TPS55289: Very big output ripple in PFM mode

Hi Johan,

Thank you for reaching out. It is IC normal behavior that output voltage ripple in PFM mode is higher than in FPWM mode. You can observe inductor current waveforms. In PFM mode, inductor current ripple is higher than in PWM in each switching cycle, causing the high voltage ripple. To optimize the ripple, please try the following advice.

1. Add more output capacitor. Add 3-4 more 10uF capacitor to see if it is better.

2.Reduce loop cross frequency. Change R55 to 10k to see if it is better.

About the schematic,

1. Recommend to add a 0402 0.1uF ceramic capacitor in Vout pin and put it as close as to IC.

2.Recommend to add RC snubber(330pF+2ohm) in SW1 and SW2.

3.Recommend to use type II compensation.

Regards,

Mulin

Thanks for your advise. I will try out your suggestions. It may take time to implement the suggestions.

I'll get back to you.

Question: The difference I see in output ripple is huge. approx 150 mV in FPWM mode and 1.9V in PFM mode. Is this enormous difference really normal behavior? Or is something else also wrong?. Also the low frequency of the ripple and the nice triangle shape i find suspicious.

Hi, I did not yet implement your recommendations, but I have a few new pictures. The pink is the SW2 signal

And in PWM mode it looks like this. You see, the output voltage has way lower ripple.

Don't pay too much attention to the overshoot. I think it is because of bad probing

And with very light load ( 2 mA)

You can see that it stops switching for a long time ( almost 4 ms) and the output voltage drops significantly until it resumes ( it drops lke 800 mV)

Hi Johan,

Since cross frequency is high, in PFM mode, inductor current would ramp much higher than in FPWM mode. Also, because output capacitor is not enough, the large inductor current will cause output voltage increase a lot in several switching cycles. It takes some time for output voltage to decrease to normal. So please follow my advice.

Regards,

Mulin

Thank you so much.

I followed some of your recommendations and it improved a lot.

I cannot follow all recommendations on this board. I will build a new one.

Adding capacitance to the output did help a lot.

I don't have room for just adding 4 0805 ceramic capacitors, so i soldered an electrolytic capacitor of 47 uF over the existing caps.

The ripple reduced drastically, now it is only approx 50 mV, but it is packed in hf nosie. and I see very big switching spikes, which i need to filter out.

The change of R55 to 10k didn't seem to do anything.

The elco has a not so low ESR, I measured it to be 0.67 Ohm. I'll order a better one. In the new design I'll use 4 ceramic capacitors

Would it be an idea to change the inductance of the coil? 

I added a 100 nF capacitor close to the pins of VOUT , but didn't see a change.

As an alternative, I changed the frequency ,, with one board, from 400 kHz to 1.8 Mhz. It was also an improvement, but of course, more switching noise.

I think increasing the output capacitance is the best option.

The snubber network is very difficult to build, becasue of the small size of the board. I''l take this into the new design.

One more thing, the TI Calculation tool comes with only 7.5 uF for a 50 mV ripple, and now we have 10 times that.

Hi Johan,

Glad to hear your update.

1. What is hf noise?

2.Add RC snubber will help reduce SW spike.

3. No need to change inductor. Make sure Isat of inductor is larger than 8A.

4.Some of my advice is for system stability, it may seem no difference, but please follow my advice.

5.Calculation tool is for PWM mode. Also, 7.5uF is effective capacitance, with DC bias effective capacitance will reduce a lot.

6.Do you have specific requirement about output voltage ripple? You don't have to care too much about it in low load PFM mode. It will reduce a lot when load increases and enters PWM(CCM) mode.

Regards,

Mulin