You can calculate the inductance using some basic formulae and math, it's simple enough to do in Excel in 5 minutes, no complex modeling needed.
First you need to decide the current ripple. This is how much the current through the inductor (and the output capacitor) changes during each cycle. Note that if your output capacitors are ideal, have infinite capacitance and zero ESR, current ripple is meaningless for output ripple voltage, which is always zero.
From the definition of inductance, V(t) = L * di/dt, solve for L.
We often decide the current ripple to be something like 25 to 50% of nominal or maximum output current. Say, for 1A output, di could be 0.3A.
You know the frequency, so you know dt. Say for example, your input is 10V and the output is 2V, so in a buck the duty cycle is 20% on, 80% off. At 1MHz, t_on is 0.2 us, and t_off is 0.8 us. When the switch is on, the inductor sees 8V over it; when the switch is off, the inductor sees 2V over it as the current goes through the freewheeling switch (active or passive). All of this makes much more sense if you draw the switch control signal, and the expected triangle waveform of the inductor current, on paper with the circuit diagram.
The hardest part for off-the-shelf inductor selection is the AC loss estimation or calculation. Data for this is never provided in any inductor datasheet, yet the range is orders of magnitude, some inductors are completely unsuitable for switch mode power supplies (say, results in 5% efficiency or an inoperational converter), some have nearly zero AC loss. The process is either:
1) Some manufacturers give you a calculator tool (nowadays web-based). You input your converter parameters and get a result out.
2) Luck and experimentation. You just buy an inductor and prototype it. Yes, really.
#2 isn't too bad for a small converter, just pick a ferrite-based, physically small, definitely shielded SMD inductor with enough saturation and thermal current rating, and it's likely good. Those look like small enclosed cubes of ferrite. I have never had bad luck with these.
Saturation current of the inductor must be at least the maximum short-circuit current the controller will provide before entering protection (see the MAX column in the datasheet!), plus half of the current ripple. People often overlook this and make marginal designs using expected output current.
And, modern converters are fine with ceramic output capacitance. For your low voltage ripple, just add enough ceramic output capacitance. With near-zero ESR, the voltage ripple, theoretically, reaches zero as you increase the capacitance. Paralleling multiple in small packages like 0603 or 0805 is good for decreasing stray inductance (which would cause part of the ripple to go past the cap).
At some point, adding a second LC stage may become cheaper than increasing C. Additionally, using very small values of L,C (of lossy types) as a secondary filter may help you filter very high frequency noise, but this isn't called "ripple".
The caveat is that most regulators enter pulse-skipping modes at low loads, effectively (from the output ripple viewpoint) meaning they work at much lower frequencies.
Home
Help
Search
Login
Register
