DC-DC verification MP1584 SY8205

**Disclaimer: I am an amateur in power supply design; this article is for reference only.**





This is an MP1584 and MP2451 power supply I designed several years ago. At that time, I had no concept of ripple. Recently, during testing, I found that under heavy load (above 2A), the ripple reached over 300mV, and in some circuits, it reached 600mV.

To reduce the ripple, I tried adding multiple large-capacity electrolytic capacitors to the input and output of the old board, but found that it did nothing to reduce the ripple.

After some research, I realized that low-ESR capacitors should be used for the input and output. My board has a 100uF electrolytic capacitor + 100nF ceramic capacitor at the input and a 22uF ceramic capacitor + 100nF ceramic capacitor at the output.

I discovered that the company's designed boards almost always used a combination of (22uF + 10uF + 100nF + 10nF ceramic capacitor) + (220uF/470uF electrolytic capacitor) or (10uF + 4.7uF + 100nF + 10nF ceramic capacitor) + (220uF/470uF electrolytic capacitor) for input and output.

After removing all the electrolytic capacitors from the old boards and replacing them with a single 10uF ceramic capacitor at the input, the ripple significantly decreased. Subsequently, changing the inductor value to a larger one further reduced the ripple; some ripple levels were reduced to 170+mV, while others remained above 300+mV.

This demonstrates that power supply capacitors must first meet low ESR requirements, and then capacitance should be considered. If the ESR is not low enough, adding large capacitors is futile. Theoretically, combining multiple capacitors can achieve an even lower ESR.

The issues with the ripple that couldn't be reduced were likely due to wiring problems, such as insufficient ground vias, excessively long ground loops, or the top and bottom layer GND wires near the DC-DC chip taking a long detour before connecting.

For example, one of our clients had a board designed and sent to Gerber for manufacturing. We discovered that the ripple was very high and the power supply quality was poor, resulting in a high product defect rate. Thousands of boards had to be reworked. The direct cause was that the client's board design had almost no GND vias, and the GND traces were thin and took a long detour before connecting. It was the kind of design where all the wires were connected and there were no error messages, so they considered it OK.




The above measurements were taken without an electrolytic capacitor. Low-ESR ceramic capacitors are used to smooth high-frequency ripple, while electrolytic capacitors are used to smooth low-frequency ripple. Without an electrolytic capacitor, the following will occur: a very high or very low voltage spike will appear when the load suddenly increases or decreases, as shown in the image below.

Adding a 470uF or larger electrolytic capacitor will effectively suppress this. If the 10uF + 100nF ceramic capacitor combination at the input still has significant ripple, add a 4.7uF and a 10nF capacitor as shown in the image above; the ripple will be perfectly suppressed. ![IMG_20240822_224221.jpg] The newly designed board uses a 4-layer design with as many ground vias as possible to ensure ground integrity (I'm not sure if this term is appropriate for me, haha). The ripple is lower and the temperature is significantly reduced. The advantage of lower temperature is that it can continuously output high current. The old board stopped outputting above 2.5A, but this version can continuously output above 3.5A, which is largely thanks to the 4-layer design. However, exceeding the rated current results in a very high temperature, and the ripple reaches over 250 Hz. Adding a solid-state capacitor in parallel didn't reduce the ripple further, so it was omitted. Ceramic capacitors produce a buzzing sound; I'll try replacing the uF ceramic capacitors with solid-state capacitors in the future. (The input electrolytic capacitor is not soldered; the output solid capacitor needs to be replaced with a large-capacity electrolytic capacitor. This will be added upon arrival.) ![IMG_20240821_001520.jpg] ![IMG_20240819_004250.jpg] Ripple at the USB port is 170+, and at the output capacitor it is tens of mV . ![Screenshot_20240819-003953.png] The MP1584 seems to be discontinued; you can replace it with the SY8205, which supports higher current, lower temperature, requires less inductance, and has lower ripple . ![IMG_20240821_001808.jpg] ![Screenshot_20240819-232849.png] **Calculation of the voltage divider resistor values ​​for the FB pin under different output voltages; download the attachment and study it yourself.**