I'm using the TPS56221 buck converter.

500 kHz switching frequency 12V in 5V out, 16A max, 14A load step

I've found out all the values I need, the problem I'm having is finding suitable capacitors. For example I need around 600uF output capacitance and 200uF input capacitance. However at 12V any capacitor I get will be more than half it's capacitor value due to the DC biasing. Even at 5V, most capacitors are half their value such as the C1210C107M4PAC7800 (100uF) that I want to use. I could put 12+ of those to get the 600uF that I need, but I'm wondering if there is a better solution here - for example the MegaCaps from TDK that keep the capacitance relatively stable with increasing voltage

Thanks in advance!


closed as off-topic by Olin Lathrop, PeterJ, Dmitry Grigoryev, Voltage Spike, winny Sep 13 '17 at 10:34

This question appears to be off-topic. The users who voted to close gave this specific reason:

If this question can be reworded to fit the rules in the help center, please edit the question.

  • \$\begingroup\$ Please ask a specific question, you'll get better answers and your question wont get closed \$\endgroup\$ – Voltage Spike Sep 11 '17 at 16:34

You're making something like a desktop PC CPU VRM, except your output current is lower, but the input power is the same. So, let's use PC mobo 105°C polymer caps.

Input caps: we have 20A 500kHz ripple. This will be handled mostly by MLCCs, but since you mention 14A load step, if we have a load step in the 10-100kHz range, then the 12V supply may or may not respond quickly enough. So we need local bulk caps.

  • 4-5 R7 or the more expensive but smaller OSCONs
  • 0805/0603 MLCCs, try to stagger the values like 10x10µF - 10x1µF. The 10µF can be 1206's. The 1µF should be small and as close as possible to the DC-DC.

Use a wide copper pour for VIN...

The MLCC's job is to close the HF current loop right at the source so the nasty fast edges don't turn into EMI. The polymer cap's job is to help the main 12V power supply.

Mind the ripple current rating of your MLCCs too! If they heat too much they will crack. I would't use less than 10 caps, that's 2 amps pumped in a tiny 0805 or 1206...

Output caps:

The load will decide how many close decoupling ceramics it needs. Here, I'm just talking about the caps at the output of the DC-DC, which again close the HF current loop and absorb the inductor ripple current. The datasheet goes for 6A ripple, so let's use that.

So, 2-3 S8 or R7 or OSCON or flat caps... plus a few 1-10µF in 0805 or 0603, say 4 of each.

If the caps use too much space, you're gonna need a higher switching frequency. If you wanted to use MLCCs only due to height restrictions, check the flat Panasonic polymer caps I posted. Also, big MLCCs tend to crack if the board is flexed.

Also, even at 95% efficiency, you'll still burn 4W in a tiny area, so keep thermal management in mind...

  • \$\begingroup\$ Thanks so much! I was originally looking only at ceramics because I didn't want the high ESR and comparatively short lifetime of Aluminum capacitors - this hit the perfect middle spot ! \$\endgroup\$ – VanGo Sep 8 '17 at 21:20
  • \$\begingroup\$ You're welcome! A surprise bonus of polymer caps is that ESR is stable with temperature, which means they work well in the cold, unlike standard electrolytics. Also here's an excellent doc, every page has tons of useful info: chemi-con.co.jp/e/catalog/pdf/… pay attention to antiresonance, the ceramic caps in parallel with the low-ESR caps should be large enough (10µF minimum), you should run a spice sim. \$\endgroup\$ – peufeu Sep 9 '17 at 0:25

Not the answer you're looking for? Browse other questions tagged or ask your own question.