this coupling. If you’re using unshielded or semi-shielded inductors, selecting an inductor that is larger in the x-y dimension but has a lower profile will reduce the airgap height and hence the fringing flux.
· Reduce the output capacitance ESL as described above.
· Don’t position the output capacitors and tracking directly next to the inductor, where the field is highest. Where space is critical, consider placing the inductor on the opposite side of the board to the rest of the regulator circuit in a clamshell construction. This moves the output capacitors away from the plane of the inductor where the magnetic field is strongest.
· Read the Analog Design Journal article, “Select inductors for buck converters to get optimum efficiency and reliability.”
· Download the application reports, “Output Ripple Voltage for Buck Switching Regulator” and “Space Optimized, ‘Clam-Shell’ Layout for Step-Down DC/DC Converters.”
What do you mean by "reverse the inductor"?
I envisioned it as desoldering the inductor, rotating it 180 degrees and then soldering it back down, but I've always been under the impression that the inductors I use are symmetrical (DFE2520 for example: www.murata.com/.../productdetail
Yes, I did mean desoldering, spinning 180deg and resoldering. As an ideal component inductors are symmetrical, but for a real world, classic bobbin wound inductor rotating has two potential effects:
Firstly it is good practice to connect the start of the winding to the SW node and the end of the winding to the output. The SW end of the coil has the full ground to Vin voltage swing and can radiate EMI. If this is the start of the winding, it is buried in the centre of the inductor with the rest of the winding around it. This has "self shielding" effect and can reduce the EMI.
In this article however, we were discussing the effect of the output ripple. This is due to the magnetic coupling from the inductor to the ESL of the output capacitors which acts like a transformer. Reversing the inductor reverses the direction of current flow in the coil and reverses the field direction and hence the polarity of this coupled signal - like switching the dot on a transformer winding.
Both these effects are seen when using lower cost, unshielded or semi-shielded inductors. The simple solution is to use a fully shielded inductor but in cost sensitive applications this may not be an option. I notice the DFE2520 inductor you cited is a chip inductor - the structure is slightly different but the effects are the same.
Thanks Jim for demystifying what typically plagues my DC/DC power designs. Given the parts are generally well chosen, how much would you think the PCB layout plays out on ripple over selecting even more expensive (better) parts? In other words, is it worth putting more engineering time into parts vs. PCB layout.
The answer to your question is .... it depends! In the article above I have tried to suggest a few different steps to mitigate each undesired ripple component. The step which is best in each case depends on the limitations within which you are working. Are you trying to achieve the smallest possible solution, or the most cost effective solution, or the most efficient solution? For example, a fully shielded inductor will allow a very small layout but will cost more; a lower cost, non-shielded inductor can be fine if the output capacitors are further away but this is a larger solution; or slowing down the switching edges can help but this sacrifices a little efficiency. The double sided layout described in this apps note (www.ti.com/.../slva630a.pdf) can certainly help in some cases. Generally, I would say that time spent optimising the layout of your design is never wasted. It will allow you to achieve the full potential of the devices and components chosen, but it can't really compensate for a poor component.
All content and materials on this site are provided "as is". TI and its respective suppliers and providers of content make no representations about the suitability of these materials for any purpose and disclaim all warranties and conditions with regard to these materials, including but not limited to all implied warranties and conditions of merchantability, fitness for a particular purpose, title and non-infringement of any third party intellectual property right. No license, either express or implied, by estoppel or otherwise, is granted by TI. Use of the information on this site may require a license from a third party, or a license from TI.
TI is a global semiconductor design and manufacturing company. Innovate with 100,000+ analog ICs andembedded processors, along with software, tools and the industry’s largest sales/support staff.