Things to consider while choosing input capacitor.

[nepdeep]

I am designing a switching power supply. In a lot of data sheets, they say the input capacitor should handle at least the following rms current
Rims=sqrt(dutycycle*(1-dutycycle))
It is very hard to find capacitor rated for higher rms current.

Eg to convert 10v to 1v
Dutycycle =0.1

For output current of 15 A(pretty high for processor core)

The capacitor should handle at lease

15*sqrt(0.1*(1-0.1))
=15* o.3 =4.5[A]

Man that is huge rating for capacitor, or where did I get it all wrong?

 

[D.A.(Tony)Stewart]

To handle large currents such as these, you need very low ESR Caps.

with 15mOhm @4.5A you can expect <80mV ripple pp
https://www.digikey.com/product-detail/en/PCF1A561MCL1GS/493-2997-2-ND/1136087

Choose lower ESR like <10mOhm, if you prefer.

 

[BradtheRad]

This topic seems to be about the output capacitor, rather than the input capacitor.

To reduce stress (intense current pulses) to the capacitor, split the burden across a bank of capacitors.

Or consider interleaving two converters. Stress on the output capacitor will take a big downward jump.

 

[dick_freebird]

Input capacitor current toggles between zero and full load current
every cycle. The ESR is a prime inefficiency term as well as any
VIN toggling being a source of control loop error (dynamic PSRR
and internal misbehaviors - like transient input supply droop
hitting undervoltage lockout on every turnon edge of the HSS?

Paralleling a number of smaller capacitors will get you to low
ESR/ESL, but this of course costs additional pennies.

Polyphase operation can level the input capacitor current
somewhat or even entirely depending on the phase overlap.

 

[BradtheRad]

Also consider using continuous current mode (CCM). This reduces the amount of current ripple in the output stage. It reduces the intensity of ampere pulses going into the smoothing capacitor.

Discontinuous mode produces greater current ripple. Therefore it causes greater stress on the capacitor.

It may seem desirable to increase the switching frequency. This allows you to reduce the coil and capacitor values, which makes them smaller and less expensive. However it does not strike us as really possible, when looking at a small capacitor, with its thin wire leads, to imagine 4.5 A going back and forth through it thousands of times per second, does it?