How to reduce TPS5430 power ripple?

[sage_xiong]

I've been learning about switching power supplies.I made my own test PCB.However, the maximum output ripple is 60mV. The example in the manual has a power ripple of only 30mV.
I want to know how to improve my design.Please point out the shortcomings of my design. Thank you!

[DanInvents]

You could use a capacitance multiplier at the output. It should reduce your voltage ripple to negligible levels. Dave showed how capacitance multipliers work in this video:

[sage_xiong]

Thanks a lot.It's very helpful.

[amaschas]

A schematic might be helpful in reviewing the board layout. Just at a cursory glance, the placement of C1 seems a little odd. Is that the boot capacitor? There looks to be a trace there going off to the right that connects to... ground? I also generally use more input capacitance with this IC. I start out with 2x10µF and adjust for performance.

[Weston]

Are you testing this at no load? It looks like the controller is switching for a cycle (the large peak you see) and then skips a few cycles while the diode and switch capacitance rings with the output inductor. You are mostly seeing components at a higher frequency than the switching frequency in that scope shot. The datsheet value was probably taken with the system operating under load, which would put the converter in CCM mode, which would avoid the ringdown period.

At these high frequencies the voltage across the electrolytic capacitor is going to be driven by the ESR and ESL more than the bulk capacitance value, which can be different between capacitors with the same capacitance and voltage rating. A ceramic capacitor in addition the electrolytic might lead to an improvement as it has very low ESR and ESL.

A small R-C snubber on the switching node would also reduce some of the high frequency ringing.

Some of this ripple may also be due to to your measurement technique. Measuring the ripple voltage is very sensitive to the probe and ground lead location, which can lead to ripple voltage measurements that are larger than the actual value. A long ground lead can resonate with the scope input capacitance, increasing the measured voltage. The ground lead loop can also couple to magnetic fields on the board, which can induce signals that don't actually exist on what what you are measuring. There is a lot of guidance about this online. Dave did a video on it at some point:


I don't think its contributing to the ripple you are seeing, but I would agree that you need more input capacitance on the PCB. The reference schematic assumes the IC is on a PCB with enough capacitance for decoupling, which is not the case for for your PCB with the long wires connecting to a power supply. The ripple current on the input of a buck converter is worse than the output so input capacitance is pretty important. An electrolytic capacitor similar to what you are using for the output filtering would work in addition to the existing ceramic capacitor. Without this capacitor you may start seeing some stability issues at higher current.

[sage_xiong]

Recently, I modified the previous design.

1. For the input, an input filter capacitor is added;
2. For the output, a tantalum capacitor and an MLCC are added for filtering;
3. Change the fixed output to an adjustable output;







Here are some test results of the modified design.

When 5V output, the power supply ripple is about 15mV.



At 9V output, the power supply ripple is about 20mV.

[sage_xiong]

Yes, I tested it with no load.

Thank you very much for your reply. Your reply has taught me a lot.

I just updated a version of the PCB based on what I've learned recently.I put it in the latest reply.

[sage_xiong]

I added input capacitors to a version of the PCB that was updated recently and it worked fine.